header logo image


Page 32«..1020..31323334..40..»

Archive for the ‘Genetics’ Category

Plotly Sponsors Development of Predictive Genetics Application in Partnership with McGill University – GlobeNewswire

Monday, October 21st, 2019

MONTREAL, Oct. 21, 2019 (GLOBE NEWSWIRE) -- Plotly, developer of the leading data science platform for creating analytic applications, today announced a partnership with McGill University to fund three Ph.D. interns in collaboration with Mitacs, a not-for-profit organization that fosters growth and innovation in Canada. The doctoral students will work with Sahir Rai Bhatnagar, Assistant Professor of Biostatistics, to create a predictive genetics application to better understand the genetic determinants of temporomandibular disorder (TMD). The tool will be based on a machine learning-driven analysis of the largest available dataset on TMD, which causes pain in the jaw.

Plotly and Mitacs are working together to provide financial support for the interns, and Plotly is contributing support for use of the companys powerful Dash software, an open source platform for building analytic web applications. Dr. Bhatnagars team will use Dash to interactively analyze the large dataset and visualize results from the machine learning model. These visualizations will provide key insight into which genetic components are driving the predictions. The team will create analytical applications in the R programming language for use by researchers studying pain and working to identify drug targets in order to develop more effective treatments.

This partnership expands an ongoing collaboration between Plotly and Mitacs. Last year, Plotly sponsored three bioinformatics interns at the Universit de Sherbrooke as they developed a visualization tool for the universitys CoBIUS Lab. The model enabled researchers to view DNA or complex molecules in 3D.

Plotly is delighted to work Mitacs to partner with McGill in support of technical talent in Qubec, said Jack Parmer, CEO of Plotly. Its important to us to give back to the communities were a part of, from open source data science to Canadian research teams. By contributing funding and use of our technology to these three promising biostatistics researchers, we hope to benefit not only the students, but patients across Qubec.

Dr. Bhatnagar commented: Funding from Plotly and Mitacs will help us bring our work out of the lab and to Canadas healthcare community. Dash will help our team visualize a trove of data and may give us, as well as the researchers at the Qubec Pain Research Network, more insight than ever before on temporomandibular disorder.

For 20 years, Mitacs has helped develop partnerships between Canadian industry and post-secondary institutions, and were happy to continue that work by joining with Plotly and biostatistics researchers at McGill University, said Eric Bosco, Chief Business Development Officer at Mitacs. Recognizing the data analysis capabilities of Plotly and biostatistics expertise of Dr. Bhatnagars team, we see this as a perfect match for both organizations to elevate their work and to help understand chronic pain alongside the Qubec Pain Research Network.

About MitacsMitacs is a not-for-profit organization that fosters growth and innovation in Canada for business and academia. Working with 70 universities, thousands of companies, and both federal and provincial governments, Mitacs builds partnerships that support industrial and social innovation in Canada. Open to all disciplines and all industry sectors, projects can span a wide range of areas, including manufacturing, business processes, IT, social sciences, design, and more. Mitacs is funded by the Government of Canada and the Government of Quebec, along with every other province, industry and not-for-profit partners, and academic partners.

About PlotlyFounded in 2013, Plotly is a data visualization company focused on taking data science out of the lab and into the business. Plotly makes it easy to create, deploy, and share interactive web apps, graphs, and visualizations in any programming language. Plotly's libraries are used by millions worldwide and embedded into mission critical applications across the Fortune 500.

ContactDanielle Toboni617-945-1915Plotly@LaunchSquad.com

See the original post here:
Plotly Sponsors Development of Predictive Genetics Application in Partnership with McGill University - GlobeNewswire

Read More...

Hacking Darwin: How the coming genetics revolution will play out – New Atlas

Monday, October 21st, 2019

Jamie Metzl is an extremely impressive man. Having held senior positions on Clinton's National Security Council and Department of State, and Joe Biden's Senate Foreign Relations Committee, he's also been Executive VP of the Asia Society, a Senior Fellow at the Atlantic Council and a former partner in Cranmere LLC, a global investment company. Today, while he's not running ultra-marathons, he's best known as a geopolitics expert, futurist and author.

Metzl writes in science fiction and scientific non-fiction, and his latest book, Hacking Darwin: Genetic Engineering and the Future of Humanity, delivers a serious, strongly-researched warning on what's likely to happen if we sleepwalk into the genetics age.

We spoke to Metzl at WCIT 2019 in Yerevan, Armenia, where he appeared as a keynote speaker, moderator and panel member.

Vahram Baghdasaryan/WCIT Yerevan 2019

"Right now were at this moment of super-convergence," Metzl tells us. "Its not any one technology thats determinative, its all these technologies happening at the same time, because theyre all influencing each other."

The first of these, Metzl outlines, is cheap sequencing of the human genome. Well need a ton of genetic information to be able to find the patterns needed to move forward, and while the cost of full genome sequencing is currently the limiting factor, it's dropping steeply, from around US$2.7 billion in 2003, to less than US$600 today. That's going to have to come down by another order or two of magnitude before it starts getting used as a matter of routine.

Secondly, 5G connectivity and the Internet of Things promises to teach us enormous amounts of information about people's health over the years, as wearable health analysis devices begin to stream back colossal piles of dynamic data about what's going on in people's bodies. Eventually, this will enable population-wide phenotypical research to be cross-checked against the genome to learn even more about how genes express themselves, individually and in concert with one another.

Thirdly, big data and analytics tools. The 2.9 billion haploid pairs making up each sequenced human genome represent about 725 megabytes of data, and dynamic health records will likely require even more storage space, in formats that can easily be cross-checked against each other at a massive scale.

Metzl notes that artificial intelligence or more precisely, deep learning is the only way we'll ever be able to meaningfully process such monstrous amounts of data, and its capabilities are rocketing forward daily. Perhaps when it's ready for serious commercial use, the speed and power of quantum computing will prove invaluable in quickly crunching through these petabytes of data.

Then of course, there are the wetter technologies: vastly improved IVF technologies that will soon enable us to generate egg and sperm cells from skin cells without needing invasive or embarrassing procedures to be carried out. Eventually, we'll have the capability to cheaply produce dozens, or even hundreds of embryos to sequence, select and implant.

And of course, gene editing tools. CRISPR/Cas9 editing is the most famous example of these, but it's already being compared to "genetic vandalism" due to its imprecise nature. More accurate and precise tools are constantly being discovered and refined to edit the genome of living subjects.

"We have to stitch together all these technologies," says Metzl, "and its already starting to happen. And itll happen increasingly until the end of time."

Vahram Baghdasaryan/WCIT Yerevan 2019

The first step, says Metzl, will be in healthcare. Our interactions with health care professionals will move from the current generalized model, to something more personal and precise as we start to understand what treatments work for people with certain genetic markers. Eventually, we'll have enough information to start engaging in predictive health care.

"You dont need to be perfect to make a huge impact on health care," says Metzl, "you just have to be better than the status quo, where nobody has that information, for it to be applied." It'll inch forward, offering probabilities rather than certainties as more and more is discovered.

Next and soon, we'll start seeing advanced embryo selection as a core part of any IVF treatment. Prospective parents will start having multiple embryos to choose from, each of which will have its genome fully sequenced so they'll be able to choose between offspring with a growing amount of information.

To begin with, this will allow parents to select against crippling genetic diseases, much the same as how parents who can afford the right scans can "select against" things like Down syndrome now.

But as science learns more and more about what individual genes, and combinations of them, do to the final outcome of a human, we'll quickly gain the ability to select for certain traits as well as against others. Will you want your child to be taller? More athletic, with a greater proportion of fast twitch muscle fibers? What about intelligence? Skin color? Eye color? Would you select for a child with a higher probability of living longer? Would you select for a child with a higher degree of extraversion, or a more even temperament?

All these things, and many more, are already known to have genetic underpinnings. And soon, parents will be able to choose between dozens, or potentially hundreds of their own biological embryos with this information at hand. All these possible kids are yours, so why wouldn't you choose the one that appears to have the best possible shot at life? Not doing so, says Metzl, could grow to be seen as a "crime against potential."

The disadvantages of having children the old-fashioned way will soon become apparent, as smarter, stronger, faster, healthier kids born from selection processes begin to dominate across a range of competitive situations, from sport to business to earning capacity and these advantages will multiply with subsequent generations, as more and more science is applied to the reproductive process.

"Embryo selection uses technologies that already exist," says Metzl. "IVF, embryo screening, and genome sequencing. Obviously we need to get better at all these things, but its happening very, very quickly."

And that's just using our naturally-occurring genetics. Soon afterward will follow precision gene editing, in which you select option J from your pre-implanted embryos, but make a few tweaks before you implant it. Here's where things start getting a little sketchy, as you're making edits to the germ line of the human species.

"Editing the genome requires the understanding that one gene might not just do one thing; it might do a lot of things," Metzl tells us. "If its a particularly harmful gene, then we know the alternative is deadly, so that decision will be easier. But when we move into the world of non-deadly single gene mutations, well, then the costs of not having a full understanding go higher."

Metzl says it's clear which direction things will go."We are going to do more and more complex genome editing," he tells us, "either to address risks, or to create enhancements - and there will be no natural boundary between the two. This is all about ethics. The science is advancing, theres nothing we can do to stop the science. The question is ethics."

The dawn of a new age of superhumans could nearly be upon us, in which a lucky, selected, edited few will have extraordinary genetic potentials in a wide range of areas. Sports could become almost meaningless, as it'll be impossible to tell a selected or edited human from a "natural born" one. Humanity will begin steering its own evolution for the first time in history, with some predictable results and some we can't see coming.

Negative results, says Metzl, could include everything from a gaping division between genetic haves and have-nots which could express itself within and between countries all the way up to eliminating all human life altogether. "We may make choices based on something we think is really good, like eliminating a terrible disease," says Metzl, "but then that genetic pattern that enabled that disease, in some other formulation, could be protective against some threat we cant even imagine, thats coming our way a thousand years from now. Thats why we need to be so respectful of our diversity. Genetic diversity, up to this point, has been our sole survival strategy. If we didnt have diversity, you could say wed still be single-celled organisms. We wouldnt, wed probably just have died. When the world changes around us, diversity is what helps us survive."

And then there's the potential of creating genetically engineered weapons. "Researchers in Canada spent $100,000 a couple of years ago," says Metzl, "to create essentially a weaponized version of horse pox in the lab, to show what could be done. Well, that could probably now be done for $20,000. In five years, you might be talking $2,000. These tools are agnostic. They dont come with their own value system. Just like nuclear power. We had to work out what are the OK uses, what are the not OK uses, and how do we structure things to we minimize the downsides."

Metzl wants people across the world to be informed about the technologies and capabilities that are barreling down the pipeline toward us, so meaningful efforts can be made to steer them in a direction that everyone can agree on, and set up clear redlines past which we agree not to venture. Each country, he says, needs to set up a national regulatory infrastructure to control the pace of these changes, and there also needs to be an international body with some teeth to make sure certain nations don't leap ahead and change the nature of humanity just due to lax regulations.

"This is always going to be changing," says Metzl. "The science is changing, the societal norms about what is and isnt OK are going to be changing too, and we need a dynamic process that can at least try to do a better job of keeping up with that rapid change."

Where does Metzl stand personally on how this next phase should be approached? "I'm a conservative person about this," he says. "I mean, four billion years of evolution is a lot. Life has made a lot of trade-offs. So if youre going against four billion years of evolution, you have to be humble. We know so little about the body. We cant let our hubris run away with us."

If you want to get informed on this incredibly complex, multilayered and potentially explosive technological revolution, Metzl's book Hacking Darwin: Genetic Engineering and the Future of Humanity (April 2019) is an outstanding summary with more examples and possible future situations laid out than you could possibly need, written in an engaging style designed to be accessible to anyone. I found it extremely enlightening and recommend it thoroughly.

Source: Jamie Metzl, WCIT Yerevan 2019

See the original post here:
Hacking Darwin: How the coming genetics revolution will play out - New Atlas

Read More...

Homology Medicines Presents Data from Investigational PKU and MLD Gene Therapy Programs that Demonstrate Preclinical Proof-of-Concept for Potential…

Monday, October 21st, 2019

BEDFORD, Mass., Oct. 21, 2019 (GLOBE NEWSWIRE) -- Homology Medicines Inc. (Nasdaq: FIXX), a genetic medicines company, announced today the presentation of preclinical data that support its investigational gene therapy programs for the treatment of metachromatic leukodystrophy (MLD) and phenylketonuria (PKU) at the American Society of Human Genetics (ASHG) 2019 Meeting.

