Potentially corrupted BAM files

[lre1234]

Hi all,
I have a question about detecting possible corruptions to BAM files. Essentially, we had a serious hard drive issue and failure across our network. While we have been able to restore our hard drives and retrieve our data, we have noticed issues with many files. for example, in some text files, some information has been changed to non ascii data, or lines deleted from them. We still do not yet know the full extend of the issues but are trying to figure it out. One, potential issue is that we have many 1000's of mapped bam files. We do not know if the files have been corrupted in anyway, and are interested in finding out. One approach (I think) would be to try to convert them back to SAM files, and if it breaks then the file is somehow corrupted. But I was wondering if anyone knew of some other method that can check the file.

I should note that we do have most of this backed up on a tape storage system that we could go back to, but this would take an exceedingly long time to do for our whole system. So we would like to maybe find the 'bad' files and replace those only. Also, I am not a software engineer, so hopefully this all makes sense

[Richard Finney]

zcat the bam file and check for errors ... example ...

rfinney@pigdog:~$ file test.bam
test.bam: gzip compressed data, extra field

rfinney@pigdog:~$ samtools index test.bam

rfinney@pigdog:~$ cat test.bam | tr " " "x" > test2.bam # corrupt the file test2.bam

rfinney@pigdog:~$ samtools index test2.bam
[E::bam_hdr_read] invalid BAM binary header
samtools index: "test2.bam" is corrupted or unsorted

rfinney@pigdog:~$ zcat test.bam > /dev/null

rfinney@pigdog:~$ zcat test2.bam > /dev/null
gzip: test2.bam: invalid compressed data--crc error
gzip: test2.bam: invalid compressed data--length error