Trouble using 'first' op on dbNSFP txt file

[IvantheDugtrio]

I having trouble removing multi-allelic annotations from the dbNSFP reference text file, as it seems to always parse these fields as 'self'. I am thinking the problem has to do with limitations in parsing a text file versus a VCF. Is there a workaround?

[brentp]

my multi-allelic, you mean it has one line per alternate allele? i think self is the only option that will work correctly for that. what is your config?

[IvantheDugtrio]

By multi-allelic, I mean one column, say HGVSc_snpEff, has multiple annotations separated by commas, each corresponding to a different transcript ID, while the rest of the file is tab-delimited.

For example, one annotation line can have an HGVSc_snpEff field that looks like:
c.44C>G,c.44C>G,c.57C>G,c.57C>G

My config is as follows:

file=dbNSFP.txt.gz
columns=[13,21,22]
ops=["first","first","first"]
names=["dbNSFP_genename","dbNSFP_HGVSc_snpEff","dbNSFP_HGVSp_snpEff"]

[brentp]

hmm. I'm not sure this can work. what does ALT look like for that dbNSFP line?

[IvantheDugtrio]

Oh, I mis-stated, the multiple annotations in the same column are separated by semicolons. An actual line from the file looks like in the attached.

The REF and the ALT are still just single entries, in this case, a C for the REF, and a T for the ALT.
T315I.txt

[liserjrqlxue]

dbNSFP has multi records of same variants which have different amino acid change.
This may also case unexpect result.