josephcmurray / vcf2omop

vcf2omop

vcf2omop

The Observational Health Data Sciences and Informatics (OHDSI) consortium develops the Observational Medical Outcomes Partnership (OMOP) common data model, which is widely used in Electronic Healthcare Record (EHR) research. The OMOP Genomic module is being developed to promote the integration of genomic and EHR data.

The OHDSI genomics workgroup has developed a staging file of genomic variants with relevance to oncology. This file assigns OMOP concept ids to each of these variants. In the OMOP CDM, each concept_id is an integer that is unique across all domains. It is intended that observations of genetic variants will be represented in the OMOP measurement table in the measurement_concept_id field.

vcf2omop is a Java command-line application that takes as input a VCF file and the OMOP staging file and annotates any variant that is represented in the staging file with the corresponding concept_id. Optionally, the app will also annotate variants with respect to their predicted effect on transcripts using the Jannovar library.

Authors:	Peter N Robinson, Daniel Danis
Version:	0.6.1 of 2021/09/10

Requirements

• Java runtime environment with version 8 or higher
• OMOP Genomic Staging File
• If you want to build vcf2omop from source, you will need maven.

TODO -- Provide information about where to get the OMOP staging file!

Setup

To build the tool, you can clone the repository and use maven.

$ https://github.com/pnrobinson/vcf2omop.git
$ cd vcf2omop
$ mvn package

This will create a Java app in the target directory. To check whether the build was successful, enter the following.

$ java -jar target/vcf2omop.jar
Usage: vcf2omop [-hV] [COMMAND]
Extract OMOP-encoded variants
  -h, --help      Show this help message and exit.
  -V, --version   Print version information and exit.
Commands:
  download  Download files
  synonyms  Generate table with all transcript "synonyms"
  omop      extract OMOP-annotated vars from VCF

The tool uses a transcript data file from the Jannovar project. Jannovar offers transcript definition files for NCBI RefSeq (hg19, hg38) and and Ensembl (hg19, hg38). Before using the app for analysis, run the download command to download the Jannovar transcript definition files.

$ java -jar target/vcf2omop.jar download

This will create a new directory called data and download the following four files to the directory

• ensembl_87_hg19.ser
• ensembl_91_hg38.ser
• refseq_curated_105_hg19.ser
• refseq_curated_109_hg38.ser

Annotate VCF files with OMOP concept ids

After setting up the app as above, all you need is a VCF file and a path to the OMOP staging file (represented as stage_genomic.csv in the following command). Then run the app with the omop command as follows.

$ java -jar target/vcf2omop.jar omop --stage <path/to/stage_genomic.csv> --vcf <path/to/vcf>

There are several options that can be seen with the -h flag.

$ java -jar target/vcf2omop.jar omop -h
Usage: vcf2omop vcf2omop [-hV] [--annot] [-a=<assembly>] [-d=<downloadDir>]
                       [--database=<genomeDatabase>] [-p=<prefix>]
                       -s=<omopStageFilePath> --vcf=<vcfPath>
extract OMOP-annotated vars from VCF
  -a, --assembly=<assembly>  genome assembly: hg19, GRCh19, hg38, GRCh38,
                             default null
      --annot                add transcript annotations via Jannovar
  -d, --data=<downloadDir>   location of download directory (default: data)
      --database=<genomeDatabase>
                           database: refseq, ensembl
  -h, --help                 Show this help message and exit.
  -p, --prefix=<prefix>      Outfile prefix
  -s, --stage=<omopStageFilePath>
                           path to OMOP stage file
  -V, --version              Print version information and exit.
      --vcf=<vcfPath>        path to VCF file

It is possible to combine the hg19 and hg38 (genome assembly) options with the refseq or ensembl options. For instance, to annotate a VCF file with variants called according to hg19 using ensembl transcripts, enter the following

$ java -jar target/vcf2omop.jar omop \
    --stage stage_genomic.csv \
    --vcf sample.vcf \
    -a hg19 \
    -d ensembl

Note that hg19 is the default and thus -a hg19 can be omitted.

The app does not check that your VCF file corresponds to the option you chose and will produce incorrect results if the options are incorrect.

Expected results

If the input VCF file contains variants in the OMOP staging file, the app will add an annotation to the INFO field of the VCF file. For instance, the INFO field of this variant

5 112177171 rs465899 G A 4052.01 PASS (...)

gets a new OMOP annotation

(...);MQ=37;MQ0=0;MQRankSum=0.485;OMOP=36746894;(...)

The remaining data in the VCF file is transmitted as is.

