[Bio-bwa-help] bwa mem and quality trimming

[Hannes S. <ha...@sv...>]

Hi all,

I was searching in the mailing list archives,
but I could not find any satisfactory reply to my issue so I would like to
post it here.

I have Illumina 100bp paired end reads and want to align them to a
reference genome. With bwa aln there was the -q option to trim low quality
ends of reads on the fly.

Now bwa mem does not provide such an option. Does it deal with low quality
parts of the reads automatically, or should the reads be trimmed beforehand?
Note that I am not talking about adapter trimming, those should have been
removed beforehand in any case.

If quality trimming is recommended, I am not clear as to whether reads that
are below 70bp in length should be discarded after trimming. (It is noted
that bwa mem is only recommended for reads >=70bp.) Does anyone have an
answer to this?

In particular, I used trimmomatic to trim if the average quality over 4bp
windows is below 15. I often get read pairs where one mate is discarded due
to length<70bp after trimming. In this case I would have to map single
reads. Intuitively, I have the feeling that the alignment of a 100 bp read
could be facilitated by having a mate, even if the mate is only 50bp after
trimming. Should I keep pairs where one mate is <70 bp and the other >70bp
or does bwa mem have troubles with this?

Thanks a lot in advance,
Hannes

PS.: There is a related discussion here, but no satisfactory reply:
http://www.biostars.org/p/90149/#90184

-- 
Dr. Hannes Svardal
Postdoctoral researcher
Nordborg group

Gregor Mendel Institute
Dr. Bohr-Gasse 3
1030 Vienna, Austria
phone: +436803252197