Tutorials

Hands-on walkthroughs of common JBrowse workflows, from first setup to advanced analysis. The first section runs on genomes.jbrowse.org , which hosts a configured JBrowse for every UCSC genome, so those pages need nothing installed and no data prepared.

genomes.jbrowse.org

Basic usage of genomes.jbrowse.org Open a hosted genome, search a gene, and work through the UCSC track catalog on hg38, from conservation at TP53 to filtering a variant catalog down to something readable

Synteny on genomes.jbrowse.org Open a UCSC liftOver track in a linear genome view and launch a synteny view from it

Proteins on genomes.jbrowse.org Take any gene to its AlphaFold structure and its cross-species protein MSA, launched from the genome view and linked back to it

RepeatMasker as one lane per class Split a hub's RepeatMasker track into a labelled lane per repeat class, without preparing any data

The HPRC pangenome on genomes.jbrowse.org A click-path through the hosted HPRC page on staging.genomes.jbrowse.org, whose locus table opens the release 2 graph, its callset and its haplotypes at twenty structurally variable loci, with nothing prepared beforehand

Getting started

JBrowse web quick start Set up a self-hosted JBrowse Web instance and add an assembly and tracks

JBrowse desktop quick start Install JBrowse Desktop, open a genome, and add your first tracks

Synteny & comparative genomics

Synteny visualization (pairwise minimap2) Align two assemblies with minimap2 and compare them in the dotplot and linear synteny views

Synteny from an ortholog table (grape, peach, cacao) Stack N genomes from a jcvi MCScan .blocks file

Synteny visualization (all-vs-all minimap2) Stack strains in a linear synteny view from one all-vs-all PAF

Rearrangements between genomes by type (SyRI) Load SyRI's classification of the differences between assembled genomes, so each syntenic, inverted, translocated or duplicated region is a ribbon colored by its type, in a stack of four Arabidopsis accessions

Comparing one genome's two haplotypes (T2T-HG002) Load T2T-HG002 v1.2 and the Q100 project's own maternal-to-paternal chain, plot one haplotype against the other genome-wide, and look at the 8p23.1 inversion in a linear synteny view

Synteny between neighbouring haplotypes (the amylase locus) Stack one assembled human haplotype of each common amylase structure, from one AMY1 copy to seven, each aligned to the haplotype under it, so every band of the stack draws the insertions and deletions of a real pairwise alignment

Driving JBrowse with an AI agent (two Drosophila genomes) Ask an AI agent, in plain words, to align two species that have no published alignment, build the views, and find where the two genomes run in opposite directions

Selection pressure between two genomes (dN/dS) Colour an ortholog track by the ratio of non-synonymous to synonymous substitution, and read selection pressure off a gene neighbourhood

Synteny by ancestral linkage group (sponge, comb jelly, jellyfish) Load the ortholog tables odp colors by ancestral linkage group, sort each genome's chromosomes against the other's, and follow the groups across six animal genomes

Synteny from a pangenome graph (eight HPRC haplotypes) Stack eight HPRC haplotypes under GRCh38, whole genome, with the pairwise alignments unpacked from the Minigraph-Cactus graph's own projection and each lane drawing that haplotype's CAT gene models

Synteny from gene symbols (44 E. coli genomes) Stack forty-four E. coli and Shigella genomes under K-12 at one operon by joining their RefSeq annotations on the gene symbol, with no alignment step

Synteny from gene symbols (eight primates) Stack the great apes, a gibbon and a macaque under a human locus by joining their RefSeq annotations on the gene symbol

Synteny from liftOver chains (hg38 and eight vertebrates) Stack eight UCSC genomes under a human locus from the liftOver chains UCSC already publishes, one indexed alignment per genome composed into one track

Synteny from MCScan anchors (grape, peach) Load a pairwise jcvi MCScan run as gene-level and block-level synteny tracks, and convert an MCScanX run into the same files

Synteny on a circle (human and mouse) Put two genomes on one circular view, draw a liftOver chain's blocks as ribbons between them, add a gene density ring per genome, and check a ribbon and a ring value against the files they came from

Synteny visualization (a polyploid against itself) Draw hexaploid oat against itself from syntenic anchors, coloured by the selection pressure between each pair of copies

Synteny visualization (OrthoFinder orthogroups) Stack genomes on OrthoFinder orthogroups, built from protein homology, as the synteny table

Pangenomes

Pangenome (pggb) Build a five-strain pggb pangenome graph and load its linear projections plus the graph itself in JBrowse

Pangenome (Minigraph-Cactus) Build a Minigraph-Cactus pangenome graph and load its linear projections in JBrowse

Pangenome (HPRC) Read the HPRC release 2 graph at a locus, find every haplotype carrying an allele, put each haplotype back in its own coordinates, and draw the graph itself.

