Is there any robust way to convert VCFs to SQL (for easier comparisons) in R or Ruby, Java and Python?

Question

[valentina_stepanenko]

Is there any modern robust way to convert VCFs into SQL (for easier comparisons) in R or Ruby, Java and Python?

[GouthamAtla]

What comparisons you would like to make that are not possible with existing tools ? Just curious.

[valentina_stepanenko]

Multiple and 4 producing readable, recordable and processable data, experience showed that GATK, NextGene and VAAST are not very good for VCF comparisons. Can you suggest any other software for VCF file comparison?

[d-cameron]

Many of the problems with VCF comparisons will not be solved by just putting the information into a database. The difficultly of VCF comparison is, at least in part, due to the ability to represent events in multiple ways. For example, a VCF containing a 1bp deletion and SNV 10bp upstream can be equivalent or not equivalent to a SNV in the deletion position and the deletion in the SNV position depending on whether the events are phased together and what the intervening sequence is. Even if you're just dealing with SNPs the problem doesn't go away as a SNV can have an upstream deletion with a compensating insertion downstream thus effectively rendering the nominal SNV coordinates incorrect. Ignoring indels, or even left-aligning them, isn't enough to resolve these issues in the general case.

Be aware that throwing VCF records into a database won't resolve all your issues.

Answer 1

[GouthamAtla]

VCFtools

vcflib

SnpSift

intersectBed Useful to quickly get overlapping stats.

[valentina_stepanenko]

thx, and what tools can convert VCFs to SQLite, MySQL or NoSQL MondoDB or Casandra?

Answer 2

[brian.repko]

ADAM will let you read VCFs and save as parquet file. Spark SQL might be available on that.