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.

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.

pnpm install
./run.sh --dry-run   # build, no upload
./run.sh             # build, upload, deploy

Thanks to the UCSC Genome Browser team for their generous data sharing policy. Written with the aid of AI tools including Claude Code.