Is mpileup limited in the number of samples?

[asimons]

I have been using mpileup with two to six samples for a while, with good results. Today I tried a run with 20 input bam files, and a small percentage of my output lines are truncated. I haven't been able to find a reference to this on the web; my apologies if it has been discussed before.

My use of mpileup is about as simple as can be; I'm just getting the pileups, not doing any calling:
samtools mpileup -f <reference> <bam-list>

With 20 input files, I expect 63 tab-separated fields per line. In a small percentage of lines, I'm getting 61 fields. When the line is short, it always has 61 fields. In a subset of my data, the total mpileup output is 26,633 lines; of those, 23 are truncated.

I have run mpileup multiple times on the full set of the data and the output files are identical.

If this is not a known problem and my subset bams would be helpful, I can point lh3 at them.

Thanks for any pointers.
-Al

[swbarnes2]

I never tried it with that many, but I think the place to start is to believe your data. Did you check to see that in those particular areas, that every single sample has at least one read? Because if a sample has no read there, displaying nothing would seem to be what mpileup would do.

[asimons]

These data are very "thin"; there are many locations that have no reads. However, the mpileup format handles that gracefully, reporting the triple "0\t*\t*" (\t represents a tab character) for any location that is not covered in a sample. So I don't think that these short/truncated lines are caused by the last sample not having a read there. These lines have the 0, but not the following two asterisks and their separating tabs.

[12jrowley2]

samtools mpileup truncated output

These data are very "thin"; there are many locations that have no reads. However, the mpileup format handles that gracefully, reporting the triple "0\t*\t*" (\t represents a tab character) for any location that is not covered in a sample. So I don't think that these short/truncated lines are caused by the last sample not having a read there. These lines have the 0, but not the following two asterisks and their separating tabs.

Has this been resolved, I am having the same problem with no asterisks after 0 with multiple sample mpileup