.name attribute

[daler]

.name attribute

[daler]

GFF definition doesn't have mandatory attribute fields, so we can't assume ID. However, many of the files I use do have ID, and it would be convenient to have available as .name. Can we use the cython equivalent of "try/except KeyError" for this?

GTF requires both 'gene_id' and 'transcript_id' fields in the attributes. So in this case a single .name field doesn't make sense.

VCF . . . I'm not very familiar with the format but it looks like no required attributes either.

[brentp]

i added so it checks a few keys here:
brentp@47f2a03

and for VCF, just use chr:start

and see the comment. so maybe we can do:

b = BedTool('a.bed', name_key='transcript_id')

then b.name uses 'transcript_id' to look up the name.

[daler]

good idea with the hierarchy of possible name attributes. this returns None for GFF lines without one of those attrs, right?

i like this combined with the idea of setting an optional name key -- sensible defaults, but flexibility for the power users

[brentp]

aye, it returns None for now if not found.
what's the right thing to do there?
i dont think raising a KeyError is the right thing.

[daler]

yeah, i think None is the pythonic thing to do (i was just confirming) . . . and i agree that raising KeyError would be annoying.

[daler]

closing this issue (it works and we have tests for it)

[brentp]

re-opening as i want to remember implement getting column 3 for vcf files. something like:

return "%s:%s" % (fields[0], fields[1]) if fields[2] in (".", "") else fields[2]

http://www.1000genomes.org/node/101