A VCF Parser written in Rust

It reads a VCF file (plain or gzipped) and parse the output to a JSON/TSV format. The input VCF needs to be normalised so that there is no more than one alt per line. By default it uses all cores available to process the input.

Features

• Unnest INFO field
• Explode CSQ-like fields
• Join exploded fields by transcript_id
• Accept arbitrarily sophisticated filters defined in a yaml file.
• Multithreaded
• Output in JSON or TSV format
• Support logic expression for filters (experimental)

Installation

Assume you have cargo installed, you can install it with:

cargo install --path .

There's a binary on marjorie already in /nfs_home/projects/departments/nnrco/genetic_department/bin

Usage

vcf_parser -i test/test.vcf >output.tsv

Or from stdin

cat test/test.vcf | vcf_parser >output.tsv

Parse multiple CSQ-like fields and join by transcript_id, and filter variants as defined in filter.yml, output json format

vcf_parser -i test/test.vcf --fields CSQ,Pangolin --fields-join Feature,pangolin_transcript -f filter.yml --output-format j >output.json

An example of filter.yml can be found in the test/ folder. You can replace eq with = or ==, le with <= or ≤, etc. Equivalently, you can now pass a logic expression as a string to the -f option. For example:

vcf_parser -i test/test.vcf -f "(info.AF <= 0.01 AND info.CSQ.IMPACT in (HIGH, MODERATE)) AND (info.CADD_PHRED >=20 OR info.Pangolin.pangolin_max_score >= 0.5 or info.Pangolin.pangolin_max_score <= -0.5)" --fields CSQ,Pangolin --fields-join Feature,pangolin_transcript

NOTE: As in the current implementation, "AND" and "OR" have the same precedence. Please use parentheses to make the logic expression unambiguous.

If you have a list of genes to filter stored in a file, you can do it now:

vcf_parser -i test/test.vcf -f "info.CSQ.Gene in (`cat genes.txt`)" --fields CSQ,Pangolin --fields-join Feature,pangolin_transcript

Options

-h #help 
-i <input.vcf[.gz]>  
-f <filter.yaml or expression>
-t <thread number>
-l # to list columns and exit
-c <columns to output>
--output-format <j|t> #j for json, t for tsv
--fields #fields to explode. default to CSQ
--fields-join #keys to join fields, in the same order