I am trying to map 150 PE Illumina reads to a mitochondrial genome, which contains a large tandem duplication (combined length = 9 kb). When aligning the duplicated regions against each other, there are just 94 SNPs between them. As a result of this similarity, reads are mis-mapping between the two.
In an attempt to solve this, I have tried the bowtie2 --very sensitive settings but the mis-mapping is still evident. I also tried filtering by mapping quality and increasing the seed length for mapping. Whilst this eliminated the problem, coverage is drastically reduced to the point that I can't use this region for the intended use of my data, which is to detect low frequency de novo mutations.
Does anybody have any other suggestions worth trying?
Many thanks in advance.