Quality Values in BowTie2

[pikeli]

Hi,

I am trying to understand how bowtie2 uses the quality values of the FASTQ files. It is my understanding after reading the paper that in the 'seed alignment phase' the QV are ignore.

Then, when it tries to align the whole read, then it uses the QVs to obtain 'scores' of the alignment and then it reports the alignments with the highest score.

However, it is not very clear how exactly they use the value of the QV. Is it like in BowTie(1)?
In the previous version [bowtie1] it is clearly explained how the QV are used (accepting up to a sum 70 in the QV of the mismatches positions, etc.)

I would really appreciate if someone could shed some light in this issue.

Thank you very much,

[dpryan]

In the case of end-to-end alignment, the mismatch penalty that's used to calculate the alignment score is dependent on the base-call phred score. I've never actually looked, but I suspect the phred scores make no difference in local alignment, since the alignment score is purely dependent on the number of matches and there is no mismatch penalty (well, it's effectively 0).