For the first time, Homology presented preclinical data from the murine model and non-human primates that demonstrated that the HMI-202 gene therapy candidate crossed the blood-brain-barrier and the blood-nerve-barrier and increased levels of arylsulfatase a (ARSA) protein to therapeutic levels. In addition, preclinical data on gene therapy candidate HMI-102 showed that a single administration resulted in sustained Phe reduction and increased tyrosine and other important downstream biochemical metabolites in the PKU murine model.

The MLD presentation is part of a growing foundation of HMI-202 data to support a future IND filing, and the PKU preclinical data supported the initiation of our Phase 1/2 trial, which is ongoing and expected to report initial data by the end of this year, said Albert Seymour, Ph.D., Chief Scientific Officer of Homology Medicines. Taken together, these presentations demonstrate the potential of our genetic medicines platform, investigational PKU and MLD gene therapies and our continued focus on advancing these treatments to help patients and their families.

Highlights from the posters include:

HMI-202 gene therapy in development for MLD

HMI-102 investigational gene therapy for PKU

This poster received a Reviewers Choice Abstract award during the ASHG Meeting.

A 5-year retrospective study of individuals with PKU treated at two specialized U.S. clinics

For more information, please visit http://www.homologymedicines.com/publications.

About Phenylketonuria (PKU)PKU is a rare, inherited inborn error of metabolism caused by mutations in the PAH gene. The current standard of care is a highly restrictive diet, but it is not always effective, and there are currently no treatments available that address the genetic defect in PKU. If left untreated, PKU can result in progressive and severe neurological impairment. PKU affects approximately 16,500 people in the U.S., and an estimated 350 newborns are diagnosed each year.

About Metachromatic Leukodystrophy (MLD)MLD is a rare lysosomal storage disorder caused by mutations in the ARSA gene. ARSA is responsible for the creation of the arylsulfatase A (ARSA) protein, which is required for the breakdown of cellular components. In MLD, these cellular components accumulate and destroy myelin-producing cells in the peripheral and central nervous system leading to progressive and serious neurological deterioration. The late infantile form of the disorder is estimated to affect 1 in 40,000 people, and it is fatal within five to ten years after onset.

About Homology Medicines, Inc.Homology Medicines, Inc. is a genetic medicines company dedicated to transforming the lives of patients suffering from rare genetic diseases with significant unmet medical needs by curing the underlying cause of the disease. Homologys proprietary platform is designed to utilize its human hematopoietic stem cell-derived adeno-associated virus vectors (AAVHSCs) to precisely and efficiently deliver genetic medicinesin vivoeither through a gene therapy or nuclease-free gene editing modality across a broad range of genetic disorders. Homology has a management team with a successful track record of discovering, developing and commercializing therapeutics with a particular focus on rare diseases, and intellectual property covering its suite of 15 AAVHSCs. Homology believes that its compelling preclinical data, scientific expertise, product development strategy, manufacturing capabilities and intellectual property position it as a leader in the development of genetic medicines. For more information, please visitwww.homologymedicines.com.

Forward-Looking Statements This press release contains forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. All statements contained in this press release that do not relate to matters of historical fact should be considered forward-looking statements, including without limitation statements regarding our expectations surrounding the potential, safety, efficacy, and regulatory and clinical progress of our product candidates; plans and timing for the release of clinical data; our beliefs regarding our manufacturing capabilities; the potential of and related advancement of our novel platform and pipeline; our goal of delivering potential cures to patients; beliefs about preclinical data; our position as a leader in the development of genetic medicines; and the sufficiency of our cash, cash equivalents and short-term investments. These statements are neither promises nor guarantees, but involve known and unknown risks, uncertainties and other important factors that may cause our actual results, performance or achievements to be materially different from any future results, performance or achievements expressed or implied by the forward-looking statements, including, but not limited to, the following: we have and expect to continue to incur significant losses; our need for additional funding, which may not be available; failure to identify additional product candidates and develop or commercialize marketable products; the early stage of our development efforts; potential unforeseen events during clinical trials could cause delays or other adverse consequences; risks relating to the capabilities and potential expansion of our manufacturing facility; risks relating to the regulatory approval process; our product candidates may cause serious adverse side effects; inability to maintain our collaborations, or the failure of these collaborations; our reliance on third parties; failure to obtain U.S. or international marketing approval; ongoing regulatory obligations; effects of significant competition; unfavorable pricing regulations, third-party reimbursement practices or healthcare reform initiatives; product liability lawsuits; failure to attract, retain and motivate qualified personnel; the possibility of system failures or security breaches; risks relating to intellectual property and significant costs as a result of operating as a public company. These and other important factors discussed under the caption Risk Factors in our Quarterly Report on Form 10-Q for the quarter ended June 30, 2019 and our other filings with the SEC could cause actual results to differ materially from those indicated by the forward-looking statements made in this press release. Any such forward-looking statements represent managements estimates as of the date of this press release. While we may elect to update such forward-looking statements at some point in the future, we disclaim any obligation to do so, even if subsequent events cause our views to change.

Read this article:
Homology Medicines Presents Data from Investigational PKU and MLD Gene Therapy Programs that Demonstrate Preclinical Proof-of-Concept for Potential...

Read More...

Research presented by Invitae at the American Society of Human Genetics Meeting Pushes Science and Practice of Genetics Forward – Yahoo Finance

Monday, October 21st, 2019

-- Advances in classification, new approaches to genetics in cancer and implications of primary and secondary findings for clinical care among wide-ranging data presentations --

HOUSTON, Oct. 17, 2019 /PRNewswire/ -- Researchers fromInvitae Corporation (NVTA), a leading medical genetics company, are presenting data showing the increasing utility of genetic information at the American Society of Human Genetics (ASHG) annual meeting this week, ranging from comprehensive screening for cancer patients, to appropriate clinical follow up for women using non-invasive prenatal screening, to the limitations of direct to consumer genetic screening health reports.

The company's research includes three platform presentations and multiple poster sessions, many performed in collaboration with leading academic researchers. Among the data presented is a study evaluating the utility of combined germline testing and tumor profiling (somatic testing) in cancer patients. Germline and somatic testing are increasingly used in precision treatment of people with cancer, although frequently are ordered separately in clinical practice. Data presented at the meeting shows a substantial number of patients with medically significant variants in hereditary cancer syndrome genes in their tumor profile carry the same variant in their germline, thereby establishing a previously unknown risk of hereditary cancer and suggesting the value of combined or concurrent testing to inform precision medicine approaches.

"The research we are presenting at this year's ASHG meeting provides meaningful insight into both the science and practice of genetics, helping identify how we as clinicians can better use deep genetic insights to help a wide array of patients, whether they are cancer patients, women having a child or healthy adults seeking to better understand their risk of disease," said Robert Nussbaum, M.D., chief medical officer of Invitae. "We are proud and grateful to be able to join our colleagues from across genetic medicine in meaningful conversations that push genetic medicine forward."

Following are research from the company and collaborators to be presented at the meeting:

Story continues

Wednesday, October 16:

Poster presentation #819W | 2:00 3:00 pm Germline testing in colorectal cancer: Increased yield and precision therapy implications of comprehensive multigene panels. Presented by Shan Yang, PhD. Invitae.

Poster presentation #2427W | 2:00 3:00 pm Harmonizing tumor sequencing with germline genetic testing: identification of at-risk individuals for hereditary cancer disorders. Presented by Daniel Pineda-Alvarez, MD, FACMG, Invitae.

Poster presentation #606W | 3:00 4:00 pm A comprehensive evaluation of the importance of prenatal diagnostic testing in the era of increased utilization of non-invasive prenatal screening. Presented by Jenna Guiltinan, MS, LCGC, Invitae.

Thursday, October 17:

Platform presentation #235 | 5:00 pm, Room 370A, Level 3 Limitations of direct-to-consumer genetic screening for hereditary breast, ovarian and colorectal cancer risk. Presented by: Edward Esplin, MD, PhD, FACMG, FACP, Invitae.

Poster presentation #763T | 2:00 3:00 pm In-depth dissection of APC pathogenic variants: Spectrum of more than 400 pathogenic variants, challenges of variant interpretation, and new observations in a large clinical laboratory testing cohort. Presented by: Hio Chung Kang, PhD, Invitae.

Poster presentation #1399T | 2:00 3:00 pm Prediction of lethality and severity of osteogenesis imperfecta variants in the triple-helix regions of COL1A1 and COL1A2. Presented by: Vikas Pejaver, PhD, University of Washington.

Friday, October 18:

Platform presentation #264 | 9:00 am, Room 361D, Level 3 Million Veteran Program Return Of Actionable Results - Familial Hypercholesterolemia (MVP-ROAR-FH) Study: Considerations for variant return to mega-biobank participants. Presented by Jason Vassy, MD, MPH, VA, Boston Healthcare System.

Platform presentation #265 | 9:15 am, Room 361D, Level 3 Comprehensive secondary findings analysis of parental samples submitted for exome evaluation yields a high positive rate. Presented by Eden Haverfield, DPhil, FACMG, Invitae.

Poster presentation #698F | 2:00 3:00 pm Reporting of variants in genes with limited, disputed, or no evidence for a Mendelian condition among GenomeConnect participants. Presented by: Juliann Savatt, MS, LGC, Geisinger.

About InvitaeInvitae Corporation(NVTA)is a leading medical genetics company, whose mission is to bring comprehensive genetic information into mainstream medicine to improve healthcare for billions of people. Invitae's goal is to aggregate the world's genetic tests into a single service with higher quality, faster turnaround time, and lower prices. For more information, visit the company's website atinvitae.com.

Safe Harbor StatementsThis press release contains forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995, including statements relating to the increasing utility of genetic information; the utility of combined germline and somatic testing; and the benefits of the company's research. Forward-looking statements are subject to risks and uncertainties that could cause actual results to differ materially, and reported results should not be considered as an indication of future performance. These risks and uncertainties include, but are not limited to: the applicability of clinical results to actual outcomes; the company's history of losses; the company's ability to compete; the company's failure to manage growth effectively; the company's need to scale its infrastructure in advance of demand for its tests and to increase demand for its tests; the company's ability to use rapidly changing genetic data to interpret test results accurately and consistently; security breaches, loss of data and other disruptions; laws and regulations applicable to the company's business; and the other risks set forth in the company's filings with the Securities and Exchange Commission, including the risks set forth in the company's Quarterly Report on Form 10-Q for the quarter ended June 30, 2019. These forward-looking statements speak only as of the date hereof, and Invitae Corporation disclaims any obligation to update these forward-looking statements.

Contact:Laura D'Angelopr@invitae.com(628) 213-3283

View original content to download multimedia:http://www.prnewswire.com/news-releases/research-presented-by-invitae-at-the-american-society-of-human-genetics-meeting-pushes-science-and-practice-of-genetics-forward-300940213.html

Read more from the original source:
Research presented by Invitae at the American Society of Human Genetics Meeting Pushes Science and Practice of Genetics Forward - Yahoo Finance

Read More...

Most genetic studies use only white participants this will lead to greater health inequality – The Conversation UK

Monday, October 21st, 2019

Few areas of science have seen such a dramatic development in the last decade as genomics. It is now possible to read the genomes of millions of people in so-called genome-wide association studies. These studies have identified thousands of small differences in our genome that are linked to diseases, such as cancer, heart disease and mental health.

Most of these genetic studies use data from white people over 78% of participants are of European descent. This doesnt mean that they represent Europe. In fact, only three nationalities make up most of the participants: the US, UK, and Iceland. Even though the UK and the US have very diverse populations, their non-white citizens have rarely been included in genetic research.

In recent years, efforts to collect multi-ethnic data have increased. One example is the UK Biobank, a collection of data from half a million British people accessible to any bona fide researcher. It includes some 35,000 DNA samples from people who are either non-European or mixed-race. Yet 92% of research papers on UK Biobank only used the data from the European-descent samples. So collecting data doesnt automatically solve the problem of non-white representation in research.

The under-representation of non-European groups is problematic for scientific and ethical reasons. The effects of gene variants that are present only in the unstudied groups remain unknown, which means important clues about the causes of diseases might be missed. Such undiscovered genes would not be included when testing for genetic diseases. So a person carrying one of them could wrongly get a negative genetic test result and might be told that they are not at increased risk of developing the disease.

