Smatools mpileup thinks my sorted bam files are not sorted

[plumb_r]

Hi
I am trying to run samtools mpileup on a large list of bam files

samtools mpileup -d 5000 -f /path/to/ref.fa \
/path/to/first.bam \
/path/to/second.bam \
| gzip > output.piledup

However, I get a log file which features a list saying

[bam_pileup_core] the input is not sorted (reads out of order
[bam_pileup_core] the input is not sorted (chromosomes out of order)

Corresponding, to each of my original input bam files and then a further list of

[bam_plp_destroy] memory leak: 2. Continue anyway.

With a line for each input file.
I have definitely used the samtools sort command to sort these files prior to using mpileup. However, if I use:

samtools view -H sorted.bam

I still get a header of @HD VN:1.0 SO:unsorted
So my questions are:

• Are my "sorted" bam files actually sorted?
• If they are not, how can I sort them if samtools sort doesn't seem to sort them?
• If they are sorted where else could the error be?

These bam files were intially aligned using bwa and converted from sam to bam with bwa
Thanks in advance for any help

[dnusol]

You could try Picard's SortSam.

HTH

[maubp]

What version of samtools do you have? Older versions of samtools never bothered to update the @HD line during 'samtools sort'.

[plumb_r]

0.1.18
I found an old thread which seemed to suggest that the @HD line might be being left as saying unsorted by samtools.
Is the mpileup totally dependent on this line saying unsorted or sorted?

[plumb_r]

For the benefit of closing this thread: the files sorted successfully with Picard SortSam. I have no idea why they wouldn't sort with samtools

[ksw9]

Hi,
I am having a similar problem using samtools phase on sorted and indexed BAM files and I keep getting the errors:

[bam_pileup_core] the input is not sorted (reads out of order)
[bam_plp_destroy] memory leak: 19. Continue anyway.

My bam file is sorted. I have also tried to use Picard SortSam, but get the error:
Error: Unable to access jarfile INPUT=Sample_Bbcap31_L002.bam

Does anyone have advice on how to proceed? Thank you!

[rkizen]

I was having the same problem with mpileup not recognizing that my bam file was sorted. I had tried sorting with samtools and with picardtools but neither seemed to solve the problem.

For my case it appears that the issue was that one read pair had a read mapping at the very beginning of the contig and the other read not mapping. These were on the reverse strand. This meant that the first two reads are shown with a coordinate of 0. I am not sure what is causing the error exactly but removing this pair of reads resolves the issue.

Perhaps someone else has seen this as well?

Version is 1.18