Minimal length of sequence for reference alignment.

Question

[mieszko91]

Hi! When I am using bowtie 2 to reference alignment, console displays warnings about length of my sequences:

Warning: skipping mate #2 of read 'SRR988073.8033840 HWI-ST833:124:6:2202:2830:97445 length=101' because it was < 2 characters long Warning: skipping mate #2 of read 'SRR988073.8078016 HWI-ST833:124:6:2202:19987:137144 length=101' because length (1) <= # seed mismatches (0)

I have trimmed my data to improve quality of my sequences, but in fastq file is many short sequences now. Is there any minimal length of sequence for reference alignment? Shoud I remove short sequences? And how long sequences shoud be?

Answer 1

[GZ1995]

Typically short reads may map to multiple genome regions. Thus, it is recommended to remove the short sequences before alignment. I usually use 35 bp as the minium post-trimming length for bowtie2.

Answer 2

[WouterDeCoster]

Well a length <2 characters is for sure far too short. Avoid too aggressive trimming, aligners commonly can handle low quality bases and suboptimal parts of a read.

[GenoMax]

This ^. There is really no need to trim based on Q scores (which may be the case here) for re-sequencing data.