Read more: How the genomics health revolution is failing ethnic minorities

Our recent work also shows that existing genetic findings might not apply equally to non-European populations. We found that some gene variants predicting high cholesterol in white populations do not lead to the same heart problems in people from rural Uganda. These findings should serve as a major warning to the field of genetics one cannot blindly apply findings from ancestrally European groups to everyone else.

It is important to support the global application of research because scientists have a moral responsibility to develop science for the benefit of the whole of humanity, not restricted by ethnic, cultural, economic or educational boundaries. Some 80% of the worlds population live in low and middle-income countries where healthcare and research are constrained by limited financial and human resources. We should not overlook this part of the world.

Studying different populations has advanced the medical field for everyones benefit. For example, the first disease gene mapped in humans was the gene for Huntingtons disease in 1983, identified through examining a large population of patients in villages surrounding Lake Maracaibo in Venezuela. The area was found to have the largest concentration of Huntingtons disease sufferers in the world, which helped them to find the gene.

More recently, a study of schizophrenia found new risk genes by using African and Latino American samples. Genetic risk scores based on results from these groups improved the ability to predict who would develop schizophrenia in all ethnic groups.

Read more: Decolonise science time to end another imperial era

Two things need to happen if we want to avoid increasing health disparities and instead share the medical benefits of genomic science across countries and ethnic groups. First, we need more large diverse studies. First steps in this direction are being taken by the Human Hereditary and Health in Africa Initiative. PAGE and All of Us are paving the way to recruit more diverse ethnic groups in the US, and East London Genes and Health focuses on people of South Asian origin in London.

And second, to make sure diverse ethnic data resources are widely used by researchers, the challenges of analysing genetic data from ancestrally diverse samples need to be addressed. While there are statistical solutions, more work is needed to make them easy to use and give clear guidance about the best approach.

Understanding how genetic risk and social inequality interact to influence disparities in disease risk and outcomes will be critical to improving public health for all.

Go here to read the rest:
Most genetic studies use only white participants this will lead to greater health inequality - The Conversation UK

Read More...

US Dairy Cows Are Very Genetically Similar. That’s Not Good : The Salt – NPR

Monday, October 21st, 2019

Unlike most dairy cows in America, which are descended from just two bulls, this cow at Pennsylvania State University has a different ancestor: She is the daughter of a bull that lived decades ago, called University of Minnesota Cuthbert. The bull's frozen semen was preserved by the U.S. Agriculture Department. Dan Charles/NPR hide caption

Unlike most dairy cows in America, which are descended from just two bulls, this cow at Pennsylvania State University has a different ancestor: She is the daughter of a bull that lived decades ago, called University of Minnesota Cuthbert. The bull's frozen semen was preserved by the U.S. Agriculture Department.

Chad Dechow, a geneticist at Pennsylvania State University who studies dairy cows, is explaining how all of America's cows ended up so similar to each other.

He brings up a website on his computer. "This is the company Select Sires," he says. It's one of just a few companies in the United States that sells semen from bulls for the purpose of artificially inseminating dairy cows.

Dechow chooses the lineup of Holstein bulls. This is the breed that dominates the dairy business. They're the black-and-white animals that give a lot of milk.

Dairy farmers can go to this online catalog and pick a bull, and the company will ship doses of semen to impregnate their cows. "There's one bull we figure he has well over a quarter-million daughters," Dechow says.

The companies rank their bulls based on how much milk their daughters have produced. Dechow picks one from the top of the list, a bull named Frazzled. "His daughters are predicted to produce 2,150 pounds more milk than daughters of the average bull," he says, reading from the website.

Farmers like to buy semen from top-ranked bulls, and the companies keep breeding even better bulls, mating their top performers with the most productive cows. "They keep selecting the same families over and over again," Dechow says.

A few years ago, Dechow and some of his colleagues at Penn State made a discovery that shocked a lot of people. All the Holstein bulls that farmers were using could trace their lineage back to one of just two male ancestors. "Everything goes back to two bulls born in the 1950s and 1960s," he says. "Their names were Round Oak Rag Apple Elevation and Pawnee Farm Arlinda Chief."

This doesn't mean that the bulls in the catalog are genetically identical. They still had lots of different mothers, as well as grandmothers. But it does show that this system of large-scale artificial insemination, with farmers repeatedly picking top-rated bulls, has made cows more genetically similar. Meanwhile, genetic traits that existed in Holstein cows a generation ago have disappeared.

"We've lost genetic variation," Dechow says. "Now, some of that variation was garbage that we didn't want to begin with. But some of it was valuable stuff."

To see what might have been lost, Dechow decided to do an experiment. He located some old semen from other bulls that were alive decades ago, with names like University of Minnesota Cuthbert and Zimmerman All-Star Pilot. You might call them heirloom bulls. The U.S. Agriculture Department keeps samples of their semen in deep-freeze storage in Fort Collins, Colo.

Dechow used that semen to impregnate some modern cows. They gave birth, and now it's possible to see some lost pieces of the Holstein family tree come to life in a barn at Penn State in the form of three cows.

Dechow leads the way to the barn. He points toward a cow that eyes us suspiciously. "Here is our old genetic lineage, [cow] number 2869," he says.

To the untrained eye, this cow looks pretty much like all the others. But Dechow sees things that others can't. "If you notice, if you look over her back see how that cow to her left is a little more bony?" he says.

Once Dechow points it out, the difference is plain to see. "So she definitely carries more body condition. She's a little bit fatter," he says.

Traditionally, dairy farmers didn't like cows with extra body fat. They thought the ideal cow was a skinny one, because she was turning all her feed into milk, not fat. So farmers chose bulls that tended to produce that kind of daughter.

"We've kind of selected for tall, thin, cows," Dechow says. "And that's a really bad combination. They're infertile, unhealthy. So we need to get away from that."

Dechow thinks the frozen semen from those long-forgotten heirloom bulls can bring back valuable genes that went missing maybe genes that would allow cows to thrive in warmer temperatures, for instance.

For this to work, though, farmers actually have to use those bulls, and they'll only do so if they're persuaded that the daughters will also produce lots of milk.

So Dechow is carefully monitoring his experimental cows. So far, he says, it's going pretty well. Two of the three cows are producing at least as much milk as the industry average.

Follow this link:
US Dairy Cows Are Very Genetically Similar. That's Not Good : The Salt - NPR

Read More...

23andMe Wants Everyone to Get Used to Sharing Their Genetic Data – VICE

Monday, October 21st, 2019

This article originally appeared on VICE US.

What Anne Wojcicki, co-founder and CEO of 23andMe, would like you to consider is this: Once upon a time, people were kinda freaked out by the idea of typing their credit card info into a website. Now this is done basically all the time. Your phone is a digital wallet; your web browsers will memorize your card numbers for you. Might your bank information get cribbed or hacked? Has it already been? Yeah, probably so, but this is How We Live Now. Its weird to think it was ever not normal.

This little historic tale is how Wojcicki addresses the current hesitance a majority of people feel to submitting their genetic material to a database by spitting into a tube and mailing that tube off to a companys lab. In a brief discussion at the TIME 100 Health Summit in New York City on Thursday, Wojcicki explained that shes actually grateful for the 2030 percent of the population thats down to spit (according to a Twitter poll TIME ran ahead of Wojcickis talk). At-home DNA kits like 23andMe are still a relatively new technology. People simply need to grow accustomed to the ideas of genetic testing and sharing genomic data with public databases, so that researchers can observe patterns across the population, and ultimately make people healthier, she said.

Its essentially an argument for sharing DNA data as good for public health, like a new-age equivalent of getting vaccinated. To drive the point home further, Eric Lander, a geneticist and director of the Broad Institute of MIT (and someone who once sat around a table with Jeffrey Epstein), mentioned a potential breakthrough in treating angiosarcoma, a rare, highly fatal cancer by using using DNAthat, Lander argued, may never have happened if it werent for peoples willingness to fork over health and genetic information.

Its hard to find a sensible, non-demonic argument against something that could lead to expedited breakthroughs in cancer treatment. But what has to be kept in mind is that 23andMe is a private company. These anecdotesvirtuous as they may soundare marketing techniques. Wojcicki rightfully believes that no singular institution will be able to harvest the amount of DNA that her company has. According to 23andMes About page, more than 10 million people have spit into its tubes and thereby handed their genetic information over. Of those 10 million, 80 percent have opted in to participate in research, via that spit data. For context: The National Institute of Health is currently in the midst of enrolling its largest DNA-related study to date (All Of Us), which will reach full enrollment at one million participants. About a year into their recruiting efforts, theyre about a quarter of the way there, according to what NIH Director Francis Collins said in a separate panel at Thursdays TIME 100 Health Summit.

Wojcicki emphasized that only those who opt in for ancestry information have their data entered into public databases, which are subject to subpoena (the likes of which helped identify the Golden State Killer). She further emphasized that, because DNA is highly similar among family members, submitting your spit implicates your relatives genetic information. (23andMe consents around this.) Thats a hell of a lot of data thats sitting around for seemingly forever, and since 23andMe is, once again, a private company, theres no telling what happens to this info if/when the company goes under, or if they decide to change their policies.

What 23andMe is sitting on now is perhaps the most valuable pool of genetic data in the world. Earlier this year, the company partnered with TrialSpark, an NYC-based research company, in order to use its large database of data to fit its opted-in consumers to studies. Mind you, Wojcicki used to work on Wall Street. She is, at her core, a businessperson. Charging customers to have their data, and then partnering with another company with an interest in that data, sounds lucrative as fuck; a hell of a business deal.

Wojcicki and Lander concluded their talk on Thursday with a heartwarming sentiment: The virtue of something like 23andMe is that consumers (or participants, however youd prefer to look at it) have access to their genetic data, rather than submitting it to a study and never getting feedback. Wojcicki refers to this as empowering; its empowering to know whats going on in the little strands that make you the person you are. It also sounds very empowering, monetarily speaking, for those in the game, marketing spit kits and trading (totally consented for!!) genetic data.

More here:
23andMe Wants Everyone to Get Used to Sharing Their Genetic Data - VICE

Read More...

National News Family tries to cure toddler with rare genetic disease Tomas Hoppough 9:36 AM – ABC15 Arizona

Monday, October 21st, 2019

A Denver family is trying to raise $3 million in order to cure their son with a rare genetic disease.

Doctors told Amber Freed that her 2-year-old son is one of 34 people in the world to have this rare neurological genetic disease.

The disease is so rare, it doesnt even have a name, Freed said. Its called SLC6A1, because that is the gene that it effects.

The disease causes Maxwell to have trouble moving and communicating, and soon it will only get worse.

The most debilitating part of the disease will begin between the ages of 3 and 4, Freed said. So, we are in a fight against time.

Maxwell has a twin sister named Riley.

I noticed early on that Maxwell wasnt progressing as much as Riley, Freed said. I noticed he couldnt use his hands. The doctors told me that every baby can use their hands. Thats when I realized there was something wrong with him.

After multiple visits to the doctor, Freed was able to find a genetic specialist to give Maxwell a diagnosis.

He looked at me and said, Something is very wrong with your son. I dont know if hes going to live, Freed said. My soul was just crushed. It was a sadness I didnt even know existed on earth. You never think something like this could happen. I left my career, and I had no other choice but to create my own miracle and to find a treatment forward to help Maxwell and all those others like him.

Freed searched for scientists trying to create a cure, which she found at the University of Texas Southwestern Medical Center in Dallas.

Were working with diseases where kids are born with a defective gene, said Steven Gray, an associate professor at UTSW in pediatrics. Our approach is to replace that gene to fix the condition at the level of their DNA. Were taking the DNA that those patients are missing and packaging that into a virus and use that virus as a molecular delivery truck to carry those genes back in their body and fix their DNA.

Its a rare disease, no one has ever heard of it, Freed said. But one rare disease messed with the wrong mother.

Freed said she has raised $1 million to help with research for the cure and will need an additional $3 million, in order to let Maxwell and many others continue to enjoy life.

I want Maxwell to have every opportunity that children should have in this life, Freed said. When he is having a good day, I just try and soak him in as much as I can. We dont take anything for granted in this house.

If you want to help donate for the cure, you can do so by visiting MilestonesforMaxwell.org or click on this GoFundMe page.

Read more here:
National News Family tries to cure toddler with rare genetic disease Tomas Hoppough 9:36 AM - ABC15 Arizona

Read More...

$2.1 million drug approved for Pearland toddler with rare genetic disease – KHOU.com

Monday, October 21st, 2019

PEARLAND, Texas A 2-year-old whose family has been fighting for a $2 million drug to battle her rare disease received approval for the drug Friday.

