jb2hubs
Turns public genome resources into ready-to-open
JBrowse 2 sessions.
What it does
- Creates JBrowse configs from the UCSC main genome browser assemblies and
the ~50,000 UCSC GenArk hubs, one per
assembly at a permanent url, for use on https://genomes.jbrowse.org or in
JBrowse Desktop.
- Builds genomes.jbrowse.org, a static site for searching assemblies by
name, accession or taxonomy and launching them, plus a protein browser,
pangenome pages, and gene and ortholog pages.
Unique features
- Genetic codes. We read the
transl_table on each CDS in a GenArk
assembly's NCBI GFF, and for UCSC assemblies we ask NCBI taxonomy for the
species' mitochondrial code. The config records the code per sequence, so
JBrowse translates mitochondrial and plastid genes with the right one.
- Circular sequences. We list the sequences the NCBI GFF marks
Is_circular=true, so the circular view draws mitochondria, plasmids and
bacterial chromosomes as rings.
- More gene tracks on UCSC assemblies. We match each NCBI-derived UCSC
assembly to its RefSeq annotation and add that GFF as a track, plus GENCODE on
hg38, hg19 and mm39.
- Gene search on unannotated hubs. GCA hubs have no NCBI annotation, so we
join UCSC's xenoRefGene alignments to NCBI's RefSeq → symbol table and index
the symbols.
- UCSC tracks look as they do on UCSC. ENCODE composites become one overlaid
multi-wiggle track in UCSC's colors, itemRgb tracks get a color legend, and
RepeatMasker can split into one row per repeat class.
- Synteny. We convert UCSC's liftOver chains into JBrowse synteny tracks
between assemblies.
- Old JBrowse versions keep working. Desktop installs and published links
keep naming a config's url, so before each upload
run.sh boots a sample of
the regenerated configs in every JBrowse release we support.
Quick start
pnpm install
./run.sh --dry-run # build, no upload
./run.sh # build, upload, deploy
Docs
Credits
Thanks to the UCSC Genome Browser team for their generous data sharing policy.
Written with the aid of AI tools including Claude Code.
