Posts

Showing posts from July 11, 2010

GSNAP

AlignersSince starting the methylcoder project, I've been using the bowtie short read aligner. It's very fast, uses very little memory, aligns Illimina, SOLID, and colorspace reads, and has enough options to keep you busy (including my favorite: --try-hard).
There's a new short-read aligner in my feed-reader each week. I wish, as a service, they'd tell me what they do that bowtie doesn't. There are 2 scenarios I know of that bowtie doesn't handle.
As with many aligners, bowtie creates an index of the reference genome. It uses a suffix array compressed with the Burrows-Wheeler Transform(BWT). It uses 32 bit integers to store the offsets into that index so it's limited to under 4 gigabases of reference sequence. This is more than enough to hold the human genome and, so is not a problem for most users, but, for BS-treated alignments, I need to store a converted (Cytosine to Thymine) forward, and reverse copy of the genome which doubles the size of the reference…