Graham McVicker edited QTL testing with WASP.tex  almost 10 years ago

Commit id: c64c972274ee51602b44d16c930a4896300429f9

deletions | additions      

       

To discover molecular quantitative trait loci (QTLs) WASP uses a statistical test, which we call the Combined Haplotype Test (CHT). As input, CHT takes genotype probabilities at known SNPs as well as mapped reads from next-generation sequencing experiments such as ChIP-seq or RNA-seq. CHT combines two types of information from next generation sequencing data: the depth of mapped reads and the allelic imbalance of mapped reads that overlap heterozygous sites. CHT models the overdispersion of read counts (both across regions and across individuals) and accounts for variability introduced by GC content and the fraction of reads that fall within peaks.   We used an early version of CHT to discover histone mark QTLs as described in the supplementary material of our paper\cite{McVicker2013}. QTLs\cite{McVicker2013}.  We have extended the statistical model, and made numerous improvements to the software implementation. These improvements include: \begin{itemize}  \item Dispersion Automatic estimation of dispersion  parametersare now estimated automatically  from the data (rather than chosen by the user) data,  \item Unknown Adjustment for unknown  covariatescan be accommodated  by the model by providing allowing  principal component loadings to be provided,  \item Tested Allowing tested  regions can to  be split across multiple genomic regions, segments,  such as exons exons,  \item Implementation Greater  efficiencyhas been improved  sothat  the model can now be run with hundreds of individuals individuals.  \end{itemize}