deletions | additions       diff --git a/Sensitivity analysis using previously identified eQTLs.tex b/Sensitivity analysis using previously identified eQTLs.tex  index 9cfe325..039d773 100644  --- a/Sensitivity analysis using previously identified eQTLs.tex  +++ b/Sensitivity analysis using previously identified eQTLs.tex 
...
\subsection{CHT sensitivity analysis}  We To evaluate the sensitivity of CHT, we  compared the numberof  eQTLs identified using CHT  at 10\% FDR using to  the CHT and number identified using  asimple  linear model. For the linear model, GC and total read depth corrections were included. We model we  divided the observed counts $x_{ij}$  by the expected number of counts $T^*_{ij}$  based on the GC content  and total read depth. depth of each region.  We then used quantile normalization to bring the distribution of counts for each individual to a standard normal distribution. Principal components were We  included principal components  as covariates in the linear model. The model and determined the  number of principal components to use for each model was determined include  by maximizing the number of significant eQTLs. \subsubsection{Re-identifying \subsubsection{Identifying  known European  eQTLs in 70 Yoruba  LCLs} We found downloaded  a subset a set  of 1895  eQTLs for which were identified in 462  European LCLs \cite{24037378} lymphoblastoid cell lines (LCLs) by the GEUVADIS project\cite{24037378}. We identified the subset of these eQTL SNPs  that were segregating in the an independent dataset of 69  Yoruba population. LCLs \cite{Stephens_Gilad_Pritchard_2010}.  We then applied the CHT and linear model to an independent dataset of 70 the RNA-seq data from the  Yoruba LCLs \cite{Stephens_Gilad_Pritchard_2010}. LCLs.  \subsubsection{Genome-wide QTL discovery in small sample sizes of ChIP-seq data}  We also applied the two models to a dataset of ChIP-seq data for the histone modification H3K27ac on 10 individuals. individuals, which we collected in a previous study \cite{McVicker_2013}.