[E::cram_next_slice] Failure to decode slice

[bef22]

I've been using the same version of crumble (0.9.1) before on the Cambridge hpc cluster without any problems on lots of cram files (Illumina PE 151, ~ 170,000,000 reads) using data mapped to the same reference genome. Now I suddenly get the same error of [E::cram_next_slice] Failure to decode slice which results in only chr1 data to be compressed.

The command I use is
crumble -O in.cram out.cram

crumble 0.9.1 works fine on small test files with ~1000 sequences. So I'm really not sure what is going on.

Best wishes,
Bettina

[bef22]

Sorry the command is:
crumble -O cram in.cram out.cram

[bef22]

I traced the problem down to the samtools version the cram was created. It works with samtools 2.20, but fails with 2.22 and 2.23. I assume it's related to the samtools cram update in 2.22 - default output CRAM version from 3.0 to 3.1. HTSlib and SAMtools have been able to read CRAM 3.1 since version 1.12, however other tools may not yet be able to cope.

Best wishes,
Bettina

[jkbonfield]

Hello Bettina,

On Fri, Feb 13, 2026 at 01:32:08AM -0800, Bettina Fischer wrote: I traced the problem down to the samtools version the cram was created. It works with samtools 2.20, but fails with 2.22 and 2.23. I assume it's related to the samtools cram update in 2.22 - default output CRAM version from 3.0 to 3.1. HTSlib and SAMtools have been able to read CRAM 3.1 since version 1.12, however other tools may not yet be able to cope.

Sorry for the slow reply. I'm glad you've found an alternatice, but this surprises me still as the first CRAM 3.1 implementation was in io_lib which crumble is built against. Do you know when your version of Crumble was built? Thanks, James

…

-- James Bonfield ***@***.***). Work days Mon-Thu only. The Sanger Institute, Hinxton, Cambs, CB10 1SA

---------------------------------------------------------------------- The Wellcome Sanger Institute is operated by Genome Research Limited, a charity registered in England with number 1021457 and a company registered in England with number 2742969, whose registered office is Wellcome Sanger Institute, Wellcome Genome Campus, Hinxton, CB10 1SA.

[jkbonfield]

Sorry I take that back: crumble uses htslib and not io_lib. So it should just be a matter of rebuilding crumble using a modern htslib release. Obviously you can use htslib to output CRAM 3.0 first and convert back again later (if desired) as an alternative. James

…

-- James Bonfield ***@***.***). Work days Mon-Thu only. The Sanger Institute, Hinxton, Cambs, CB10 1SA

---------------------------------------------------------------------- The Wellcome Sanger Institute is operated by Genome Research Limited, a charity registered in England with number 1021457 and a company registered in England with number 2742969, whose registered office is Wellcome Sanger Institute, Wellcome Genome Campus, Hinxton, CB10 1SA.