I am interested in using Bowtie to map reads back onto a de novo Illumina assembly of a small bacterial genome. I used a subset of quality-filtered reads to build the de novo assembly using Velvet, and so I'd like to use Bowtie to map the complete read dataset to my draft assembly as a way to validate the assembly and to detect polymorphisms.
Because of my interest in SNP detection, I was curious about the effects of the mismatch policy. The Bowtie paper and documentation explains that, in addition to the number of mismatches in the first L bases of an alignment, Bowtie also considers the lowest sum of quality values at all mismatched positions as a hallmark of the optimal alignment. Low quality values would be true for a sequencing error, but if the mismatched base is actually a real polymorphism, the quality value will be high. How does this affect Bowtie's use for SNP identification?
My read dataset comes from a population rather than a strictly clonal strain, so I might expect a higher level of variation.
Any advice or insight would be greatly appreciated!