Skip to main content
Premium Trial:

Request an Annual Quote

NHGRI Precision Health Research Program Using Reverse Phenotyping to Better Understand Genomics


NEW YORK – The National Human Genome Research Institute (NHGRI) has launched a new precision health research program with the aim of reverse engineering genomic data in order to more accurately predict diseases or other traits based on the presence of specific genetic variants.

The technique is called reverse phenotyping, or genotypic ascertainment phenotyping. It involves using individuals' genome data to generate hypotheses about possible associations between specific genetic variants and specific phenotypes and then testing the individuals the data came from to see if they indeed carry those phenotypes. Unlike forward-looking genetics techniques, which seek to find the genetic basis of a phenotype or trait, reverse phenotyping seeks to identify a phenotype that may arise as a result of particular genetic sequences.

Leslie Biesecker, chief of the NHGRI Medical Genomics and Metabolic Genetics branch and the new program's director, said that the agency already has cohorts of individuals sequenced and ready to go. And because these individuals have consented to be recontacted when necessary, it's a simple matter to look at the sequence data, make a hypothesis about what disease or trait a person may have based on a particular variant, and then bring that patient to the NIH Clinical Center and run an experiment to test for the presence or absence of that phenotypic attribute.

"That's the fundamental research model for what predictive genomic medicine needs to be. What we want to do is predict disease based on genomic or other omic attributes and test its validity and robustness and clinical utility," he said.

Though it seems to be gaining in popularity, there's only a smattering of published studies in the last few years that employ the reverse phenotyping technique. The most recent paper, a January study in the Clinical Journal of the American Society of Nephrology, showed that reverse phenotyping could increase the accuracy of certain diagnostic tests. Another study published in Pediatric Neurology in 2017 showed the approach could be used to help researchers further define and understand a new and rare mitochondrial syndrome caused by a specific gene mutation.

"That is how we're going to build the evidence base that would justify genomic screening healthcare approaches," Biesecker said. "That's what we need to be able to do."

The power of the reverse phenotyping approach is that there are plenty of genes and variants to work on, he added. A large degree of what will be accomplished in the bounds of the NHGRI's program will be driven by the individual interests of different researchers.

Biesecker's own group is working on a trait called malignant hyperthermia, which is a type of severe reaction that occurs in response to particular medications used during general anesthesia. Susceptibility can occur due to at least six genetic mutations, with the most common one being in the RYR1 gene. These genetic variations are often inherited, but the condition has also been associated with de novo mutations, as well as several inherited muscle diseases, such as central core disease.

"It's a fascinating trait because it's halfway between a disease and a pharmacogenetic attribute, in that people who have this trait, if they receive general anesthesia, they have a substantial chance of having a catastrophic reaction to that anesthesia and even dying in the operating room," Biesecker said. "And now we're starting to be able to identify who these people are. And there's a very straightforward way to manage them so that this catastrophic event doesn't happen. And that's a great model for predictive medicine."

Josh Denny — the new CEO of the All of Us Research Program, and an expert in clinical informatics and precision medicine whose research group will reside within the NHGRI Intramural Research Program — is also joining this new precision health initiative. Denny's general research approach is to use clinical informatics databases in order to sort and identify individuals who have unrecognized presentations of a known genetic disorders or who have novel genetic disorders, Biesecker said, adding, "So, we're coming at it from complementary angles and converging on the notion of defining and identifying disease and susceptibility based on informatics attributes, either genomic or phenotypic."

All of Us will be one of the research programs that the NHGRI initiative will draw its data from. Because the individuals who sign up for All of Us have agreed to be recontacted after sequencing, Biesecker's group will be applying to use the data to apply the reverse phenotyping approach to find people in that cohort who have these malignant hypothermia variants, and evaluate the predictive utility of doing so. They'll also be looking to access data from the UK Biobank, BioVU, ClinSeq, and others databases, and study selected cohort participants from the NIH Clinical Center.

Importantly, Biesecker said, taking this approach to precision medicine could help researchers overcome the limitations of the Genome Aggregation Database (gnomAD) — the revolutionary, and publically available, database that aggregated and harmonized exome and genome sequencing data from a variety of large-scale sequencing projects — or Exome Aggregation Consortium (ExAC) as it was first known.

"We all love and use ExAC, and now predominantly gnomAD, every day, and it's wonderful," he said. "But it is a severe limitation of that resource that [although] finding a variant in gnomAD allows you to measure its frequency in a certain kind of a population, we cannot know what the phenotype is that's associated with the individual who has that variant."

Biesecker hopes that the NHGRI's initiative can take gnomAD's work one step further and start making those genotype-phenotype connections more concretely. The program's approach also has the potential to more firmly resolve lingering issues around certain variants of unknown significance, making determinations one way or another whether they are indeed associated with diseases or traits.

Further, he said, one of the major criticisms of predictive medicine is that the predictive power of those findings is being overestimated, because most of the existing data about variants in those genes is based on families that have a high prevalence of these diseases. But this program could go a long way toward ameliorating this problem and taking some of the teeth out of the criticism.

"There is an inherent bias towards individuals who have higher-penetrance variants, because if they had a lower-penetrance variant in that gene, then they wouldn't have been identified as a highly or severely affected family, [and they wouldn't have been] in those studies," Biesecker said. "That's what leads people to say our studies are biased towards high penetrance and that we're overestimating the penetrance. Genomic ascertainment research allows us to actually measure what that penetrance really is, because we are going to find the people with the variants and ask what is the penetrance without that ascertainment bias."

And because the researchers will have access to data from programs such as All of Us, they're also hoping to have access to a more diverse group of individuals to study, which could also help ameliorate the problem of the results of disease studies being too skewed toward individuals of European ancestry.

This is especially important for researchers looking to develop polygenic risk scores for various diseases, according to Biesecker, as it's extremely likely that such calculations could be significantly affected by the geographic origin of the individuals who are studied and to whom they are applied.

"It arguably may be less important for rare high-penetrance variants, only because those population-specific modifiers relatively have less effect in the presence of a very high-penetrance single-gene variant," he said. "But it will be important, for sure, and it'll be differentially important in different diseases with different degrees of penetrance. You've got to have the diversity to answer all those questions."

Certain details are still being worked out, he noted. The program will likely have its own internal working databases while studies are underway, but as results are generated that are broadly useful, those results are likely to be added right back into databases like All of Us rather than into a new custom-built database made for this program.

"We'd rather see information flow right back to where it was generated from so that everyone has access to it as soon as possible," Biesecker said.

He also noted that the program is looking to recruit at least one more principal investigator, as well as some career development professionals. And the initiative is also aiming to develop some core facilities that will support precision health research for many researchers in the NHGRI's intramural program.

Ultimately, he added, what the program is proposing to do is to use big data and large phenotypic and genotypic datasets to ask and answer questions in a way that's complementary to existing specific hypothesis testing science. "It will answer different questions in powerful ways and allow us to leverage each other upward towards doing a better job of understanding the molecular biology of human disease," Biesecker said.