Filter PE reads together

[nr23]

Hi,

I'm working with RNA-Seq using 65 bp paired end reads. I'm having huge problems with mapping - after trimming and filtering reads (using fasts etc) only a small fraction (~20%) map to my reference genome (using bow tie/tophat). I'm attempting to map to a different species, which would explain this low number, but I also see that 0.01% of my PE reads pair 'properly' during alignment (using Samtools flagstat). I'm wondering if, during QC steps the reads have become unpaired, and thus fail to find mates during mapping.

Could anyone advise on this? Is there a way to filter PE reads together?

ANy help would be greatly appreciated,

N

[dpryan]

There are a number of read quality trimmers that can properly handle paired-end data. Trimmomatic is one, trim_galore/cutadapt is another and there are likely a slew more if you search. Reads becoming unpaired could certainly cause the issues that you're describing.

[nr23]

Brilliant - thanks. Is there any chance that this lack of pairing in mapping is also responsible for the low mapping % I'm seeing?

[dpryan]

Could be. Not knowing if the reads are truly being mis-paired or the similarity of the species to which you're aligning makes it hard to say exactly.