NEW YORK (GenomeWeb News) – Much of what has come to be known as "missing heritability" could be a consequence of genetic interactions that cause the total heritability of certain traits or diseases to appear larger than it actually is, a study set to appear online this week in the Proceedings of the National Academy of Sciences suggests.
Using Crohn's disease and schizophrenia as examples, researchers from the Broad Institute and Brigham and Women's Hospital provided statistical illustrations of how population-based estimates of heritability and an insufficient appreciation of the possible presence of genetic interactions can lead to what they call "phantom heritability." As such, they explained, epistatic and other interactions could artificially inflate the number of genetic variants expected to contribute to some conditions.
"Our results cast doubt on current conclusions about missing heritability, because they ignore effects of genetic interactions," the study's senior author Eric Lander, Broad founding director and head of the institute's genome biology program, and co-authors wrote. "[W]e show that simple and plausible models can give rise to substantial phantom heritability."
To address such potential errors, the team developed a top-down heritability estimate method, also presented in the paper, that uses information from isolated populations to get a sense of the actual heritability for a given trait even when genetic interactions exist.
In parallel with genome-wide association and other studies that have uncovered scores of loci linked to common traits and diseases in recent years, the team explained, there has been an ongoing debate over just how many influential variants remain undiscovered. Despite the growing tally of disease-associated sites in the genome, analyses often suggest that a large proportion of variants influencing traits and diseases have not yet been found — an apparent missing heritability problem that has dogged association studies.
"[E]arly GWAS were puzzling because they appeared to explain only a small proportion of the 'heritability' of the traits," the study's authors noted. "With larger GWAS, the proportion of heritability explained has grown … but most of the heritability remains unexplained for most traits."
Lander and his colleagues reasoned that this apparently large gap between explained and unexplained heritability may be at least partly explained by over-estimates in the overall heritability.
This heritability — the combined effects that known and undiscovered genetic variants are predicted to have on a trait — is generally estimated using indirect evidence gleaned from population patterns, the researchers explained. But in most cases, they argued, such estimates have been based on models that don't account for possible interactions between genes and pathways.
"[G]enetic interactions may greatly inflate the apparent heritability without being readily detectable by standard methods," they wrote. "Thus, current estimates of missing heritability are not meaningful, because they ignore genetic interactions."
Indeed, the team outlined several scenarios involving genetic interactions that could skew heritability estimates, including a model known as the limiting pathway model that involves one or more rate-limiting processes.
Determining the extent to which genetic interactions influence a given trait is expected to be challenging in its own right, the researchers explained. In their Crohn's disease example, for instance, they estimated that it may be necessary to sample some half a million people to get a sufficient handle on the genetic interactions that contribute to the condition.
Even so, the team proposed a top-down theorem for estimating heritability and detecting genetic interactions in isolated populations in which individuals share large stretches of DNA that are identical by descent.
And while their analyses point to the possible presence of phantom heritability, the study authors were quick to note that this phenomenon isn't expected to explain all of the missing heritability in GWAS. On the contrary, they noted, the search for disease-associated variants is expected to be ongoing and will require further research.
Moreover, they cautioned that the focus on accounting for all of the heritability for a trait or disease should not overshadow the other information that can be gained from GWAS and other medical genetic studies.
"Ultimately," they concluded, "the most important goal for biomedical research is not explaining heritability — that is, predicting personalized patient risk — but understanding pathways underlying disease and using that knowledge to develop strategies for therapy and prevention."