GeneBEDs resources

GeneBED resources are transcript-level annotations. all contains every transcript isoform; deduplong keeps one longest isoform per gene. The def link points to the registry-defined default version.

Bundled GENCODE builds include hg19, hg38, mm9, mm10, and mm39. Other assemblies are resolved from GENCODE, UCSC, or Ensembl.

Commands

sjcab-peak2anno-db install-genebed
sjcab-peak2anno-db install-genebed mm10 vM22 -g gencode.vM22.annotation.gtf.gz
sjcab-peak2anno-db download-genebed hg38 v31 -o annotations
sjcab-peak2anno-db download-genebed hg19 v31lift37 -o annotations
sjcab-peak2anno-db install-genebed hg38 -name human

For Ensembl, def, default, current, and latest resolve the current release. Vertebrate and Ensembl Genomes catalogues are cached by release; def links to the current release. Assembly, species, name, and assembly_accession can be used as species identifiers when unique.

For UCSC builds, GTF metadata and liftOver chain URLs are cached in {db-path}/ucsc/gtf_builds.tsv. Sizes files use the assembly name whenever possible; .sizes.clean uses primary chromosomes from the assembly report. Use -name with install-genebed to store a selected GeneBED set under a custom species name; the mapping is recorded in custom.name.tsv.

Layout

{db-path}/genebed/{species}/{version}/all.gene.bed
{db-path}/genebed/{species}/{version}/deduplong.gene.bed
{db-path}/genebed/{species}/def -> {version}
{db-path}/sizes/{species}.sizes
{db-path}/sizes/{species}.sizes.clean

GeneBED generation from a local GTF

sjcab-peak2anno-db download-genebed mm10 vM22 \
  --gtf-path gencode.vM22.annotation.gtf.gz \
  --output-dir annotations

Use --gene-type repeatedly to restrict GENCODE types. See FeatureBEDs for derived regions and Deduplication for transcript selection.

GTF resolution URLs

The resolver checks these sources for the GTF associated with a GeneBED: