vcfeval+hap.py qfy.py FP output

[Laura mora]

Hi all

I am performing a benchmarking of the GIAB samples with an in-house pipeline and I am currently using rtg tools vcfeval and qfy.py from hap.py to compare truth and query datasets.

I was a bit confused by the documentation of hap.py as it mentions that vcfeval comparison engine does not output "local matches" so FP.AL column from qfy.py scripts should be zero. 

In my hands, this is never the case. I found that in the rtg core 3.8.2 release announcement the following improvement is mentioned:

 vcfeval: GA4GH output mode now supports loose positional matching of

  variants (within +/-30bp by default, and adjustable via

  --Xloose-match-distance).

Does this mean that from that version onwards "local matching" (using hap.py nomenclature) is possible with rtg vcfeval and thus hap.py will classify FP.AL calls?

Thanks,

Laura

PS: I cannot find the --Xloose-match-distance option in the documentation.

[Len Trigg]

Hi Laura,

On Thursday, May 29, 2025 at 8:55:38 AM UTC+12 lauramo...@gmail.com wrote:

Does this mean that from that version onwards "local matching" (using hap.py nomenclature) is possible with rtg vcfeval and thus hap.py will classify FP.AL calls?

That is correct. The initial version of the GA4GH output mode didn't have the local matching support, and the hap.py documentation can't have been updated to take the newer release into account.

 

PS: I cannot find the --Xloose-match-distance option in the documentation.

Yes, any of the flags listed via "--Xhelp" and "--XXhelp" flags are not included in the main documentation, and may have sharp edges. Ask here if something sounds interesting.

Cheers,

Len.

 

[Laura mora]

Thanks for the clarification Len!