Sequencing reads (75 bases, single reads) were mapped to the human hg19 reference genome using an in-house mapping pipeline (Breese & Liu, 2013) that utilizes bfast-0.7.0a (Homer, Merriman, & Nelson, 2009). In brief, reads were truncated where the average quality fell below 10 within a window size of 5, and then reads of <35 bases were discarded. Reads mapped to rRNA/tRNAs were discarded. The remaining reads were mapped to reference genome hg19 and to a splice-junction library created in-house. The genomic and splice-junction library maps were then merged. Reads mapped to multiple positions in the genome were excluded from further analysis. The gene-based expression levels were calculated using bamutils from NGSUtils, based on the RefSeq gene annotation of hg19. The average number of reads per sample was 28 million. Data deposited in GEO, GEO series <awaiting accession number>.
