Custom reference when aligning male/ female NGS data to hg19 reference

[ron128]

Hello everybody. I just wanted to ask if somebody has any experience in aligning sex specific data to hg19 genome build. I recently came accross a weird problem (which when you look at it indepth makes more sense) wherein I aligned my transcriptome sequencing data from primary breast cancer tumours to hg19, and a lot of the reads (approx 15000) were actually aligning to the y chromosome. I am using tophat 2.0 and bowtie 2.0 for the alignment.

Has anybody faced this issue previously? Does it make sense to create a custom sex specific reference by removing the Y chromosome if your sequencing sample is female? I know 15000 is too small a percentage of reads when compared with the entire transcriptome having millions of reads, all the same this is an interesting question. Or is it that it really dosent matter a lot, and I am making a mountain out of a molehole? ANy thoughts would be mightily appreciated! thanks

[alexdobin]

Within the ENCODE consortium we use male/female references to avoid the false positives you saw (reads mapping to uniquely chrY in a female sample), as well as to avoid multi-mappers that map to both chrX and chrY in a female sample. Moreover, the autosomal (PAR) regions on chrY in the male genome are masked out.