README.md

metadata

license: cc0-1.0
task_categories:
  - tabular-classification
tags:
  - genomics
  - dbsnp
  - rsid
  - cross-build
  - GRCh37
  - GRCh38
  - bioinformatics
  - sqlite
pretty_name: Allelix rsID Identity Table
size_categories:
  - 1M<n<10M

Allelix rsID Identity Table

Pre-built SQLite cache mapping rsID allele positions between GRCh37 and GRCh38 genome builds, derived from dbSNP build 157 for use with Allelix.

What's in the file

rsid_identity.sqlite.gz is a gzipped SQLite database containing the rsid_identity table with 4,851,120 rows covering 3,046,373 rsIDs that appear in at least one Allelix annotation database (ClinVar, GWAS Catalog, SNPedia, PharmGKB).

Schema

CREATE TABLE rsid_identity (
    rsid TEXT NOT NULL,
    chrom TEXT NOT NULL,
    pos37 INTEGER NOT NULL,
    ref37 TEXT NOT NULL,
    alt37 TEXT NOT NULL,
    pos38 INTEGER NOT NULL,
    ref38 TEXT NOT NULL,
    alt38 TEXT NOT NULL,
    palindromic INTEGER NOT NULL
);

CREATE INDEX idx_rsid_identity_rsid ON rsid_identity(rsid);
CREATE INDEX idx_rsid_identity_pos37 ON rsid_identity(chrom, pos37);
CREATE INDEX idx_rsid_identity_pos38 ON rsid_identity(chrom, pos38);

Columns

Column	Description
rsid	dbSNP rsID (e.g. rs429358)
chrom	Chromosome (1-22, X, Y, MT; no "chr" prefix)
pos37	Genomic position on GRCh37
ref37	Reference allele on GRCh37
alt37	Alternate allele on GRCh37
pos38	Genomic position on GRCh38
ref38	Reference allele on GRCh38
alt38	Alternate allele on GRCh38
palindromic	1 if the allele set equals its complement (strand-ambiguous), 0 otherwise

Multi-allelic sites

An rsID with multiple alternate alleles has one row per allele. For example, rs1801133 (MTHFR C677T) has three rows for its three observed alternate alleles (A, C, T).

Inclusion criteria

An rsID is included only when it has exactly one primary-assembly placement per build, on the same chromosome, with allele sets that match on the same strand or (for SNVs) after complementing. Exclusions from the 3,073,779 rsID universe:

Reason	Count
Only placed on GRCh37	196
Only placed on GRCh38	296
Multiple placements on one build	685
Alleles unpairable between builds	205
Chromosome changed between builds	2

How Allelix uses this

Allelix downloads this file automatically during allelix db update. The identity table serves three purposes:

1. Reference allele population for array data on both GRCh37 and GRCh38.
2. rsID recovery for rsID-less VCF files on both builds, via locus-based lookup.
3. GRCh37 → GRCh38 coordinate translation for cross-build enrichment against gnomAD, AlphaMissense, and CADD. Palindromic sites abstain from translation to avoid strand-ambiguous mismatches.

Building from source

The cache can be rebuilt from dbSNP VCFs using the build script included with Allelix:

python scripts/build_rsid_identity_cache.py \
    --data-dir /path/to/dbsnp/vcfs \
    --output rsid_identity.sqlite \
    --work-dir /tmp/identity-build \
    --gzip

The build streams both GRCh37 and GRCh38 dbSNP VCFs through an 8 MB Bloom filter, then joins candidates in SQLite. Peak memory is ~40 MB; runtime is ~3 hours on a 1 GB VPS.

Integrity

Property	Value
File	rsid_identity.sqlite.gz
Size (compressed)	148 MB
Size (uncompressed)	436 MB
SHA-256	2a00787cd0be4e7572efd21686013c49f3d0180137cd4403730524d50a5db285
Rows	4,851,120
Distinct rsIDs	3,046,373

Source and license

• Source: NCBI dbSNP build 157 (GRCh37 VCF: GCF_000001405.25; GRCh38 VCF: GCF_000001405.40)
• License: CC0 1.0 / Public Domain — derived from NCBI dbSNP, a United States Government Work in the public domain
• Citation: Sherry ST, et al. "dbSNP: the NCBI database of genetic variation." Nucleic Acids Res. 2001;29(1):308-311.