Search
Question: CollapsedVCF and ExpandedVCF rownames and vcfs with large ref/alt columns
0
gravatar for Sean Davis
11 months ago by
Sean Davis21k
United States
Sean Davis21k wrote:

With some newer VCF files, particularly those that have come from structural variant callers, the rownames of the resulting `VCF` objects (when the ID column is missing) can become problematic or excessively large. To give a concrete example, we have VCFs with insertions of dozens or hundreds of bases that lead to very long rownames. While human-readable rownames are useful in some cases, would it be possible to maintain uniqueness while enforcing some further constraints on the constructed rownames to keep them manageably short and printable?   

ADD COMMENTlink modified 11 months ago by Valerie Obenchain ♦♦ 6.4k • written 11 months ago by Sean Davis21k
1

Or perhaps also have an option to drop rownames altogether...

ADD REPLYlink written 11 months ago by Michael Lawrence9.8k
2
gravatar for Valerie Obenchain
11 months ago by
Valerie Obenchain ♦♦ 6.4k
United States
Valerie Obenchain ♦♦ 6.4k wrote:

Yes, I can imagine that's the case. The pasting of CHROM:POS_REF/ALT was a reasonable solution when ALT was just a few bases. I'm open to other suggestions of how to handle this - what information would you like to see as the row names?

We do have the option to turn off rownames with readVcf(..., row.names=FALSE).

Val

ADD COMMENTlink written 11 months ago by Valerie Obenchain ♦♦ 6.4k

Thanks, Val.  row.names=FALSE is a good solution for our use case. If unique IDs are of interest, hashing approaches would be potentially interesting to folks, but perhaps that is best left as a user-level decision. Part of my rationale for asking was to see what others thought. 

ADD REPLYlink written 11 months ago by Sean Davis21k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 218 users visited in the last hour