-This track merges variants from all individual variant frequency databases into a single -bigBed file with predicted protein consequences and cross-database filtering. It contains -over 1.1 billion variants from 26 source databases worldwide. For a summary of -all available databases, see the -Variant Frequencies supertrack page. +This track merges variants from 28 sequencing-based variant frequency databases into a +single bigBed file with predicted protein consequences and cross-database filtering. It +contains 1.34 billion variants from WGS, WES, and long-read sequencing cohorts worldwide. +For a summary of all available databases, see the +SNV Frequencies supertrack page.
++Two companion combined tracks split out the cohorts that don't belong in a general +sequencing-based summary: +
+Each variant is annotated with its predicted consequence on protein-coding genes (using bcftools csq with Ensembl -gene models), and colored by severity. -Allele counts and frequencies are shown for each source database and, where available, -broken down by ancestry or population group. +gene models), and colored by severity. Allele counts and frequencies are shown for each +source database and, where available, broken down by ancestry, population, or phenotype.
Variants are colored by their most severe predicted consequence:
| Color | Consequence class | Examples |
|---|---|---|
| Red | Protein-truncating / Loss-of-function | stop_gained, frameshift, splice_donor, splice_acceptor, stop_lost, start_lost |
| Blue | Missense / In-frame | missense, inframe_insertion, inframe_deletion, protein_altering |
| Green | Synonymous | synonymous, stop_retained |
| Grey | Non-coding / Intergenic | intron, non_coding, intergenic, UTR |
The "AA change" field uses bcftools csq notation: 23I>23V means position 23 changed from Isoleucine (I) to Valine (V) (missense). 23I alone (no arrow) means position 23 is Isoleucine and unchanged (synonymous). A "*" indicates a stop codon (e.g. 45R>45* is a stop_gained).
This track supports extensive filtering via the track settings page. Click on the track title or use the "Configure" button to access filters:
stop_gained,frameshift is selected by either the "Stop Gained"
or the "Frameshift" filter. The "Other" bucket catches the less
common Sequence Ontology
consequence terms emitted by bcftools csq that don't fit the named
buckets above. Examples include
splice_region (variant near a splice site but outside the canonical
donor/acceptor),
start_lost / stop_lost (variant disrupts the start codon
or replaces the stop codon with a coding amino acid),
stop_retained (variant changes the stop codon but keeps it a stop),
inframe_insertion / inframe_deletion (in-frame indel
that adds or removes whole codons), and
coding_sequence (CDS variant where the precise impact is undetermined).
If you include "Other" in the filter selection, no records will be
hidden by the consequence filter.How to find protein-truncating variants: Set the Consequence filter to include only "Stop Gained", "Frameshift", "Splice Donor", and "Splice Acceptor". These will appear as red items in the track display.
The Source Database filter lets you restrict to variants present in specific databases. For example, select only "GREGoR" to see variants found in the rare disease cohort. This filter uses OR logic: selecting multiple databases shows variants found in any of the selected databases.
-Several databases provide ancestry-specific allele frequencies:
+Three sources also expose phenotype-stratified counts: +
++The disease-related Disease-related Databases Combined +track exposes additional phenotype splits for SCHEMA (Schizophrenia case vs control). +
-Variant frequency VCF files from 26 databases were stripped of their INFO fields
-(to reduce size), normalized with bcftools norm (splitting multi-allelic sites),
-and merged with bcftools merge. The merged VCF was then annotated with predicted
-protein consequences using bcftools csq with the
+Variant frequency VCF files from 28 sequencing-based databases were stripped of their INFO
+fields (to reduce size), normalized with bcftools norm (splitting multi-allelic
+sites), and merged with bcftools merge. The merged VCF was then annotated with
+predicted protein consequences using bcftools csq with the
Ensembl
-GRCh38 release 115 gene annotation (GFF3).
+GRCh38 release 115 gene annotation (GFF3). The same pipeline, run on different subsets of
+source VCFs, produces the
+Disease-related Databases Combined and
+Genotyping Array Databases Combined tracks.
The annotated VCF was converted to bigBed format using a custom Python script
(vcfToBigBed.py) that reads frequency data from each source VCF in parallel,
matches variants by position/ref/alt, and writes a BED file with consequence coloring,
per-database allele counts and frequencies, and population breakdowns.
The database configuration (which VCFs to include, field mappings, and population definitions)
is stored in two TSV files
(databases.tsv and
populations.tsv)
so that future updates only require editing these files.
The track's makeDoc file documents how each source VCF was converted. Scripts are available from Github.
The data can be explored interactively with the Table Browser or the Data Integrator. For programmatic access, our REST API can be used; the track name is varFreqsAll.
Because the merged callset includes data from multiple sources whose redistribution licenses differ, the combined bigBed is not available for download from our download server. The combined track can be reconstructed from the individual source VCFs using the conversion scripts on GitHub together with the build documentation. Where individual source data is downloadable from UCSC, the per-subtrack description page indicates the path on our download server.
This track is only possible thanks to the data from millions of volunteers around the world, who donated blood, signed consent forms and provided health information about themselves and sometimes their families. Click on any of the individual tracks in the -Variant Frequencies supertrack to see the specific +SNV Frequencies supertrack to see the specific credits for each project. Thanks to Alex Ioannidis, UCSC, for the motivation for this track and to Andreas Lahner, MGZ, for feedback.
For primary citations of each source dataset, see the References section on the -Variant Frequencies supertrack page. The merged-track +SNV Frequencies supertrack page. The merged-track build itself uses the following tools:
Danecek P, McCarthy SA. BCFtools/csq: haplotype-aware variant consequences. Bioinformatics. 2017 Jul 1;33(13):2037-2039. PMID: 28205675; PMC: PMC5870570
McLaren W, Gil L, Hunt SE, Riat HS, Ritchie GR, Thormann A, Flicek P, Cunningham F. The Ensembl Variant Effect Predictor. Genome Biol. 2016 Jun 6;17(1):122. PMID: 27268795; PMC: PMC4893825