consensus of CIGAR strings

[Flo1000]

Hi guys,

I am writing an tool for amplicon sequencing, which needs to condense reads mapping to the same position. At the moment, I just take the most frequent read, which is obviously far from perfect, especially in low quality or low coverage situations.
While it is easy to calculate the consensus of the base string and to approximate the quality, the CIGAR string is much harder. Does anyone has an algorithm or even source code to do this?

Best

Flo

[Brian Bushnell]

One solution is to do multiple sequence alignment... I'm sure there are tools for this exact problem, but I don't know what they are, specifically. I think the Mothur package has something for that, though, specifically for creating a consensus of reads from a given amplicon.

Another approach would be to gather all the reads mapped to the same position, and assemble them. This might be computationally more efficient than multiple sequence alignment. Possibly, an OLC assembler like MIRA would be best.