SAMTools index stats strange result?

[tatinhawk]

Hello Everyone,

I made an index and it's respective index stats from a previously sorted BAM File. The BAM file contains the alignment of single-end SOLiD reads to the human genome. What I don't understand is that in the "# of unmapped reads" column of the indexstats output, you can find references with values that differ from 0. (e.g. "gi|251831106|ref|NC_012920.1| 16569 19656 3"). This is weird, becuase I don't understand how a read can be "unmapped" preferentially to a specific reference, instead of being counted as part of the "*" row?


Cheers,
Martin

[daoud]

It might be that these reads map at the end of a chromosome. At least for BWA the chromosomes are actually concatenated head to tail prior to mapping. When a read spans such a junction, it will be marked as unmapped while retaining the chromosome annotation.

So if all these reads are at the end of a chromosome, this is probably your answer.

Grtz,

Daoud

[tatinhawk]

It might be that these reads map at the end of a chromosome. At least for BWA the chromosomes are actually concatenated head to tail prior to mapping. When a read spans such a junction, it will be marked as unmapped while retaining the chromosome annotation.

So if all these reads are at the end of a chromosome, this is probably your answer.

Grtz,

Daoud

Thanks, that must be the case.

[swbarnes2]

The other reason an unmapped read would have a mapping position is that .sam file specs call for unmapped reads to be given the mapping coordinates of their mapped mates, if they have one. This is so they will sort next to each other.

If the 4 flag is checked, then the read didn't map, no matter what anything else says.