Bowtie call to get unique, multi-hits and nonmatching reads

[PFS]

Hello there,

I want to estimate (1) the number of reads that map uniquely (in one place only), (2) the number of reads that map in multiple places and (3) the number of reads that do not map at all.

Is this the correct bowtie call?

-m 1 --un nomatch_reads.txt -- max multihits_reads.txt

Thanks!

[kwatts59]

First find the number of reads that map at least once. To do this, run bowtie with "-k 1" option. Lets call this number K1.

Then run bowtie with the "-k 2" option. This will report up to two alignments per read. Lets call this number K2.

Subtracting the K1 from K2 gives the number of reads that mapped two or more times. Lets call this number K2p.

To find the number of reads that mapped exactly once, subtract K2p from K1.

To find the number of reads that did not map at all, just subtract K1 from the total number of reads in your sample.

[chadn737]

Not sure if this is what you want, but a few months back I wrote a very simple perl script (probably not very good, I'm a biologist not a programmer, so my scripts tend to be brute force approaches) that will take a sam file and spit out the names of the reads that map more than once and how many times they mapped.

Now I can't even remember what I was using it for, but maybe it will help you get to where you need.