Ag3 metadata and SNP data sample count mismatch

[tristanpwdennis]

There seems to be a mismatch between the number of samples returned by sample_metadata() and the samples available in snp_calls() for Ag3.

• sample_metadata(sample_sets="3.0") returns 3,453 samples
• snp_calls(region="3R", sample_sets="3.0") has 3,081 samples
• Difference: 372 samples exist in metadata but not in SNP data

This occurs both on my local and on colab - I first picked it up because it was causing my PR to fail CI .

It seems to stem from the 'AG1000-X' lab crosses (so happens in later releases too - the tests were just specifying 3.0...).

See in colab here.

[leehart]

This is probably related to the fact that the AG1000G-X sample set ought to go through an extra QC filter. I suspect this has been missed at some point, resulting in the metadata containing records for AG1000G-X samples that actually failed QC.

We will need to trace the origin of the misalignment. I believe there should be 297 released samples in AG1000G-X, rather than 669.

[jonbrenas]

I am 99% certain that I created the issue when I re-released the general metadata for all samples to remove the duplicated locations. I'll try to correct that ASAP.

[jonbrenas]

Both functions now return 3081 samples.

[leehart]

Hi @tristanpwdennis . Could you please verify that this issue has now been resolved for you? Thanks to the fix to the sample metadata by @jonbrenas .

[tristanpwdennis]

Yes - both functions return 3081 and tests are passing this now. Thanks @jonbrenas for picking this up so quickly!