Discerning between mismatch and N

Question

[jordi]

Hi,

I'm using bowtie 1.1 to run alignment on paired end reads (40bp). I would like to know if there is any option in bowtie that allow you to modify the maximum number of nucleotides undetermined (N) allowed independently from the number of real mismatches (when the nucleotide read is not the same as in the reference genome) when aligning.

Thank you,

[Martombo]

why don't you filter the reads according to the number of N prior to the mapping with bowtie?

[jordi]

Thanks for the reply Martino,

My purpose is to allow more N in the reads than real mismatches. For example, MaxN=5 and v=1.

Answer 1

[Sean Davis]

I think you may need to post-filter using the MD tag.

Answer 2

[Devon Ryan]

I don't believe that what you want to do is possible with bowtie v1.1. Even with v2, the mismatch penalty from an N isn't considered independently from other bases. The best you could do is use bowtie2 and set --np 0 and use a suffciently high --n-ceil, which would be the part that filters by N. You would then need to select an --n-ceil function and a --score-min function that are multiples of each other, which should be simple enough and would then effectively score filter by number of Ns seperately from the number of mismatches.