Abstract
Although many genetic factors have been successfully identified for human diseases in genome-wide association studies, genes discovered to date only account for a small proportion of overall genetic contributions to many complex traits. Association studies have difficulty in detecting the remaining true genetic variants that are either common variants with weak allelic effects, or rare variants that have strong allelic effects but are weakly associated at the population level. In this work, we applied a goodness-of-fit test for detecting sets of common and rare variants associated with quantitative or binary traits by using whole genome sequencing data. This test has been proved optimal for detecting weak and sparse signals in the literature, which fits the requirements for targeting the genetic components of missing heritability. Furthermore, this p value-combining method allows one to incorporate different data and/or research results for meta-analysis. The method was used to simultaneously analyse the whole genome sequencing and genome-wide association studies data of Genetic Analysis Workshop 18 for detecting true genetic variants. The results show that goodness-of-fit test is comparable or better than the influential sequence kernel association test in many cases.
Highlights
According to the Catalog of Genome-Wide Association Studies updated by the National Human Genome Research Institute, approximately 7260 single-nucleotide polymorphisms (SNPs) have been identified for 770 traits in 1360 publications as of November 2012
To assess how the single-nucleotide variants (SNVs) group’s size may have affected the performance of these tests, we split chr3 into segments of fixed windows with 1 of 3 widths: 10 kilobase pairs, 100 kbp, and 500 kbp. At those 3 levels, the grouping strategy resulted in 19,472, 1950, and 391 windows, among which 87, 37, and 20 windows contained true SNVs that were either nonsynonymous or regulatory to systolic blood pressure (SBP) according to the Genetic Analysis Workshop 18 (GAW18) simulation [13]
We defined a SNV as a rare variant if it had a minor allele frequency less than 5%
Summary
According to the Catalog of Genome-Wide Association Studies updated by the National Human Genome Research Institute, approximately 7260 single-nucleotide polymorphisms (SNPs) have been identified for 770 traits in 1360 publications as of November 2012. Researchers believe that a significant proportion of heritability of many complex traits is still missing [1,2]. The remaining genetic variants to be detected are either common variants with small allelic effects, or rare variants with relatively strong allelic effects. In both cases, the genetic effects are weak at the population level. Only a small proportion of the avalanche of candidate variants are likely associated with a trait, which is a problem closely related to sparse signal discovery in statistics. It is very challenging to detect weak and sparse genetic effects via association
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have