Abstract:
|
In genetic studies, recent research work has provided evidence that rare/ultra-rare variants tend to be more deleterious. Statistical methods such as burden analysis, SKAT or SKATO have been developed trying to identify rare causal risk variants responsible for complex traits or disease. These methods suffer several limitations. Simple burden analysis by pooling together all variants will significantly lower the power if true causal variants only take a small portion since the signal will be greatly diluted. SKAT, however doesn't take into account information of biological annotations. Thus motivated by these issues, we have proposed a mixture model to account for the uncertainty of being causal for a variant. The model can also take advantage of external information by linking the probability of a variant being causal with the various annotations we have, such as allele frequency and likely deleterious effects. Simulation studies have demonstrated the effectiveness of the method and it is being applied to analyze exome sequencing data of Autism Spectrum Disorders (ASD).
|