Krista James has Spinal Muscular Atrophy, or SMA, a life-threatening condition that severely impacts kids muscle movements. Her family has been fighting for the toddler to receive Zolgensma, a cutting-edge gene therapy that treats the disease at the genetic level.

RELATED: Pearland family fighting to get $2.1 million drug for toddler with rare genetic disease

Zolgensma is the most expensive drug to ever receive FDA approval. The treatment is priced at $2.125 million.

Despite Kristas doctor telling Medicaid its what the toddler needs, the request for coverage was denied until Friday, which happened to be Kristas 2nd birthday. Texas Health and Human Services approved the family's appeal.

The family told KHOU 11 they are beyond blessed and considered the approval the best birthday present ever.

ALSO POPULAR ON KHOU.COM

Send us a news tip | Download our app |#HTownRush Newsletter

Continue reading here:
$2.1 million drug approved for Pearland toddler with rare genetic disease - KHOU.com

Read More...

UCLH and UCL begin trial of genetic analysis in blood pressure patients – University College London Hospitals

Monday, October 21st, 2019

While this has been done as part of a number of studies in the past, the UCLH and UCL approach is unique in that they will also be returning this information to the patients as well as the doctors to understand how this information is useful and how it should be best communicated with doctors and patients.

For the UCLH BRCs AboutMe initiative, researchers will analyse patients DNA and other personal biological information to try to understand why some people have high blood pressure and others dont, and if it could be used to tailor treatment and whether it would be feasible for such testing to be rolled out across the NHS for other disease areas.

Some NHS patients have already had their DNA sequenced as part of the 100,000 Genomes Project done by Genomics England, but that project has so far focussed on rare diseases and cancer. UCLH and UCL researchers aim to extend this sequencing effort for common conditions. This effortand return of genomic information has not been done in the NHS before, within routine care.

The ultimate aim is that any NHS patient could have their DNA analysed in order to predict their chance of developing different conditions, prevent or reduce their risk of disease, and diagnose patients earlier with data returned to patients in a personalised report. Data from patients may also help in future with the development of drugs or tailoring of treatment.

For the initial study, research processes will be carried out in parallel with routine care to minimise disruption to patients. For instance, the doctor will seek the patients consent in the course of a clinic appointment, and the phlebotomist will take blood samples for research at the time of taking samples for standard tests. Blood samples will be processed by UCLHs laboratory.

The BRCs AboutMe initiative seeks to embed genomics and testing of other personal biological information into the NHS.

Researchers will be working with patients to look at how genetic information, including information on the risk of developing different conditions, can be best communicated to patients within the NHS environment.

See more here:
UCLH and UCL begin trial of genetic analysis in blood pressure patients - University College London Hospitals

Read More...

Food As Medicine: What Biochemistry And Genetics Are Teaching Us About How To Eat Right – Forbes

Thursday, October 10th, 2019

We often talk about genetics as if its set in stone. She just has good genes or He was born with it are common phrases.

However, over the past decade, biochemists and geneticists have discovered that your genetic expression changes over time. Based on environmental factors, certain genes may be strongly expressive while others are dormant.

In fact, a 2016 study of human longevity found that only 25% of health outcomes are attributable to genetics. The other 75% of outcomes are attributable to environmental factors. Among those environmental factors, diet and nutrition play a major role.

An entire branch of scientific research has now exploded around nutrigenomics, the study of the interaction between nutrition and genetics. Scientists now understand that genes set the baseline for how your body can function, but nutrition modifies the extent to which each gene is expressed.

As more data comes in about the types and quality of food that improve health outcomes, high-tech farmers are also entering the nutrigenomics conversation. Using precision agriculture, they hope to produce food thats targeted to deliver a nutrient-rich, genetically beneficial diet.

Implications Of Nutrigenomics

Researchers have found that theres no such thing as a perfect diet. Dietary recommendations are not one-size-fits-all. Each individual needs different nutritional choices for optimal health and gene expression. In addition, each person is different in the extent to which their genes and health are impacted by their diet.

Geneticists and nutritionists are working together to study the dietary levers that most impact genetic expression. If theyre successful, it may be possible to prevent and treat disease through individualized nutrition tailored to your genetic profile. Indeed, you may walk into a doctors office and leave with a dietary prescription customized to your DNA.

In the near future, instead of diagnosing and treating diseases caused by genome or epigenome damage, health care practitioners may be trained to diagnose and nutritionally prevent or even reverse genomic damage and aberrant gene expression, reports Michael Fenech, a research scientist at CSIRO Genome Health and Nutrigenomics Laboratory.

The initial results of nutrigenomics studies are promising. A healthy, personalized diet has the potential to prevent, mitigate, or even cure certain chronic diseases. Nutrigenomics has shown promise in preventing obesity, cancer and diabetes.

If Food Is Medicine, Food Quality Matters

Nutrient abundance or deficiency is the driving factor behind nutrigenomics. Foods that have grown in poor conditions have a lower nutritional density. In turn, eating low-quality foods can have a significant impact on human gene expression. In order to take advantage of the findings of nutrigenomics, consumers need access to high-quality, nutrient-dense foods.

Similar to human health, plant health is impacted by the combination of genes and nutrient intake. Healthy soil, correctly applied fertilization techniques, and other forms of environmental management lead to healthy crops.

However, applying these custom growing techniques at a large scale is a major challenge. Agriculture technology (AgTech) will play a big role in allowing farmers to precisely manage the growing conditions and nutrient delivery for their crops. In turn, this precision farming will make crops more nutritious and targeted for nutrigenomics-driven diets.

Making Food Thats Better For Us

Plant health relies on nutrient uptake from the soil. In order to ensure plants receive the nutrients they need, farmers need to precisely apply additives where theyre needed. With in-ground sensors, advanced mapping of crop quality across a field, and other technologies, farmers can target their applications of water and nutrients to match plant needs. The days of broadly applying generic fertilizer to entire fields are coming to an end.

Farmers play an integral role in providing access to diverse, nutritious food, explains Remi Schmaltz, CEO of Decisive Farming. Nutrient deficiency in plants and the soil can contribute to the deficiencies found in humans. The opportunity exists to address these deficiencies through precision nutrition delivered by the agriculture sector.

Additionally, CRISPR and other technologies allow us to experiment with the genetic makeup of plants, increasing nutrition and flavor, both pluses for consumers. In recent years, genetic modification has produced disease-resistant bananas, more flavorful tomatoes, lower gluten wheat, non-browning mushrooms and sustainable rice. While there has been a lot of skepticism over genetically-modified crops, multiple studies have shown that GMOs are safe for consumption and can even improve plant health and nutrition.

Using Biochemistry And Big Data To Create Better Food And Healthier People

Nutrigenomics will completely change how we think about health and disease prevention. Indeed, personalized diet recommendations that are tailored to your genes could be a new form of medicine for chronic illnesses.

Nevertheless, a key part of making nutrigenomics effective is having access to high-quality, nutrient-dense foods. AgTech is using the internet of things, AI, precision farming and gene editing to make nutrient-dense food more readily available. The benefits to public health from these efforts could change the way we think about medicine, longevity and what it means to be healthy.

Continue reading here:
Food As Medicine: What Biochemistry And Genetics Are Teaching Us About How To Eat Right - Forbes

Read More...

The thorny ethics of collecting genetic data – Quartz

Thursday, October 10th, 2019

In 2009, researchers collected DNA from four elderly men in Namibia, each from one of the many San indigenous communities scattered across southern Africa. A year later, analyses of the mens DNA were published in the journal Naturealongside that of South African human rights activist Desmond Tutu. The intention, in part, was to increase the visibility of southern, indigenous Africans in genetic-based medical research. Soon after, a nongovernmental organization representing indigenous minorities in Southern Africa took issue with the consent procedures used to gather the data and wrote to Natures editors accusing the papers authors of absolute arrogance, ignorance, and cultural myopia.

The San case highlights the thorny ethics of collecting genetic data. Yet today, to make medicine more equitable, scientists see the importance of sampling DNA from more diverse populations. Most genetic research uses DNA from descendants of Europeans, which means the related medical applicationssuch as genetic tests to see the likelihood of developing a certain disease, called polygenic risk assessments can only benefit those populations. In 2018 in the United States, for example, the National Institutes of Health launched All of Us, a research program that aims to collect DNA, electronic health records, and other data, from about one million Americans with emphasis on including many different groups of people.

When we do genetic studies, trying to understand the genetic basis of common and complex diseases, were getting a biased snapshot, said Alicia Martin, a geneticist at the Massachusetts General Hospital and the Broad Institute, a biomedical and genomics research center affiliated with Harvard and MIT.

Research to capture these snapshots, called genome-wide association studies, can only draw conclusions about the data thats been collected. Without studies that look at each underrepresented population, genetic tests and therapies cant be tailored to everyone. Still, projects intended as correctives, like All of Us and the International HapMap Project, face an ethical conundrum: Collecting that data could exploit the very people the programs intend to help.

Researchers with All of Us have already collected data from about 1,600 Native Americans, some of whom live in cities outside of sovereign lands, where tribal approval is not necessary for genetic research, according to Krystal Tsosie, a geneticist at Vanderbilt University who is co-leading a study in collaboration with a tribal community in North Dakota .

Obviously theres an interest in monetizing biomarkers collected from diverse populations and underrepresented populations, Tsoise said, so without adequate protections, the concern becomes about exploitation.

Medical genetic research generally works like this: Geneticists use powerful computers to compare the genomes of people affected by a particular disease to healthy controls. Researchers mark genetic patterns that are common in people with, say, diabetes, but not the controls, as associated with the disease. The more samples geneticists feed to the algorithms, the more likely that the findings reflect reality.

But studies restricted to descendants of Europeans will only find associations between diseases and variants that are common in European ancestry populations, said Martinif those variants are common enough to be found.

Scientists use the results to develop polygenic risk scores, which count the risky variants on someones genome to estimate their susceptibility to a disease. But if studies dont use the genomes of non-white populations, the tests wont pick up on the problematic variants in different groups of people. One 2019 Nature Genetics study, on which Martin was an author, determined that these blind spots reduce the accuracy of polygenic tests by approximately two and five times in South or East Asian, and black populations, respectively.

In many cases, the groups whose DNA is missing have worse health care outcomes compared to their white counterparts, and genetic medicine could worsen these disparities. I think theres a huge responsibility, said Martin. If we look at the history of the field, over the past decade weve gone from participants in genetic studies being 96% European ancestry to about 80 percent. Weve shifted gears a little bit, but not nearly enough to be able to serve minority populations. Jantina De Vries, a bioethicist at the University of Cape Town, agreed that representation in genomics research can bring health benefits, particularly if it is paired with measures to build research capacity so that, eventually, there are researchers at every level within the groups themselves.

Collecting broader genetic samples poses a host of challenges. Efforts to collect and study the genomes of indigenous peoples, for example, have been controversial since the early 90s. The first such project, called the Human Genome Diversity Project (HGDP), was meant to explore the full range of genome diversity within the human family by collecting DNA samples from about 500 distinct groups, with an emphasis on indigenous peoples that might soon vanish. Indigenous-rights organizations criticized the project, taking issue with being treated as mere objects of scientific interest and potential for commercialization. All of Us, more recently, has run into similar objections from the National Congress of American Indians.

The concerns are linked to the long history of exploitative encounters between researchers and vulnerable populations. The Tuskegee Studyin which the US Public Health Service withheld treatment from African American men with syphilislasted from 1932 until 1972, ending less than 20 years before the HGDP proposal. And in 1989, researchers from Arizona State University collected DNA samples from the Havasupai Tribe and reused the data for research to which the participants hadnt consented: on schizophrenia, inbreeding, and migration history. Tsosie said this context has created a climate in which weve seen tribes deciding to disengage from biomedical research completely.

All the geneticists and ethicists Undark spoke with agreed that community engagement is crucial to establish trust. But they didnt agree on the degree of the engagement. Some believed that gaining the consent of communities is necessary for ethical research, while others said it was enough to have respect and open dialogue between researchers and the people theyd like to study.

But both approaches are difficult in the context of collecting and analyzing genetic data, since geneticists take DNA from individuals to make conclusions about entire populations. For instance, the San paper in Nature extrapolated findings regarding individual genomes to discuss the genomes of the broader communities. Ones genome is not their own specifically; ones genome is informed by their recent ancestry, their family structures, and their more distant ancestry, said Tsosie. Geneticists are never talking about an individual thats siloed.

