Bowtie mapping and seed length

[nilmot13]

Hi,

I've been using bowtie function on Galaxy. I just have a quick question with regards to setting seed length when mapping.

I've trimmed my sequencing reads to let say 25bp because of poor read quality. The default mapping parameter state seed length is 28bp.

Would this affect the mapping result?
Will the mapping even be valid taking more than the sequence reads a seed?
Is it better to reduce seed length to something like 20bp?

Many thanks for your help

[colindaven]

Yep, maybe 18-20bp would be a good idea. As far as I remember the 28bp seed is important for the first rapid scan, and you allow by default 2 mismatches (-n switch). Thereafter, the -e switch becomes important for mismatches.
I heard a seed of 40bp is appropriate for ~76bp reads, so you should shorten it for yours.

Have a look at the bowtie docs, they're good.

[epi]

It seems you applied a rigorous quality filter to make your reads 25bp, or this was not a particular good sequencing run. Anyways, try -v which rids you of seed length altogether.

Hi,

I've been using bowtie function on Galaxy. I just have a quick question with regards to setting seed length when mapping.

I've trimmed my sequencing reads to let say 25bp because of poor read quality. The default mapping parameter state seed length is 28bp.

Would this affect the mapping result?
Will the mapping even be valid taking more than the sequence reads a seed?
Is it better to reduce seed length to something like 20bp?

Many thanks for your help

[nilmot13]

Thank you all for the useful response.

The sequencing run we've done is paired. Read1 has good quality shown in statistical plots which is about 75 bp. Read2 however has much low read quality and the median shown in stats plots suggested anything more 30 bp had low phred score and high amount of variation.

I'll adjust my mapping parameter accordingly then and see what happens.