Discrepancy in Heritability Estimates: Twin Studies Reveal Significant Genetic Influence Unaccounted for by Molecular Methods

A recent social media post by "Crémieux" has highlighted a persistent challenge in genetics: the significant underestimation of molecular heritability for numerous phenotypes compared to estimates derived from twin studies. The tweet points to a wide array of traits, from eye and hair color to facial morphology and various biological activities, where molecular methods fall short. This phenomenon, known as the "missing heritability problem," suggests that current genetic analyses may not fully capture the complex genetic architecture underlying many human characteristics.

According to the post, "Molecular heritability estimates are extreme underestimates for a lot of phenotypes, like eye color, hair color, facial morphology, dermatoglyphics, salivary amylase, serum bilirubin, butyrylcholinesterase activity, alpha-1 antitrypsin level, anosmias and hyperosmias, hair whorl, widow's peak, mid-digital hair/unibrow/synophrys, Darwin's tubercle, and more." This observation is particularly striking for physical traits, where twin correlations are notably high, often reaching around 0.98 for monozygotic (identical) twins and 0.49 for dizygotic (fraternal) twins. Such high twin correlations indicate a strong genetic component.

The discrepancy arises because traditional twin and family studies, which compare trait similarities between relatives, often yield much higher heritability estimates than those found by genome-wide association studies (GWAS). While GWAS identify specific genetic variants (SNPs) associated with traits, they typically explain only a fraction of the heritability observed in family studies. The tweet's author speculates, "I wonder if this is all just due to poor phenotyping," suggesting that imprecise or insufficient measurement of traits could be a major factor.

Experts in the field widely acknowledge that "poor phenotyping," or the inadequate measurement and definition of traits, can indeed dilute genetic signals and hinder the detection of associated variants. Beyond phenotyping, other proposed explanations for missing heritability include the cumulative effect of a vast number of common variants each with a small effect, the role of rare genetic variants, structural variations in the genome, epigenetic modifications, and complex gene-gene (epistasis) and gene-environment interactions that are difficult to model. A recent study published in the Journal of Proteome Research in May 2025 further illustrated this, finding that twin-based heritability estimates for plasma proteins were, on average, twice as high as SNP-based estimates, with half of the total heritability remaining unaccounted for.

Addressing this gap requires continued advancements in genomic technologies, improved methods for capturing and analyzing diverse genetic variations, and, crucially, more precise and comprehensive phenotyping. Researchers are exploring strategies such as whole-genome sequencing, advanced statistical models for gene interactions, and the integration of multi-omics data to unravel the intricate genetic underpinnings of complex traits and diseases.