The gap between individual and collective consent is partly responsible for the continued friction between genetic science and indigenous peoples. Collective consent, said Tsosie, who is herself Navajo, is more culturally consistent with how tribal groups govern themselves. In 2017, Andries Steenkamp, a San leader, and Roger Chennells, a lawyer, wrote that the Nature study failed in this regard by only getting informed consent from the indigenous individuals who participated.

Not everyone agrees that collective consent can or should be a requirement for all genetic studies. For instance, de Vries said, it depends what sort of community were talking about, drawing a contrast between small, rural, communities and larger populations spread across several cities or countries. If were talking about the entire Yoruba population, who would you even talk to? she added. The Yoruba are an ethic group of more than 20 million individuals, most of whom live in Nigeria, with smaller populations in Benin, Togo, and across several diaspora communities. De Vries believes the onus lies on researchers to think in terms of respecting communities, rather than in terms of collective consent.

Gaining collective consent involves logistical hurdles, especially for large-scale projects. The NIHs All of Us program, for example, wasnt able to get input from each of the 573 federally-recognized tribes. According to Tsosie, during the planning stages, there was talk of gaining tribal input, but that plan seemed to be abandoned early on. The All of Us website does have a section on tribal engagement, but only offers formal consultation and listening sessions for ongoing projects, not guidance on how to approach these issues before a project starts.

Among non-indigenous policymakers and scientists, Tsosie noted, theres a magical notion that stakeholders from every tribe can be brought together in one room when, in reality, that is not how we make consensus decisions for ourselves.

Even more difficult than logistics, perhaps, may be conceptualizing the genetic studies to begin withfor example, deciding which people belong in which groups. One of the greatest political acts, acts of power, that we perform as human beings is dividing ourselves up for the purpose of knowing and governing ourselves, said Jenny Reardon, a sociologist who specializes in genomics at the University of California, Santa-Cruz.

Globally, indigenous peoples are so culturally distinct from one another that a single understanding of a community wont resonate with everyone. Finding a method for data collection that crosses all indigenous groups is going to be really hard, said Vanessa Hayes, a geneticist at the Garvan Institute of Medical Research and the University of Sydney in Australia who conducts fieldwork in South Africa. Because, straightaway, thats assuming all indigenous people are the same. Without common ground, scientists must do the hard work of understanding each unique community. As Hayes put it, every group that you work with, you have to respect that group, and take the time to understand what is important in that group.

Hayes was one of the authors of the 2010 Nature study on the San, and she was responsible for obtaining consent, gathering samples, and discussing the results with the community. While Steenkamp and Chennells suggested the researchers were hasty in their data collection and ignored governance structures, Hayes countered that, at the time of the study, shed already been working in these communities for more than a decade and they were working directly with government agencies. Shed been in contact with the Working Group of Indigenous Minorities in Southern Africa (WIMSA)the NGO which would eventually criticize the studybefore it began. But, she said, when I went back to the community and asked if they knew who WIMSA was, they said no. I asked them if they wanted WIMSA to represent them, and they said, Hell no.

(As an organization, WIMSA is currently being restructured. The South African San Council, which now represents the San communities of South Africa, declined an interview with Undark, citing a requirement for financial compensation and a signed contract.)

Hayes followed the principles of collective consent, she said, just at a lower level than formal institutions like WIMSA or the San Council: Their decision was made as a group. They are the group, they are the band, they are the family. She added, No one can represent them that is not them.

The difficulties in defining a group make collective consent even more challenging.

In the clearest of circumstances, where an established organization exists, approval processes can be difficult to navigate and can take months. But within some indigenous and minority groups, issues of representation remain controversial. Often, a scientist will have to invest a lot of time interacting with potential subjects in order to judge what consent procedures are appropriate. Few scientists have the necessary time and resources.

There is no easy way to choose which organizations to deal with, especially when there are internal disagreements about representation. Or, as Reardon put it: The folks that are trying to democratize the science are going to have the same problem as the people who were attempting to treat it as Were just going to go out and get these groups, and study them from a scientific perspective.

Although the repeated controversies surrounding research and indigenous groups may have slowed their inclusion in genetic science, the researchers Undark spoke with said ensuring these concerns are heard and addressed is a vital part of the work. Indigenous groups are demanding a greater say in research that concerns them, whether under the All of Us program or conducted by individual researchers in Africa. Resolving the ethical ambiguities is no easy task, but, as Hayes asked: Why should it be easy?

This article was originally published on Undark. Read the original article.

View original post here:
The thorny ethics of collecting genetic data - Quartz

Read More...

Innolea Created to Focus on Genetics and Genomics of Oil Crops – Seed World

Thursday, October 10th, 2019

Since January 1, 2019, the research activities in oilcrop genetics and genomics of the company Biogemma moved to a new company, Innolea, under the control of three major French seed companies in oil crops: Euralis Semences, Limagrain, RAGT Semences and the innovation fund for oil and protein crops, Sofiprotol.

Innolea is supported by experienced teams andknow-how developed for more than 20 years by Biogemma and benefit of theinfrastructure of Mondonville (Haute-Garonne) site, which is now the headquarter,as well as the support of its shareholders.

Innoleas direction was given to Bruno Grzes-Besset who was research coordinator in Biogemma, and the presidency to Jean-Marc Ferullo, the research director of Euralis Semences.

Biogemmas research programs in rapeseed and sunflower, the main oil crops in France and Europe, are continued in Innolea. The sunflower program is led by Delphine Fleury and the rapeseed program by Sbastien Faure.

To support the French oil industry, Innoleasactivities focus on the genetics of traits of interest with direct applicationsto breeding new varieties. The main research areas will be the exploration ofgenetic diversity of the main oil crops of France and Europe and the characterisationof beneficial genes.

By creating this new company, the French oil and protein industry and the seeds companies demonstrate their engagement in supporting competitive research in applied genomics and pre-breeding. This leading-edge knowledge will enable offering new varieties to the French market, with improved agronomy and particularly plant resistance to diseases and pests.

Read more:
Innolea Created to Focus on Genetics and Genomics of Oil Crops - Seed World

Read More...

Why is your dog so aggressive? The answer is largely in its DNA – CTV News

Thursday, October 10th, 2019

TORONTO -- Dog owners, take note: If you're fed up with your pet's personality, there may not be much you can do about it.

A new study confirms something many people have long suspected, namely that canine behaviours are largely based on the animals' genetic makeup.

Researchers at four American universities analyzed genetic information and behaviour logs for 14,000 dogs from 101 breeds and determined that 60 to 70 per cent of the differences in personality traits between breeds can be explained by the genes they inherited from their parents.

Genetics were found to contribute most strongly to traits such as trainability, aggression toward strangers and attention-seeking. This fits with the idea that these were some of the most or least sought-after attributes during the early stages of breeding, making them essentially hard-wired into the breeds' DNA.

Because most of these breeds have only been in existence for 300 years or less, they have not had enough time to develop the sort of genetic diversity seen in species with longer histories.

This helps explain why greyhounds and Siberian huskies are some of the least aggressive dogs toward strangers, for example, or why Yorkshire terriers and toy poodles have trouble coping with separation.

The researchers were also able to find 131 genetic variations that appear to be linked to breed behaviour. They say this suggests that no individual gene is solely responsible for dogs' personality traits, and instead the relationships between genes come into play just as they do with humans.

Even some of the genes located by the researchers as being associated with dogs' neurological development mirror similar genes that have been found in humans.

"Dogs exhibit striking parallels to traits in humans," the study reads.

"For example, common genetic mechanisms contribute to individual differences in social behaviour in dogs and humans."

This suggests that further study of dog genetics could help illustrate how personality traits develop in humans, and which traits are more or less likely to be inherited, the researchers said.

The study was published Oct. 1 in the Proceedings of the Royal Society B.

Read the original here:
Why is your dog so aggressive? The answer is largely in its DNA - CTV News

Read More...

Blue Devil of the Week: Searching for Answers in the Genetic Code – Duke Today

Thursday, October 10th, 2019

Name: Sue Jinks-Robertson

Position: Professor, Vice Chair and Director of the Cell and Molecular Biology Program

Years at Duke: 13

What she does at Duke: Jinks-Robertson has many duties in the Department of Molecular Genetics and Microbiology. She oversees the Cell and Molecular Biology Training Program which features around 100 graduate students and involves around 130 faculty members. She also co-directs the Cancer Genetics and Genomics program at Duke Cancer Institute.

But Jinks-Robertson is most at home in her lab, where she studies the genetic makeup of yeast.

Her team examines yeast DNA, looking for the factors behind mutations or changes in sequence. This research is important is because the DNA of yeast is essentially the same that found in many other organisms, including humans.

If we understand how this works in yeast, we can get information about what can go wrong in humans, Jinks-Robertson said.

The research is of great value in the fight against cancer, since it can occur when cells with genetic flaws multiply. Therapies that help identify and repair these flaws can be critical in battling the disease.

The basis for the therapies comes from the very basic work done in the trenches with an organism like yeast, Jinks-Robertson said.

What she loves about Duke: When she arrived at Duke after two decades on the faculty at Emory University, Jinks-Robertson was struck by the affection and loyalty her new colleagues both staff and faculty had for the university.

Soon, she too came to have a similar relationship with the university. She said its hard to pin down one specific reason for her connection with Duke, but she knows its there.

I dont know if its some of the physical structures around, like the Gardens, the Chapel, theres a central focus, of course theres basketball, its hard to put your finger on what it is, Jinks-Robertson said. Its just a very nice place to work. You feel connected to something bigger than yourself.

A memorable day at work: This spring, Jinks-Robertson was preparing for a major grant application when she got a call from colleague Thomas Petes with exciting news.

Petes, the Minnie Gellar Professor of Molecular Genetics and Microbiology, told her that shed been elected to the National Academy of Sciences, a 156-year old organization comprised of the nations top scientific minds.

It was a big surprise, Jinks-Robertson said. If youre a scientist, at least in this country, its a great recognition.

Jinks-Robertson was one of two Duke scientists elected to the 2019 class. Susan Alberts, the Robert F. Durden Professor of Biology, also earned election to the academy.

The nicest part of it was that I was hit with a flood of emails and phone calls, it was really wonderful, Jinks-Robertson said.

Special object/memorabilia in her workspace: On a shelf in her office, Jinks-Robertson has a collection of gifts given to her by former students who came to Duke from other countries. Theres a statue of Saraswati, the Hindu goddess of learning, which was given to her by a student from India. Theres also a vase from Russia, a screen from China and small house from the Philippines.

I like to think it shows I was successful in training the next generation, Jinks-Robertson said.

First ever job: A native of Panama City, Florida, Jinks-Robertson grew up around the water. As a child, she swam and water skied often. After she graduated from high school, she spent the next two summers working as a mermaid at Gulf World Marine Park, a popular attraction in Panama City.

We didnt have tails, but we had on scuba tanks and dove in saltwater tanks and fed the fish as people watched, Jinks-Robertson said of the mermaid role, which also had her swimming with dolphins and sea lions. It was fun.

Best advice received: In 1986, when she was finishing up her time as a post-doctoral researcher at the University of Chicago working with Thomas Petes, who many years later helped bring her to Duke, Jinks-Robertson began looking for faculty positions.

I was pregnant with my first child and I was concerned about that, Jinks-Robinson said. His advice was, If its a problem, its not a place you want to be. He really put me at ease and told me I shouldnt worry about that. Hes always been very supportive of women in science.

Something most people dont know about her: Much of Jinks-Robertsons work is done with a sleeping labradoodle at her feet. With soft, curly light brown hair, Gracie is Jinks-Robertsons constant companion, often accompanying her to work.

Its calming, Jinks-Robertson said. I walk her every day, so it gets me moving and out of my chair. Shes good company.

Is there a colleague at Duke who has an intriguing job or goes above and beyond to make a difference? Nominate that person for Blue Devil of the Week.

Read the original here:
Blue Devil of the Week: Searching for Answers in the Genetic Code - Duke Today

Read More...

Advances in the Diagnosis of Type 1 von Willebrand Disease: Genetic Testing – Hematology Advisor

Thursday, October 10th, 2019

Von Willebrand disease (VWD) is the most common hereditary bleeding disorder but one of the most difficult to diagnose, especially type 1 VWD. Recurrent challenges include the need to complete several assays of von Willebrand factor (VWF) activity and lack of consensus surrounding the acceptable standard for diagnosis. Consequently, improving current diagnostic techniques, as well as implementing new methods, is essential to ensure patients are provided optimal care.

