samtools mpileup segmentation fault

[adrian]

hi:

I clearly do not understand behavior of samtools mpileup. it is frustrating at times for not knowing the exact reasons for segmentation fault.

1. is seg-fault in mpileup related to a bam file? If so how do I know why it segfaults?

here is my segfault:
samtools mpileup -d 100000 -f hg19.fasta -g bc1.bam | bcftools view -e -g -c -v -N -i -1 -I -

(here I clearly gave -I -- to avoid indels)

chr3 10183782 . T G 5.29 . // info truncated///
[afs] 0:1003.293 1:3.825 2:8.882
Segmentation fault


2. I take a different bam file and enforce -I -- I get segfault with a different message. In a previous analysis this bam file went through bcftools pipe.

[bcf_sync] incorrect number of fields (6 != 5) at 2:10191496
[afs] 0:1438.765 1:13.635 2:23.601
Segmentation fault


Is this due to corrput bcftools or samtools.




3. Cannot exactly understand the reason for segfault in both 1 and 2.


Is this because I have some problem in bam file or bcftools got corrputed somehow?

This bam is generated through (PGM) tmap aligner.



thanks for your help.

-Adrian

[yina]

Notice that it´s necesary to sort the bam file before calling the pileup:

samtools sort alignment.bam > alignment.sort

hope it helps

[swbarnes2]

Rather than piping into bcftools, why don't you try just making a pileup file. Then you'll know whether or not bcftools is respnsible.

One quick thing worth checking: remake the reference index with samtools faidx, and confirm that it gets made with no error. mpileup will check to see if the file exists, and it will make it if it doesn't see a file with the right name, but if the index file was made wrongly, mpileup won't know what to do.

[bioman1]

Hi,
I think first you can create mpileup file without error and then try for bcftools.
Follows these steps

1.samtools faidx reference.fa (this will produce reference.fai file)
2.samtools import reference.fai sample.sam sample.sam.bam
3.samtools sort sample.sam.bam sample.sorted.bam
4.samtools index sample.sorted.bam sample.sorted.bam.bai
5.samtools mpileup reference.fa sample.sorted.bam.mpileup

Please be careful in giving file extensions

For tyring bcfltools you must need both sorted.bam & bam.bai file

I hope this helps

[swbarnes2]

Hi,
I think first you can create mpileup file without error and then try for bcftools.
Follows these steps

1.samtools faidx reference.fa (this will produce reference.fai file)
2.samtools import reference.fai sample.sam sample.sam.bam
3.samtools sort sample.sam.bam sample.sorted.bam
4.samtools index sample.sorted.bam sample.sorted.bam.bai
5.samtools mpileup reference.fa sample.sorted.bam.mpileup

Please be careful in giving file extensions

For tyring bcfltools you must need both sorted.bam & bam.bai file

I hope this helps

samtool import is deprecated, and you don't actually need an .bam index to run mpileup. And running sort like that will make a file called sample.sorted.bam.bam. And you forgot to specify a input .bam in your pileup command.

[cmmay1]

Segmentation error FIX

When you try the command, samtools faidx ref.fasta
You may get this error message: [fai_build_core] different line length in sequence 'scaffold50'. Segmentation fault.

Use sed '/^$/d' myFile > tt to reformat the linebreaks.

You can reformat text files in the BASH using several SED commands such as cut/sort/replace/find.

[melop]

I'm extremely annoyed by this problem too.
The problem seems to have been caused by bcftools when it's trying to convert the uncompressed bcf output from samtools mpileup. This might be caused by some unknown glitches in the bam file. It's extremely hard to pinpoint the problematic read though because usually it takes days to get to that point and when it hangs there's no record of where it stopped. Disabling indel calling by supplying -I seems to solve some but not all "incorrect number of field" error.
The bottom line is: if there's a glitch in the bam file or the uncompressed bcf output, bcftools should just give a warning, make a record of it instead of failing altogether wasting several days of computation. Someone should improve the source code.

[figo1019]

I'm extremely annoyed by this problem too.
The problem seems to have been caused by bcftools when it's trying to convert the uncompressed bcf output from samtools mpileup. This might be caused by some unknown glitches in the bam file. It's extremely hard to pinpoint the problematic read though because usually it takes days to get to that point and when it hangs there's no record of where it stopped. Disabling indel calling by supplying -I seems to solve some but not all "incorrect number of field" error.
The bottom line is: if there's a glitch in the bam file or the uncompressed bcf output, bcftools should just give a warning, make a record of it instead of failing altogether wasting several days of computation. Someone should improve the source code.

I am also fed up of similar error
[afs] 0:99774.498 1:110.438 2:115.064
[bcf_sync] incorrect number of fields (4 != 5) at 15:16294867

Have you found the reason for it . If you can share that that would be a great help

Regards

[melop]

Unfortunately I haven't found a solution for it. Perhaps try remaking the bam files would help?

I am also fed up of similar error
[afs] 0:99774.498 1:110.438 2:115.064
[bcf_sync] incorrect number of fields (4 != 5) at 15:16294867

Have you found the reason for it . If you can share that that would be a great help

Regards