Some MuTect files are improperly joined

[lbeltrame]

When testing the pipeline after fixing the cluster problem, I noticed that the GATK errored out much later on with this:

The provided VCF file is malformed at approximately line number 4: The VCF specification does not allow for whitespace in the INFO field

Looking at the VCF file, I noticed the following:

##fileformat=VCFv4.1
## No variants; no reads aligned in region
#CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO
chr10   89685461        .       G       T       .       REJECT  AC=0;AF=0.00;AN=1;DP=456        GT:AD:DP:FA     0:454,1:456:2.198e-03

As you can see, it's an empty VCF that's however been "filled" somehow. The reason is that MuTect manages to write data before erroring out (with that famous Java 7 error we discussed), therefore generating a wrong VCF file.

I'm not sure there's an easy solution for this, given that the GATK is now requiring Java 7 and MuTect does not work yet properly with it.

On the plus side, this gives me more motivation to implement support for other paired callers. ;)

[lbeltrame]

Part of the sorting error is being fixed in broadinstitute/mutect#2 (although perhaps a different way will be implemented in the end).

[chapmanb]

Luca;
Thanks for the report. This is a bug in how I do the concatenation of files. The final concatenated files gets the header of the first file, which will be problematic if the first file is empty. I checked in a fix to avoid this so hopefully it'll work cleanly now. Let me know if you run into any other issues.

[lbeltrame]

Thanks Brad, it works correctly now (run ended successfully).