Fixed length trimming of paired end reads for MISO

[sjuzhet]

Hey y'all,

I'm in possession of a dataset that needs some pretty aggressive trimming to get it to align properly, but because MISO requires fixed read lengths, I'm trying to trim a fixed length from all of my reads. However, I'm not really sure how to approach it! Should I trim a bit off the 5' or 3' or both of either or both pairs? Anyone have any advice on how to go about doing this?

[Brian Bushnell]

If you map your reads with BBMap using the "mhist" flag:

bbmap.sh ref=ref.fa in=reads.fq mhist=mhist.txt out=mapped.sam

...you can plot that output to see the mismatch rate by position in the reads, which will help you determine the best fixed length to trim. Probably you only need to trim the right end; normally the left end is pretty good for Illumina randomly-sheared libraries.

However, if you generate the sam files using BBMap in the first place, you don't really need to trim the reads first, as it's very sensitive and designed for low-quality data. It's still a good idea to trim if, say, the last 15 bases are complete garbage but it's very rare that that's necessary.