1. Reading the graph 2. Who carries what 3. Haplotype coordinates 4. The graph as the picture

Pangenome (HPRC) part 6, mitochondrial lineages from the graph Read a 9 bp deletion off the pangenome graph of the human mitochondrial chromosome, find who carries it, and cluster the graph's 234 haplotypes into the lineages of the human mitochondrial tree

Pangenome (mouse) Open the mouse strain pangenome, read Nnt against a reference that is itself one of the strains, then rank the graph for its densest bubble and descend into it

Pangenome (cattle) Open the bovine super-pangenome, see where the graph shows sequence it cannot attribute, and deconstruct it into the callset that names which assembly carries what

JBrowse 2 Pangenome (preparing your own graph) Turn a pangenome graph you already have into the small indexed files JBrowse opens by locus, and the database that holds every haplotype's walk

Pangenome (HPRC) part 5, repeat lengths across haplotypes Measure the ABCA7 intronic VNTR in every HPRC haplotype straight from the graph, set TRGT's read-based genotypes on the same bars, and find the samples where reads lost an allele the assemblies kept

Structural variation

Structural variants (1000 Genomes) Read one whole-gene deletion across the 1000 Genomes cohort, then check the genotypes against the reads that produced them

CNV across a population (1000 Genomes) Read k-mer depth copy number for every individual in the 1000 Genomes panel, and pack the whole panel into one Zarr store

Low-mappability regions (SMN) Check whether a locus can support the calls made on it, using the mappability, coverage and problematic-region tracks genomes.jbrowse.org already publishes

Structural variants (Dog10K) Genotype four classes of structural variant across dog breeds and read each against the gene it sits in

Structural variants from Hi-C Find a translocation in ENCODE Hi-C by opening two chromosomes in one view, and read the domain, loop and compartment calls over the same matrices

Cancer genomics

Structural variants (Cancer GIAB) Build a tumor/normal HiFi site from raw reads, and read its benchmark SV and CNV calls against the alignments that support them

CNV cohort (TCGA) Plot somatic copy number across a thousand tumors, one row per sample

Complex rearrangements and derivative alleles Follow a rearrangement that takes several junctions to make across a breakpoint split view, and show the derivative allele an assembly built against the reference

Gene fusion calls and the DNA behind them Triage a fusion caller's short-read calls against long RNA reads from the same cells, then find the DNA breaks the transcript junctions do not sit on

Mutation cohort (TCGA) Read somatic point mutations across a thousand tumors as a genotype matrix, grouped by clinical annotation

Reviewing a whole SV callset Render every junction in a somatic SV callset as a breakpoint image, triage the directory, and check the calls against the matched normal

Population genomics

Selection scans (Drosophila DGRP) Fst, diversity, and Tajima's D scans plus per-sample inversion genotypes from one VCF

Phased trio analysis (1000 Genomes) Paint a child's inherited haplotype blocks from hap-ibd IBD segments, one row per parental copy, and read the crossovers off the track

LD at a selective sweep (human) Precompute an LD triangle with PLINK, and cut a window that shows the block's edges

LD across an inversion (mosquitoes) Read precomputed PLINK LD over a 22 Mb inversion, and load the same inversion genotyped per mosquito

QTL mapping (BXD mice) Chromosome-painting and a QTL Manhattan plot from GeneNetwork BXD data

A loss-of-function allele across breeds (Dog10K) Locate a nonsense variant from the reference sequence, then read its genotypes across dog breeds and wolves

A selected haplotype (Dog10K) Scan the Dog10K panel for allele-frequency differences between breeds at both ends of a trait, then slice one peak out of the SNV callset and cluster its genotype matrix

Local ancestry (Dog10K) Paint wolf-derived haplotype blocks in two wolfdog breeds, against 219 other breeds and eight held-out wolves, from the Dog10K phased panel

Epigenomics & single cell

Methylation (long-read) Per-read, aggregate, and allele-specific 5mC from HG002 nanopore reads

Methylation (bisulfite) A WGBS/EM-seq pipeline from SRA reads to per-read CpG/CHG/CHH methylation coloring

ChromHMM chromatin states Paint many-cell-type ChromHMM states in one multi-row track

Single-cell ATAC pseudobulk Aggregate single-cell ATAC into per-cell-type coverage BigWigs

AlphaGenome predictions Predict expression, accessibility, splicing and 3D contacts over a locus, then score a variant against the reference

Single-cell RNA pseudobulk Aggregate single-cell RNA into per-cell-type coverage BigWigs, and link them to a UMAP

Transcriptomics & proteins

RNA-seq visualization Spliced reads, splice arcs, and strand-specific coverage in RNA-seq

Differential transcript usage Build a GFF3 carrying a per-transcript statistic in its attribute column, and configure a gene track to paint it

TP53 from prediction to crystal Open an AlphaFold model, the DNA-bound core domain and the MDM2 complex of p53 in one view, all linked to the same transcript, and read a hotspot back to its codon

Genes & annotation

Reviewing a gene prediction (Tiberius) Sort the models a gene finder disagrees with an existing annotation about into four kinds, and build a static review page that links each one back into JBrowse

Grammar of graphics

A grammar of graphics over a BED (RepeatMasker Alu age) Declare a BED column as the height of a bar and another as its colour, count features per bin zoomed out, and read a density sidecar past the fetch budget

A grammar of graphics over a BAM (NA12878 insert size) Declare a read's depth, insert size and mapping quality as the channels of a plot, with no variant caller in between, and scan a chromosome for a deletion

Configuration & embedding

Cookbook Copy-paste recipes for the most common JBrowse 2 configuration tasks: colors, labels, tooltips, tracks, themes, and more

Display settings Change a track's height, color and read pairing, then make the change stick in a link, a session file, or config.json

Embedding JBrowse Embed the linear genome view component in a custom web page

JBrowse 2 Embedded component examples ↗ Live, copy-paste examples for the embeddable React components: themes, per-feature colors, sessions, text search, and more.

$ jbrowse JBrowse CLI with Desktop Assemble a config.json with the jbrowse CLI and open it in JBrowse Desktop

Gene density and transposon density along a chromosome Whole-chromosome bands for 30,000 genes and 1.3 million Alu copies, from a features-per-kilobase sidecar the track draws where its features are too many to fetch