Jerome Kelleher

I think this is ready for a look now. There is a little duplication in the tests I've added for the sample VCF and existing ones which have been fixed...

Sigh - skipping those tests pushes the required coverage below 100% so the build still fails. Any suggestions here? Temporarily push coverage down, and create an issue to track setting...

Thanks @tomwhite - I'm happy to go with your preferred option. If we can fix the other problems fairly easily in this PR then that would be simplest. I'm just...

OK, this is ready for another look @tomwhite. I've temporarily worked around some fiddly corner cases - hopefully we'll still have 100% coverage. #1197 is a tricky one, but I...

OK! Looks like it's ready for a final look now @tomwhite. @timothymillar, this is a reasonably big change so would be good to get your eyes on it as well...

Related: - #1138 - #1129 - #1132 Upstream dask issue: https://github.com/dask/dask/issues/10654

CALL_AD and CALL_PL are pretty weighty - is there an argument for doing these super heavy fields on their own in iterative passes? (OK you read in the VCF multiple...

Could reindex with tabix as a quick workaround, and see if the code works in that case?

Right - can't have an index anywhere else except in the same dir. Sigh.

I think this can be fixed by adding some functionality to cyvcf2. I made a start on something similar here: https://github.com/brentp/cyvcf2/pull/295 Alternatively, we are actually fully parsing the indexes locally,...