In a review article published in Current Opinion in Hematology, Veronica H Flood, MD, of the department of pediatrics at the Medical College of Wisconsin in Milwaukee, and colleagues summarized the current literature surrounding the diagnosis of type 1 VWD. They also reviewed new advances in genetic testing for VWF, which could serve as a potential alternative to conventional laboratory methods.

Overview of Genetic Dysfunction

In contrast to type 2 VWD, type 1 VWD may include genetic defects in the coding region of the VWF gene. These mutations vary from insertions and deletions to point mutations that produce missense or nonsense mutations. With conventional sequencing methods, insertions and deletions can be missed, which has historically precluded the clinical use of genetic-based diagnostic techniques. These limitations are not typically seen in type 2 VWD as genetic defects are usually present in the DNA region specific to the impacted functional region.

Because of the high degree of polymorphism seen in the VWF gene, entire genome or exome sequencing may be required for diagnosis; in other instances, the VWF gene may be analyzed directly if a particular coagulation defect is suspected. In type 1 VWD, certain high frequency variants have been linked to disease etiology, but recent data have highlighted the potential role of novel variants in type 1 VWD. The high degree of variability seen in the VWF gene is certainly a key contributor to the disease phenotype, but not all defects will ultimately lead to VWD.

Modifier Genes and Diagnosis

In addition to defects in VWF, several genes independent of the VWF locus have been shown to affect VWF levels. The most described modifier gene is ABO, though others such as CLEC4M, STAB2, and STXBP5 also exist. Blood group O levels of less than 50 IU/dL are routinely used to establish a diagnosis of VWD, but some individuals with blood type O also meet this criteria despite being healthy. Some experts have proposed that low VWF may be more suitably described as a risk factor for bleeding instead of as the basis for bleeding.

Originally posted here:
Advances in the Diagnosis of Type 1 von Willebrand Disease: Genetic Testing - Hematology Advisor

Read More...

Datar Cancer Genetics Announces Positive Results With 42.9% Objective Response Rate, and 90.5% Disease Control Rate in Heavily Pre-treated Patients…

Thursday, October 10th, 2019

LONDON, MUMBAI, India and BAYREUTH, Germany, Oct. 9, 2019 /PRNewswire/ --

- RESILIENT Protocol captures Encyclopedic information from tumors and uses Artificial Intelligence (AI) to optimize treatments;

- Objective Response and Disease Control Comparable to or better than most Immunotherapy options;

- Unique Ultra-personalised combination of drugs already approved for cancer;

- RESILIENT Protocol commercially launched

Datar Cancer Genetics, a cancer research company, today announced positive data from the phase II/III RESILIENT trial intended to validate clinical benefit for cancer patients who have run out of treatment options under the present standard of care guidelines. The study achieved its primary and secondary end points of Objective Response Rate, Progression Free Survival and Disease Control Rate respectively.

Drug resistant cancers present a serious clinical challenge since there are virtually no treatments available and the prognosis is invariably poor. As a large proportion of all patients with advanced cancers ultimately progress towards this phase, life extending treatment options for these patients are urgently required.

The RESILIENT Protocol is designed to analyse all functional layers of a cancer cell i.e., DNA, RNA, proteins and germline genetics as also the chemoresistance/sensitivity of live tumour cells. This data is integrated through an Artificial Intelligence algorithm to derive treatment regimens which are most efficacious and yet show the least risk of toxicity.

RESILIENT is the world's first and only prospective Precision Oncology trial where drug combinations were selected on multi-analyte integration. Most prior trials based on a single molecular alteration for drug selection had dismal outcomes. 143 patients started treatment and 126 patients were evaluable as per study criteria. All patients underwent PET-CT and Brain MRI scans prior to start of treatment to establish extent of disease. Treatment response was determined by follow-up PET-CT and MRI scans.

In the majority (90.5%) of patients, further spread of cancer was effectively halted. In 42.9% of patients, treatments also led to a significant decrease in the extent of cancer. Remarkably, among the 12 patients where disease progressed, no new metastases were reported in 9 patients. There were no serious treatment related adverse events or deaths. Most patients reported stable to improved quality of life.

The data of the RESILIENT Trial is published in the peer revived oncology journal 'Oncotarget' (https://doi.org/10.18632/oncotarget.27188)

"The RESILIENT trial marks a watershed moment for molecular oncology as it unequivocally proves that patients who have failed even 2 to 3 lines of treatment can benefit from already approved drugs if comprehensive tumour analysis is used to guide treatments. Patients in the United Kingdom and all over the world have much to gain from the outcome of this trial," said Dr. Tim Crook, Medical Oncologist at the St. Luke's Cancer Centre, Royal Surrey County Hospital, Guildford, UK, who is one of the authors of the publication.

Datar is a leading cancer research corporation specialising in tumour analysis for better diagnosis, treatment decisions, and management of cancer. Datar's research initiatives are poised to bring about meaningful, patient-friendly and practice changing advancements in cancer treatment. Datar is also pursuing Adoptive Cell Immunotherapy for Multiple Solid Organ Cancers.

For more information please contact:

Dr. Vineet Datta - drvineetdatta@datarpgx.com

Dr Stefan Schuster - drstefanschuster@datarpgx.com

Datar Cancer Genetics info@datarpgx.org

Related Links

https://doi.org/10.18632/oncotarget.27188

https://doi.org/10.1093/annonc/mdz268.061

Encyclopedic Tumor Analysis – Exacta

SOURCE Datar Cancer Genetics Ltd

Read the rest here:
Datar Cancer Genetics Announces Positive Results With 42.9% Objective Response Rate, and 90.5% Disease Control Rate in Heavily Pre-treated Patients...

Read More...

Genetics of Diabetes | ADA

Thursday, October 3rd, 2019

You've probably wondered how you developed diabetes. You may worry that your children will develop it too.

Unlike some traits, diabetes does not seem to be inherited in a simple pattern. Yet clearly, some people are born more likely to develop diabetes than others.

Type 1 andtype 2 diabeteshave different causes. Yet two factors are important in both. You inherit a predisposition to the disease, then something in your environment triggers it.

Genes alone are not enough. One proof of this is identical twins. Identical twins have identical genes. Yet when one twin hastype 1 diabetes, the other gets the disease at most only half the time. When one twin has type 2 diabetes, the other's risk is at most 3 in 4.

In most cases of type 1 diabetes, people need to inherit risk factors from both parents. We think these factors must be more common in whites because whites have the highest rate of type 1 diabetes.

Because most people who are at risk do not get diabetes, researchers want to find out what the environmental triggers are. One trigger might be related to cold weather. Type 1 diabetes develops more often in winter than summer and is more common in places with cold climates. Another trigger might be viruses. Perhaps a virus that has only mild effects on most people triggers type 1 diabetes in others.

Early diet may also play a role. Type 1 diabetes is less common in people who were breastfed and in those who first ate solid foods at laterages.

In many people, the development of type 1 diabetes seems to take many years. In experiments that followed relatives of people with type 1 diabetes, researchers found that most of those who later got diabetes had certain autoantibodies in their blood for years before. (Antibodies are proteins that destroy bacteria or viruses. Autoantibodies areantibodies'gone bad' that attack the body's own tissues.)

If you are a man with type 1 diabetes, the odds of your child developing diabetes are 1 in 17. If you are a woman with type 1 diabetes and your child was born before you were 25, your child's risk is 1 in 25; if your child was born after you turned 25, your child's risk is 1 in 100.

Your child's risk is doubled if you developed diabetes before age 11. If both you and your partner have type 1 diabetes, the risk is between 1 in 10 and 1 in 4.

There is an exception to these numbers. About 1 in every 7 people with type 1 diabetes has a condition called type 2 polyglandular autoimmune syndrome. In addition to having diabetes, these people also have thyroid disease and a poorly working adrenalgland. Some also have otherimmune systemdisorders. If you have this syndrome, your child's risk of getting the syndromeincluding type 1 diabetesis 1 in 2.

Researchers are learning how to predict a person's odds of getting diabetes. For example, most whites with type 1 diabetes have genes called HLA-DR3 or HLA-DR4. If you and your child are white and share these genes, your child's risk is higher. (Suspect genes in other ethnic groups are less well studied. The HLA-DR7 gene may put African Americans at risk, and the HLA-DR9 gene may put Japanese at risk.)

Other tests can also make your child's risk clearer. A special test that tells how the body responds toglucosecan tell which school-aged children are most at risk.

Another more expensive test can be done for children who have siblings with type 1 diabetes. This test measures antibodies toinsulin, to islet cells in thepancreas, or to anenzymecalled glutamic acid decarboxylase. High levels can indicate that a child has a higher risk of developing type 1 diabetes.

Type 2 diabetes has a stronger link to family history and lineage than type 1, and studies of twins have shown that genetics play a very strong role in the development of type 2 diabetes.

Yet it also depends on environmental factors.Lifestyle also influences the development of type 2 diabetes.Obesitytends to run in families, and families tend to have similar eating and exercise habits.

If you have a family history of type 2 diabetes, it may be difficult to figure out whether your diabetes is due to lifestyle factors or genetic susceptibility. Most likely it is due to both. However, dont lose heart. Studies show that it is possible to delay or prevent type 2 diabetes by exercising and losing weight.

Have you recently been diagnosed with type 2 diabetes?Join our free Living With Type 2 Diabetes program and get the information and support you need to live well with diabetes.

Type 2 diabetes runs in families. In part, this tendency is due to children learning bad habitseating a poor diet, not exercisingfrom their parents. But there is also a genetic basis.

If you would like to learn more about the genetics of all forms of diabetes, the National Institutes of Health has publishedThe Genetic Landscape of Diabetes. This free online book provides an overview of the current knowledge about the genetics of type 1 and type 2 diabetes, as well other less common forms of diabetes. The book is written for healthcare professionals and for people with diabetes interested in learning more about the disease.

Continue reading here:
Genetics of Diabetes | ADA

Read More...

Ph.D. in Genetics at Texas A&M University

Thursday, October 3rd, 2019

Please save the date for oursecond Career Club of the fall semester!We are very excited to listen to Dr. Robyn Baldens talk about: Medical Science Liaison and other opportunities at Merck nextFriday,September 20th at12:00 PMinNMR/Rm. N127

Dr. Balden is a physician scientist and Regional Medical Scientific Director for Anesthesia/Surgery, South/Central US Medical Affairs division of Merck Research Labs. This role integrates internal and external scientific exchange and collaboration in order to facilitate and support clinical and drug development programs and maximize patient safety and outcomes related to existing pharmaceuticals including clinical trials, investigator-initiated studies, medical education, and scientific content creation.Her role at Merck began in 2018 as Associate Director, Medical Science Liaison for Anesthesia/Surgery, South/Central US, subsequent to gaining experience conducting medical research and directing business development for clinical trials at the Texas Center for Drug Development in Houston, TX. At the Texas Center for Drug Development she engaged in medical affairs focusing on coordination of clinical research for various therapeutic areas, serving as a supporting investigator for clinical trials, scientific discussion and account management with key physician leaders, and development of medical educational materials. Prior to this role she was a surgical intern, resident anesthesiologist, and clinical scholar at Cedars-Sinai Medical Center in Los Angeles, CA, where she initiated clinical studies for novel anesthetic regimens.

Dr. Balden received her MD and PhD in Neuroscience from Texas A&M Health Science Center College of Medicine. Her passions involve the intersection of medicine and science with neuroimmunology and neuroendocrinology. She also collaborates with advocacy and student organizations, has written several academic papers on Vitamin D, and served as a member of the Vitamin D Councils Board of Directors contributing as a volunteer writer, podcast contributor, and graphic designer for the Vitamin D Council. Shelives with her family in Houston, TX and enjoys painting, design, traveling, scuba diving, outdoors, live music, reading, cooking, and gardening.

Continue reading here:
Ph.D. in Genetics at Texas A&M University

Read More...

