Hi,
We recently obtained some cram data (Human Whole Exome Seq) from our collaborators. While converting it to bam (using the same reference as they had used), we obtain the following error -
[E::cram_decode_slice] MD5 checksum reference mismatch for ref 0 pos 248638165..248848443
[E::cram_decode_slice] CRAM: bc527343c7ffc103111f3a694b004e2f
[E::cram_decode_slice] Ref : 46798b5cfca45c46a84b7419f8b74735
[E::cram_next_slice] Failure to decode slice
[main_samview] truncated file
The command used for converting cram to bam is -
/path/to/samtools-1.9/samtools view -b -T /path/to/hg38.fa -o /path/to/bam/test_1.bam /path/to/cram/test_1.cram
An important thing to note is that while we get this error, the download had completed successfully without any errors or warnings. I downloaded the data again since my interpretation of the error was that one of the cram slices has not downloaded properly, but the error persists with the same position being mismatched.
After throwing this error, the pipeline continues to run, but the resultant vcf files are around 8 MB (We had earlier obtained vcf files around 20 MB from the same pipeline for different samples). I am guessing that the size difference could be due to coverage difference, but this is the same error I get while converting to fastq, so I cannot do a fastqc either. Has anyone encountered such a problem, and if so, what was the solution that worked for you?
Thanks in advance.
P.S. Since the data is confidential, I have not shared the actual MD5 checksum here.
