Bowtie 2 parameters, these should be generous correct?

[KnowNothing2]

I previously ran my fastq file with parameters I can't remember and it gave me ~82% coverage.

Today (2 months later), I attempted even a less stringent run using

--local --norc -N1 -L 15

(i.e. 15 nucleotide seed with one allowable mismatch)

This only gave me ~35% coverage.

I have no clue what happened.

[gringer]

That seed length looks far too short. Even bowtie2's "sensitive" mode (--sensitive) uses a seed length of 20. If you have too small a seed, then bowtie2 will find too many false hits (from a matching seed) and give up on extending them. See the "-D" option:

http://bowtie-bio.sourceforge.net/bowtie2/manual.shtml#input-options

If you want a less stringent run, it's easier to just use the standard modifier parameters (e.g. '--very-sensitive', or '--very-sensitive-local').

[KnowNothing2]

Awesome thanks, so then why didn't I get 100% coverage with a 15 seed. How will a 20 seed perform more optimally?

[gringer]

Because bowtie2 gives up on a read once there have been too many seed hits that had failed extensions. As I mentioned before, see the "-D" option:

http://bowtie-bio.sourceforge.net/bowtie2/manual.shtml#effort-options

[KatrinSameith]

Dear gringer, and other users,

I would like to pick up this question once more. I had a similar question as KnowNothing2, and am not sure, I understand the -D parameter. Would someone explain it a bit more in detail to me?

Lets say I have a read of 150bp.
And with L=10 & i=S,1,0.5, I have one 10bp seed every 7bp (=1+0.5*sqrt(150)).

How does it work from here? Will bowtie2 extend each seed one by one, and if D=15 fail, it gives up, because it seems that all seeds are no good and it takes too long to find the good alignment?

Any comment is appreciated. Thanks for your help,
Katrin