Efficient Implementation of Penalized … – genetics.org

Monday, May 6th, 2019

Abstract

Polygenic Risk Scores (PRS) combine genotype information across many single-nucleotide polymorphisms (SNPs) to give a score reflecting the genetic risk of developing a disease. PRS might have a major impact on public health, possibly allowing for screening campaigns to identify high-genetic risk individuals for a given disease. The Clumping+Thresholding (C+T) approach is the most common method to derive PRS. C+T uses only univariate genome-wide association studies (GWAS) summary statistics, which makes it fast and easy to use. However, previous work showed that jointly estimating SNP effects for computing PRS has the potential to significantly improve the predictive performance of PRS as compared to C+T. In this paper, we present an efficient method for the joint estimation of SNP effects using individual-level data, allowing for practical application of penalized logistic regression (PLR) on modern datasets including hundreds of thousands of individuals. Moreover, our implementation of PLR directly includes automatic choices for hyper-parameters. We also provide an implementation of penalized linear regression for quantitative traits. We compare the performance of PLR, C+T and a derivation of random forests using both real and simulated data. Overall, we find that PLR achieves equal or higher predictive performance than C+T in most scenarios considered, while being scalable to biobank data. In particular, we find that improvement in predictive performance is more pronounced when there are few effects located in nearby genomic regions with correlated SNPs; for instance, in simulations, AUC values increase from 83% with the best prediction of C+T to 92.5% with PLR. We confirm these results in a data analysis of a case-control study for celiac disease where PLR and the standard C+T method achieve AUC values of 89% and of 82.5%. Applying penalized linear regression to 350,000 individuals of the UK Biobank, we predict height with a larger correlation than with the best prediction of C+T (65% instead of 55%), further demonstrating its scalability and strong predictive power, even for highly polygenic traits. Moreover, using 150,000 individuals of the UK Biobank, we are able to predict breast cancer better than C+T, fitting PLR in a few minutes only. In conclusion, this paper demonstrates the feasibility and relevance of using penalized regression for PRS computation when large individual-level datasets are available, thanks to the efficient implementation available in our R package bigstatsr.

POLYGENIC risk scores (PRS) combine genotype information across many single-nucleotide polymorphisms (SNPs) to give a score reflecting the genetic risk of developing a disease. PRS are useful for genetic epidemiology when testing polygenicity of diseases and finding a common genetic contribution between two diseases (Purcell et al. 2009). Personalized medicine is another major application of PRS. Personalized medicine envisions to use PRS in screening campaigns in order to identify high-risk individuals for a given disease (Chatterjee et al. 2016). As an example of practical application, targeting screening of men at higher polygenic risk could reduce the problem of overdiagnosis and lead to a better benefit-to-harm balance in screening for prostate cancer (Pashayan et al. 2015). However, in order to be used in clinical settings, PRS should discriminate well enough between cases and controls. For screening high-risk individuals and for presymptomatic diagnosis of the general population, it is suggested that, for a 10% disease prevalence, the AUC must be >75% and 99%, respectively (Janssens et al. 2007).

Several methods have been developed to predict disease status, or any phenotype, based on SNP information. A commonly used method often called P+T or C+T (which stands for Clumping and Thresholding) is used to derive PRS from results of Genome-Wide Association Studies (GWAS) (Wray et al. 2007; Evans et al. 2009; Purcell et al. 2009; Chatterjee et al. 2013; Dudbridge 2013). This technique uses GWAS summary statistics, allowing for a fast implementation of C+T. However, C+T also has several limitations; for instance, previous studies have shown that predictive performance of C+T is very sensitive to the threshold of inclusion of SNPs, depending on the disease architecture (Ware et al. 2017). In parallel, statistical learning methods have also been used to derive PRS for complex human diseases by jointly estimating SNP effects. Such methods include joint logistic regression, Support Vector Machine (SVM) and random forests (Wei et al. 2009; Abraham et al. 2012, 2014; Botta et al. 2014; Okser et al. 2014; Lello et al. 2018; Mavaddat et al. 2019). Finally, Linear Mixed-Models (LMMs) are another widely used method in fields such as plant and animal breeding, or for predicting highly polygenic quantitative human phenotypes such as height (Yang et al. 2010). Yet, predictions resulting from LMM, known e.g., as gBLUP, have not proven as efficient as other methods for predicting several complex diseases based on genotypes [see table 2 of Abraham et al. (2013)].

We recently developed two R packages, bigstatsr and bigsnpr, for efficiently analyzing large-scale genome-wide data (Priv et al. 2018). Package bigstatsr now includes an efficient algorithm with a new implementation for computing sparse linear and logistic regressions on huge datasets as large as the UK Biobank (Bycroft et al. 2018). In this paper, we present a comprehensive comparative study of our implementation of penalized logistic regression (PLR), which we compare to the C+T method and the T-Trees algorithm, a derivation of random forests that has shown high predictive performance (Botta et al. 2014). In this comparison, we do not include any LMM method, yet, L2-PLR should be very similar to LMM methods. Moreover, we do not include any SVM method because it is expected to give similar results to logistic regression (Abraham et al. 2012). For C+T, we report results for a large grid of hyper-parameters. For PLR, the choice of hyper-parameters is included in the algorithm so that we report only one model for each simulation. We also use a modified version of PLR in order to capture not only linear effects, but also recessive and dominant effects.

To perform simulations, we use real genotype data and simulate new phenotypes. In order to make our comparison as comprehensive as possible, we compare different disease architectures by varying the number, size and location of causal effects as well as disease heritability. We also compare two different models for simulating phenotypes, one with additive effects only, and one that combines additive, dominant and interaction-type effects. Overall, we find that PLR achieves higher predictive performance than C+T except in highly underpowered cases (AUC values lower than 0.6), while being scalable to biobank data.

We use real genotypes of European individuals from a case-control study for celiac disease (Dubois et al. 2010). This dataset is presented in Supplemental Material, Table S1. Details of quality control and imputation for this dataset are available in Priv et al. (2018). For simulations presented later, we first restrict this dataset to controls from UK in order to remove the genetic structure induced by the celiac disease status and population structure. This filtering process results in a sample of 7100 individuals (see supplemental notebook preprocessing). We also use this dataset for real data application, in this case keeping all 15,155 individuals (4496 cases and 10,659 controls). Both datasets contain 281,122 SNPs.

We simulate binary phenotypes using a Liability Threshold Model (LTM) with a prevalence of 30% (Falconer 1965). We vary simulation parameters in order to match a range of genetic architectures from low to high polygenicity. This is achieved by varying the number of causal variants and their location (30, 300, or 3000 anywhere in all 22 autosomal chromosomes or 30 in the HLA region of chromosome 6), and the disease heritability (50 or 80%). Liability scores are computed either from a model with additive effects only (ADD) or a more complex model that combines additive, dominant and interaction-type effects (COMP). For model ADD, we compute the liability score of the i-th individual aswhere is the set of causal SNPs, are weights generated from a Gaussian distribution or a Laplace distribution , is the allele count of individual i for SNP j, corresponds to its standardized version (zero mean and unit variance for all SNPs), and follows a Gaussian distribution . For model COMP, we simulate liability scores using additive, dominant and interaction-type effects (see Supplemental Materials).

We implement three different simulation scenarios, summarized in Table 1. Scenario N1 uses the whole dataset (all 22 autosomal chromosomes 281,122 SNPs) and a training set of size 6000. For each combination of the remaining parameters, results are based on 100 simulations except when comparing PLR with T-Trees, which relies on five simulations only because of a much higher computational burden of T-Trees as compared to other methods. Scenario N2 consists of 100 simulations per combination of parameters on a dataset composed of chromosome six only (18,941 SNPs). Reducing the number of SNPs increases the polygenicity (the proportion of causal SNPs) of the simulated models. Reducing the number of SNPs (p) is also equivalent to increasing the sample size (n) as predictive power increases as a function of (Dudbridge 2013; Vilhjlmsson et al. 2015). For this scenario, we use the additive model only, but continue to vary all other simulation parameters. Finally, scenario N3 uses the whole dataset as in scenario N1 while varying the size of the training set in order to assess how the sample size affects predictive performance of methods. A total of 100 simulations per combination of parameters are run using 300 causal SNPs randomly chosen on the genome.

In this study, we use two different measures of predictive accuracy. First, we use the Area Under the Receiver Operating Characteristic (ROC) Curve (AUC) (Lusted 1971; Fawcett 2006). In the case of our study, the AUC is the probability that the PRS of a case is greater than the PRS of a control. This measure indicates the extent to which we can distinguish between cases and controls using PRS. As a second measure, we also report the partial AUC for specificities between 90 and 100% (McClish 1989; Dodd and Pepe 2003). This measure is similar to the AUC, but focuses on high specificities, which is the most useful part of the ROC curve in clinical settings. When reporting AUC results of simulations, we also report maximum achievable AUC values of 84% and 94% for heritabilities of 50% and 80%, respectively. These estimates are based on three different yet consistent estimations (see Supplemental Materials).

In this paper, we compare three different types of methods: the C+T method, T-Trees and PLR.

The C+T method directly derives PRS from the results of Genome-Wide Associations Studies (GWAS). In GWAS, a coefficient of regression (i.e., the estimated effect size ) is learned independently for each SNP j along with a corresponding P-value . The SNPs are first clumped (C) so that there remain only loci that are weakly correlated with one another (this set of SNPs is denoted ). Then, thresholding (T) consists in removing SNPs with P-values larger than a user-defined threshold . Finally, the PRS for individual i is defined as the sum of allele counts of the remaining SNPs weighted by the corresponding effect coefficientswhere are the effect sizes (P-values) learned from the GWAS. In this study, we mostly report scores for a clumping threshold at within regions of 500kb, but we also investigate thresholds of 0.05 and 0.8. We report three different scores of prediction: one including all the SNPs remaining after clumping (denoted C+T-all), one including only the SNPs remaining after clumping and that have a P-value under the GWAS threshold of significance (, C+T-stringent), and one that maximizes the AUC (C+T-max) for 102 P-value thresholds between 1 and (Table S2). As we report the optimal threshold based on the test set, the AUC for C+T-max is an upper bound of the AUC for the C+T method. Here, the GWAS part uses the training set while clumping uses the test set (all individuals not included in the training set).

T-Trees (Trees inside Trees) is an algorithm derived from random forests (Breiman 2001) that takes into account the correlation structure among the genetic markers implied by linkage disequilibrium (Botta et al. 2014). We use the same parameters as reported in table 4 of Botta et al. (2014), except that we use 100 trees instead of 1000. Using 1000 trees provides a minimal increase of AUC while requiring a disproportionately long processing time (e.g., AUC of 81.5% instead of 81%, data not shown).

Finally, for PLR, we find regression coefficients and that minimize the following regularized loss function(1)where , x denotes the genotypes and covariables (e.g., principal components), y is the disease status to predict, and are two regularization hyper-parameters that need to be chosen. Different regularizations can be used to prevent overfitting, among other benefits: the L2-regularization (ridge, Hoerl and Kennard (1970)) shrinks coefficients and is ideal if there are many predictors drawn from a Gaussian distribution (corresponds to in the previous equation); the L1-regularization (lasso, Tibshirani 1996) forces some of the coefficients to be equal to zero and can be used as a means of variable selection, leading to sparse models (corresponds to ); the L1- and L2-regularization (elastic-net, Zou and Hastie 2005) is a compromise between the two previous penalties and is particularly useful in the situation (p is the number of SNPs), or any situation involving many correlated predictors (corresponds to ) (Friedman et al. 2010). In this study, we use a grid search over . This grid-search is directly embedded in our PLR implementation for simplicity. Using should result in a model very similar to gBLUP.

To fit PLR, we use an efficient algorithm (Friedman et al. 2010; Tibshirani et al. 2012; Zeng and Breheny 2017) from which we derived our own implementation in R package bigstatsr. This algorithm builds predictions for many values of , which is called a regularization path. To obtain an algorithm that does not require to choose this hyper-parameter , we developed a procedure that we call Cross-Model Selection and Averaging (CMSA, Figure S1). Because of L1-regularization, the resulting vector of estimated effect sizes is sparse. We refer to this method as PLR in the results section.

To capture recessive and dominant effects on top of additive effects in PLR, we use simple feature engineering: we construct a separate dataset with three times as many variables as the initial one. For each SNP variable, we add two more variables coding for recessive and dominant effects: one variable is coded 1 if homozygous variant and 0 otherwise, and the other is coded 0 for homozygous referent and 1 otherwise. We then apply our PLR implementation to this dataset with three times as many variables as the initial one; we refer to this method as PLR3 in the rest of the paper.

We use Monte Carlo cross-validation to compute AUC, partial AUC, the number of predictors, and execution time for the original Celiac dataset with the observed case-control status: we randomly split 100 times the dataset in a training set of 12,000 individuals and a test set composed of the remaining 3155 individuals.