If the user chooses the --annot option, then the VCF file is additionally annotated with transcript level annotations for each variant. For instance,

$ java -jar target/vcf2omop.jar omop \
    --stage stage_genomic.csv \
    --vcf src/main/resources/sample-hg19.vcf \
    --annot

This command will add Jannovar annotations such as the following to the INFO fields of variants:

JANNOVAR=ENST00000507810.1(EPB41L4A):n.953-3219A>C (non_coding_transcript_intron_variant)
JANNOVAR=intergenic_variant
JANNOVAR=ENST00000257430.4(APC):c.136-230C>A (p.(=);coding_transcript_intron_variant)
JANNOVAR=ENST00000231136.1(PCDHB6):c.1908C>G (p.(H636Q);missense_variant)
JANNOVAR=ENST00000434307.2(SPAG11A):c.310T>G (p.(*104Eext*20);stop_lost)
JANNOVAR=ENST00000230354.6(TBP):c.213_215dup (p.(Q95dup);disruptive_inframe_insertion)

These annotations will be added to all variants (not just those with OMOP annotations).

Generate synonyms

This new feature generates a table of 'synoynms'.

$ java -jar target/vcf2omop.jar synonyms [-a hg19]

Expected results

This command will generate a file called synonyms-vcf2omop-GRCh19.tsv. Here is the output for the first two variants -- we see all of the transcript-level variants that correspond to the indicated genomic variant. One could run analogous commands to generate synonyms with refseq ids or for hg38.

omop.id chrom pos ref alt gene.symbol hgvs.genomic hgvs.transcript hgvs.protein 1801782 19 4099314 G A MAP2K2 g.4099314G>A ENST00000262948.5:c.804C>T p.(=) 1801782 19 4099314 G A MAP2K2 g.4099314G>A ENST00000394867.4:c.513C>T p.(=) 1801782 19 4099314 G A MAP2K2 g.4099314G>A ENST00000599021.1:c.29+1703C>T p.(=) 1801782 19 4099314 G A MAP2K2 g.4099314G>A ENST00000593364.1:n.751C>T 1801782 19 4099314 G A MAP2K2 g.4099314G>A ENST00000595715.1:n.619C>T 1801782 19 4099314 G A MAP2K2 g.4099314G>A ENST00000600584.1:n.1364C>T 1801782 19 4099314 G A MAP2K2 g.4099314G>A ENST00000601786.1:n.1105C>T 1801782 19 4099314 G A MAP2K2 g.4099314G>A ENST00000597263.1:n.169+1703C>T 1802019 9 21974760 C T CDKN2A g.21974760C>T ENST00000304494.5:c.67G>A p.(G23S) 1802019 9 21974760 C T CDKN2A g.21974760C>T ENST00000380151.3:c.67G>A p.(G23S) 1802019 9 21974760 C T CDKN2A g.21974760C>T ENST00000446177.1:c.67G>A p.(G23S) 1802019 9 21974760 C T CDKN2A g.21974760C>T ENST00000498124.1:c.67G>A p.(G23S) 1802019 9 21974760 C T CDKN2A g.21974760C>T ENST00000579122.1:c.67G>A p.(G23S) 1802019 9 21974760 C T CDKN2A g.21974760C>T ENST00000361570.3:c.317-3553G>A p.(=) 1802019 9 21974760 C T CDKN2A g.21974760C>T ENST00000530628.2:c.194-3553G>A p.(=) 1802019 9 21974760 C T CDKN2A g.21974760C>T ENST00000579755.1:c.194-3553G>A p.(=) 1802019 9 21974760 C T RP11-145E5.5 g.21974760C>T ENST00000404796.2:c.348-54672C>T p.(=) 1802019 9 21974760 C T CDKN2A g.21974760C>T ENST00000494262.1:c.-3-3553G>A p.(=) 1802019 9 21974760 C T CDKN2A g.21974760C>T ENST00000498628.2:c.-3-3553G>A p.(=)

TODO -- Decide what to do with variants that do not intersect with a transcript, e.g., 35981064: 5:1295228G>A (GRCh19) Currently the app says Could not annotate entry OmopStagedVariant: 35981064: 5:1295228G>A (GRCh19)! but this is not an error.!

To run the demo

We have spiked in 2 OMOP-relevant variants for a demo. The VCF file is available at

src/main/resources/sample-hg19.vcf

The corresponding VCF file is located here:

src/main/resources/sample.vcf

Therefore, run the app as follows.

$ java -jar target/vcf2omop.jar vcf2omop --vcf src/main/resources/sample.vcf

This is the output (both to the command line and to a file, which by default is called vcf2omop.tsv).

36740245 hg38 chr1 92836283 G A RPL5 6125 MISSENSE_VARIANT g.92836283G>A c.418G>A p.(Gly140Ser) 35981554 hg38 chr7 5982885 C T PMS2 5395 MISSENSE_VARIANT g.5982885C>T c.2113G>A p.(Glu705Lys)

Run the app with the --all flag to see all transcripts.