Missing heritability problem

In genetics, the missing heritability problem^{[1][2][3][4][5][6]} refers to a difference between heritability estimates obtained from early genome-wide association studies (GWAS) and heritability estimates from twin and family data across many physical and mental traits, including diseases, behaviors, and other phenotypes.

An influential review article^[7] in 2008 noted that the amount of phenotypic variance explained by significant loci in GWAS studies up to that point was usually far less than expected based on family studies. This gap was referred to as "missing heritability". Using height as a model trait, a paper in 2010 showed that most of the missing heritability can be explained by the presence of large numbers of low variants whose effect sizes were too small to detect at the sample sizes that were then available.^[8] This conclusion has subsequently been confirmed using much larger sample sizes, including a study of 5.4 million individuals that identified around 12,000 independent variants that affect human height.^[9] While studies of height have particularly large power due to their very large sample size, other complex traits likely have similar genetic architecture. Thus, the missing heritability problem is largely resolved by the presence of tens of thousands of variants of small effects that could not be detected in early GWAS studies.