CHIT: an allele-specific method for testing the association between molecular quantitative traits and phenotype-genotype interaction.

MOTIVATION: Allele specific differences in molecular traits can be obtained from next generation sequencing data and could potentially improve testing power, but such information is usually overlooked in association studies. Furthermore, the variation of molecular quantitative traits (e.g., gene expression) could result from the interaction effect of genotypes and phenotypes, but it is challenging to identify such interaction signals in complex disease studies in humans due to small genetic effect sizes and/or small sample sizes.
RESULTS: We develop a novel statistical method, the combined haplotype interaction test (CHIT), which tests for association between molecular quantitative traits and phenotype-genotype interactions by modeling the total read counts and allele-specific reads in a target region. CHIT can be used as a supplementary analysis to the regular linear interaction regression. In our simulations, CHIT obtains non-inflated type I error rates, and it has higher power than a standard interaction quantitative trait locus approach based on linear regression models. Finally, we illustrate CHIT by testing associations between gene expression obtained by RNA-seq and the interaction of SNPs and atopy status from a study of childhood asthma in Puerto Ricans, and results demonstrate that CHIT could be more powerful than a standard linear interaction expression quantitative trait loci (eQTL) approach. AVAILABILITY: The CHIT algorithm has been implemented in Python. The source code and documentation are available and can be downloaded from https://github.com/QiYanPitt/CHIT. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.