Differential header stringency depending on file format

[ernfrid]

I noticed this in I'm observing differential error checking depending on whether or not the input file is a CRAM or a BAM. The files in question have multiple sample names listed (a different one for each @RG line). When the input is a CRAM, no error is thrown. When the input is a BAM, I see: panic: bam reagroup: more than one RG for /build/test.bam

At the moment, it seems as if indexcov doesn't check CRAM headers? i.e. https://github.com/brentp/goleft/blob/master/indexcov/indexcov.go#L202-L231

I assume this error is thrown because the assumption is that there is a single sample for the whole file and there isn't handling of multiple samples. What is being reported when these problem CRAMs are provided? Stats for all the samples pooled together?

[ernfrid]

For context, I'm seeing this error when running Smoove.

[brentp]

yes, for CRAM it will report the sum of all samples. it should be checked in CRAM too. the index can't know about the different samples.

[brentp]

smoove won't work with multiple samples per bam (I don't think lumpy will either).

[ernfrid]

Yeah, that's what I'd expect. They're not actually multiple samples though...just mislabeled single samples. If by happy circumstance it was all erroneously analyzed as a single sample, then that may not be the worst thing...