We compared PLR with the C+T method using simulations of scenario N1 (Table 1). When simulating a model with 30 causal SNPs and a heritability of 80%, PLR provides AUC of 93%, nearly reaching the maximum achievable AUC of 94% for this setting (Figure 1). Moreover, PLR consistently provides higher predictive performance than C+T across all scenarios considered, except in some cases of high polygenicity and small sample size, where all methods perform poorly (AUC values below 60% Figure 1 and Figure 3). PLR provides particularly higher predictive performance than C+T when there are correlations between predictors, i.e., when we choose causal SNPs to be in the HLA region. In this situation, the mean AUC reaches 92.5% for PLR and 84% for C+T-max (Figure 1). For the simulations, we do not report results in terms of partial AUC because partial AUC values have a Spearman correlation of 98% with the AUC results for all methods (Figure S3).

In practice, a particular value of the threshold of inclusion of SNPs should be chosen for the C+T method, and this choice can dramatically impact the predictive performance of C+T. For example, in a model with 30 causal SNPs, AUC ranges from <60% when using all SNPs passing clumping to 90% if choosing the optimal P-value threshold (Figure S4).

Concerning the threshold of the clumping step in C+T, we mostly used the common value of 0.2. Yet, using a more stringent value of 0.05 provides equal or higher predictive performance than using 0.2 in most of the cases we considered (Figure 2 and Figure 3).

Our implementation of PLR that automatically chooses hyper-parameter provides similar predictive performance than the best predictive performance of 100 models corresponding to different values of (Figure S8).

We tested the T-Trees method in scenario N1. As compared to PLR, T-Trees perform worse in terms of predictive ability, while taking much longer to run (Figure S5). Even for simulations with model COMP in which there are dominant and interaction-type effects that T-Trees should be able to handle, AUC is still lower when using T-Trees than when using PLR (Figure S5).

We also compared the two PLRs in scenario N1: PLR vs. PLR3 that uses additional features (variables) coding for recessive and dominant effects. Predictive performance of PLR3 are nearly as good as PLR when there are additive effects only (differences of AUC are always <2%) and can lead to significantly greater results when there are also dominant and interactions effects (Figures S6 and S7). For model COMP, PLR3 provides AUC values at least 3.5% higher than PLR, except when there are 3000 causal SNPs. Yet, PLR3 takes two to three times as much time to run and requires three times as much disk storage as PLR.

First, when reproducing simulations of scenario N1 using chromosome six only (scenario N2), the predictive performance of PLR always increase (Figure 2). There is a particularly large increase when simulating 3000 causal SNPs: AUC from PLR increases from 60% to nearly 80% for Gaussian effects and a disease heritability of 80%. On the contrary, when simulating only 30 or 300 causal SNPs with the corresponding dataset, AUC of C+T-max does not increase, and even decreases for a heritability of 80% (Figure 2). Second, when varying the training size (scenario N3), we report an increase of AUC with a larger training size, with a faster increase of AUC for PLR as compared to C+T-max (Figure 3).

Joint PLRs also provide higher AUC values for the Celiac data: 88.7% with PLR and 89.1% with PLR3 as compared to 82.5% with C+T-max (Figure S2 and Table 2). The relative increase in partial AUC, for specificities larger than 90%, is even larger (42 and 47%) with partial AUC values of 0.0411, 0.0426, and 0.0289 obtained with PLR, PLR3, and C+T-max, respectively. Moreover, logistic regressions use less predictors, respectively, at 1570, 2260, and 8360. In terms of computation time, we show that PLR, while learning jointly on all SNPs at once and testing four different values for hyper-parameter , is almost as fast as the C+T method (190 vs. 130 sec), and PLR3 takes less than twice as long as PLR (296 vs. 190 sec).

We tested our implementation on 656K genotyped SNPs of the UK Biobank, keeping only Caucasian individuals and removing related individuals (excluding the second individual in each pair with a kinship coefficient >0.08). Results are presented in Table 3.

Our implementation of L1-penalized linear regression runs in <1 day for 350K individuals (training set), achieving a correlation of >65.5% with true height for each sex in the remaining 24K individuals (test set). By comparison, the best C+T model achieves a correlation of 55% for women and 56% for men (in the test set), and the GWAS part takes 1 hr (for the training set). If using only the top 100,000 SNPs from a GWAS on the training set to fit our L1-PLR, correlation between predicted and true heights drops at 63.4% for women and 64.3% for men. Our L1-PLR on breast cancer runs in 13 min for 150K women, achieving an AUC of 0.598 in the remaining 39K women, while the best C+T model achieves an AUC of 0.589, and the GWAS part takes 15hr.

In this comparative study, we present a computationally efficient implementation of PLR. This model can be used to build PRS based on very large individual-level SNP datasets such as the UK biobank (Bycroft et al. 2018). In agreement with previous work (Abraham et al. 2013), we show that jointly estimating SNP effects has the potential to substantially improve predictive performance as compared to the standard C+T approach in which SNP effects are learned independently. PLR always outperforms the C+T method, except in some highly underpowered cases (AUC values always <0.6), and the benefits of using PLR are more pronounced with an increasing sample size or when causal SNPs are correlated with one another.

When there are many small effects and a small sample size, PLR performs worse than (the best result for) C+T. For example, this situation occurs when there are many causal variants (3K) to distinguish among many typed variants (280K) while using a small sample size (6K). In such underpowered scenarios, it is difficult to detect true causal variants, which makes PLR too conservative, whereas the best strategy is to include nearly all SNPs (Purcell et al. 2009).

When increasing sample size (scenario N3), PLR achieves higher predictive performance than C+T and the benefits of using PLR over C+T increase with an increasing sample size (Figure 3). Moreover, when decreasing the search space (total number of candidate SNPs) in scenario N2, we increase the proportion of causal variants and we virtually increase the sample size (Dudbridge 2013). In this scenario N2, even when there are small effects and a high polygenicity (3000 causal variants out of 18,941), PLR gets a large increase in predictive performance, now consistently higher than C+T (Figure 2).

The choice of hyper-parameter values is very important since it can greatly impact the performance of methods. In the C+T method, there are two main hyper-parameters: the and the thresholds that control how stringent are the C+T steps. For the clumping step, appropriately choosing the threshold is important. Indeed, on the one hand, choosing a low value for this threshold may discard informative SNPs that are correlated. On the other hand, when choosing a high value for this threshold, too much redundant information is included in the model, which adds noise to the PRS. Based on the simulations, we find that using a stringent threshold leads to higher predictive performance, even when causal SNPs are correlated. It means that, in most cases tested in this paper, avoiding redundant information in C+T is more important than including all causal SNPs. The choice of the threshold is also very important as it can greatly impact the predictive performance of the C+T method, which we confirm in this study (Ware et al. 2017). In this paper, we reported the maximum AUC of 102 different P-value thresholds, a threshold that should normally be learned on the training set only. To our knowledge, there is no clear standard on how to choose these two critical hyper-parameters for C+T. So, for C+T, we report the best AUC value on the test set, even if it leads to overoptimistic results for C+T as compared to PLR.

In contrast, for PLR, we developed an automatic procedure called CMSA that releases investigators from the burden of choosing hyper-parameter . Not only this procedure provides near-optimal results, but it also accelerates the model training thanks to the development of an early stopping criterion. Usually, cross-validation is used to choose hyper-parameter values and then the model is trained again with these particular hyper-parameter values (Hastie et al. 2008; Wei et al. 2013). Yet, performing cross-validation and retraining the model is computationally demanding; CMSA offers a less burdensome alternative. Concerning hyper-parameter that accounts for the relative importance of the L1 and L2 regularizations, we use a grid search directly embedded in the CMSA procedure.

We also explored how to capture nonlinear effects. For this, we introduced a simple feature engineering technique that enables PLR to detect and learn not only additive effects, but also dominant and recessive effects. This technique improves the predictive performance of PLR when there are nonlinear effects in the simulations, while providing nearly the same predictive performance when there are additive effects only. Moreover, it also improves predictive performance for the celiac disease.

Yet, this approach is not able to detect interaction-type effects. In order to capture interaction-type effects, we tested T-Trees, a method that is able to exploit SNP correlations and interactions thanks to special decision trees (Botta et al. 2014). However, predictive performance of T-Trees are consistently lower than with PLR, even when simulating a model with dominant and interaction-type effects that T-Trees should be able to handle.

The computation time of our PLR implementation mainly depends on the sample size and the number of candidate variables (variables that are included in the gradient descent). Indeed, the algorithm is composed of two steps: first, for each variable, the algorithm computes an univariate statistic that is used to decide if the variable is included in the model (for each value of ). This first step is very fast. Then, the algorithm iterates over a regularization path of decreasing values of , which progressively enables variables to enter the model (Figure S1). In the second step, the number of variables increases and computations stop when an early stopping criterion is reached (when prediction is getting worse on the corresponding validation set, see Figure S1).

For highly polygenic traits such as height and when using huge datasets such as the UK Biobank, the algorithm might iterate over >100,000 variables, which is computationally demanding. On the contrary, for traits like celiac disease or breast cancer that are less polygenic, the number of variables included in the model is much smaller so that fitting is very fast (only 13min for 150K women of the UK Biobank for breast cancer).

Memory requirements are tightly linked to computation time. Indeed, variables are accessed in memory thanks to memory-mapping when they are used (Priv et al. 2018). When there is not enough memory left, the operating system (OS) frees some memory for new incoming variables. Yet, if too many variables are used in the gradient descent, the OS would regularly swap memory between disk and RAM, severely slowing down computations. A possible approach to reduce computational burden is to apply penalized regression on a subset of SNPs by prioritizing SNPs using univariate tests (GWAS computed from the same dataset). Yet, this strategy was shown to reduce predictive power (Abraham et al. 2013; Lello et al. 2018), which we also confirm in this paper. Indeed, when using only the 100K most significantly associated SNPs, correlation between predicted and true heights is reduced from 0.656/0.657 to 0.634/0.643 within women/men. A key advantage of our implementation of PLR is that prior filtering of variables is no more required for computational feasibility, thanks to the use of sequential strong rules and early stopping criteria.

Our approach has one major limitation: the main advantage of the C+T method is its direct applicability to summary statistics, allowing to leverage the largest GWAS results to date, even when individual cohort data cannot be merged because of practical or legal reasons. Our implementation of PLR does not allow yet for the analysis of summary data, but this represents an important future direction. The current version is of particular interest for the analysis of modern individual-level datasets including hundreds of thousands of individuals.

Finally, in this comparative study, we did not consider the problem of population structure (Vilhjlmsson et al. 2015; Mrquez-Luna et al. 2017; Martin et al. 2017), and also did not consider nongenetic data such as environmental and clinical data (Van Vliet et al. 2012; Dey et al. 2013).

In this comparative study, we have presented a computationally efficient implementation of PLR that can be used to predict disease status based on genotypes. A similar penalized linear regression for quantitative traits is also available in R package bigstatsr. Our approach solves the dramatic memory and computational burdens faced by standard implementations, thus allowing for the analysis of large-scale datasets such as the UK biobank (Bycroft et al. 2018).

We also demonstrated in simulations and real datasets that our implementation of penalized regressions is highly effective over a broad range of disease architectures. It can be appropriate for predicting autoimmune diseases with a few strong effects (e.g., celiac disease), as well as highly polygenic traits (e.g., standing height) provided that sample size is not too small. Finally, PLR as implemented in bigstatsr can also be used to predict phenotypes based on other omics data, since our implementation is not specific to genotype data.

We are grateful to Flix Balazard for useful discussions about T-Trees, and to Yaohui Zeng for useful discussions about R package biglasso. We are grateful to the two anonymous reviewers who contributed to improving this paper. The authors acknowledge LabEx Pervasive Systems and Algorithms (PERSYVAL)-Lab [Agence Nationale de Recherche (ANR)-11-LABX-0025-01] and ANR project French Regional Origins in Genetics for Health (FROGH) (ANR-16-CE12-0033). The authors also acknowledge the Grenoble Alpes Data Institute, which is supported by the French National Research Agency under the Investissements davenir program (ANR-15-IDEX-02). This research was conducted using the UK Biobank Resource under Application Number 25589.

Available freely online through the author-supported open access option.

This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Read more from the original source:
Efficient Implementation of Penalized ... - genetics.org

Read More...

Page 32«..1020..31323334..40..»


2024 © StemCell Therapy is proudly powered by WordPress
Entries (RSS) Comments (RSS) | Violinesth by Patrick