Unexpected behaviour when parsing lists in INFO field

[holtgrewe]

I have a VCF file with an INFO field that has A many values, but these values can be empty, as demonstrated below. Reading this with HTSJDK leads to the FI and FS fields having one atomic value instead of a list of two values.

Is this intented behaviour?

1       69360   .       C       A,G     .       .       RV=C>*,C>W;GN=OR4F5;UG=OR4F5_HUMAN;VI=stop_gain,;UV=C90*,C90W;FI=,high;FS=,4.765

[yfarjoun]

going over old issues...

The missing string should be the empty string, but it isn't clear the the double array is legal since the spec doesn't seem to allow for missing individual fields in an array. so that would have to be an error.

I think that if strict pasrsing is requested, this record should be deemed not conformant due to the float array, and the string array should have returned one empty string.

[holtgrewe]

Thanks for considering this issue.

I don't see in the specs where empty strings are disallowed.

[yfarjoun]

I agree.. an empty string should have been created, but not a missing float