NEW YORK (GenomeWeb) – A UK-led team has used a new computational model to take a closer look at the proportion of variation observed in a given phenotype that can be explained by adding together known SNP contributors to that phenotype — an approach that the group expects will clarify complex trait heritability.
"By examining a large collection of real data sets, we derive approximate relationships between the expected heritability of a SNP and [minor allele frequency], levels of [linkage disequilibrium] with other SNPs, and genotype certainty," corresponding author and co-first author Doug Speed, a researcher at the University College London's UCL Genetics Institute, and his co-authors wrote. "This provides us with an improved model for heritability estimation and a better understanding of the genetic architecture of complex traits."
As they reported in Nature Genetics today, Speed and his colleagues came up with a computational model dubbed LDAK, which they applied to genetic data for 42 complex human traits, spanning anthropomorphic, physiological, cardiac, metabolic, and blood chemistry-related traits and conditions.
The researchers reasoned that since estimated SNP heritability hinges in part on SNP effect size assumptions, they might get a better sense of the SNP heritability by using real human data sets to test and compare different models.
"Previous attempts to assess the validity of assumptions have used simulation studies, but this approach will tend to favor assumptions similar to those used to generate the phenotypes," the authors noted. "Instead, we have compared different heritability models empirically, by examining how well they fit real data sets."
Based on patterns detected with genotyping data for 19 conditions considered in genome-wide association studies performed for the Wellcome Trust Case Control Consortium and other groups, the team estimated that common SNPs typically account for more than 40 percent of the variability in human traits, on average — a SNP heritability estimate that exceeds that reported using so-called GCTA software or the GCTA-LDMS model.
The researchers also presented evidence that the LDAK model may offer an improved view of the ties between heritability and linkage disequilibrium relative to models used previously, suggesting that "for many traits, common SNPs explain considerably more phenotypic variance than previously reported, which represents a major advance in the search for missing heritability."