Custom browser tutorial

When your genome is one of the hosted hubs, JBrowseR(assembly = "hg38") is all you need. This tutorial covers the other case: building a browser for a genome you host yourself, with your own tracks, gene-name search, and theme.

library(JBrowseR)

Describe the assembly

An assembly is a list with a name and a uri. JBrowse derives the index locations (.fai, plus .gzi for bgzipped FASTA) from the URL, so you only point at the FASTA itself. Add reference-name aliases so chr1/1 both resolve.

hg19 <- list(
  name = "hg19",
  uri = "https://jbrowse.org/genomes/hg19/fasta/hg19.fa.gz",
  aliases = list("GRCh37"),
  refNameAliases = list(uri = "https://jbrowse.org/genomes/hg19/hg19_aliases.txt")
)

Add tracks

A track is a list too. uri alone is enough — the view infers the track type and adapter from the file extension, and derives the index location. You do not need to set assemblyNames on each track; the view fills it in from the assembly you load.

my_tracks <- list(
  list(
    uri =
    "https://jbrowse.org/genomes/hg19/GRCh37_latest_genomic.sort.gff.gz",
    name = "NCBI RefSeq Genes"
  )
)

Enable gene-name search

Hub assemblies include search; for a custom assembly, point at your own Trix index files with a Trix adapter and pass it in aggregateTextSearchAdapters. Now location can be a gene name.

hg19_search <- list(
  type = "TrixTextSearchAdapter",
  textSearchAdapterId = "hg19-index",
  assemblyNames = list("hg19"),
  ixFilePath = list(uri = "https://jbrowse.org/genomes/hg19/trix/hg19.ix"),
  ixxFilePath = list(uri = "https://jbrowse.org/genomes/hg19/trix/hg19.ixx"),
  metaFilePath = list(uri = "https://jbrowse.org/genomes/hg19/trix/meta.json")
)

Put it together

JBrowseR(
  assembly = hg19,
  tracks = my_tracks,
  aggregateTextSearchAdapters = list(hg19_search),
  configuration = list(theme = list(palette = list(primary = list(main = "#311b92")))),
  location = "MYC"
)

Show results computed in R

track_data_frame() turns a data frame into an in-browser track with no files and no server — the natural way to put an analysis you ran in R onto the genome. The frame needs chrom, start and end columns; an optional score column makes it a quantitative track.

regions <- data.frame(
  chrom = c("10", "10"),
  start = c(29838737, 29850000),
  end   = c(29840000, 29855000),
  name  = c("regionA", "regionB"),
  score = c(42, 88)
)

JBrowseR(
  assembly = hg19,
  tracks = list(track_data_frame(regions, "my_regions")),
  location = "10:29,838,737..29,855,000"
)

Reacting to clicks in Shiny

When rendered inside Shiny, clicking a feature sets input$<outputId>_selected_feature to the feature’s data, so you can build tables, plots, or links from the current selection. The id is namespaced per output, so several browsers on one page do not overwrite each other’s.

# server side
output$browser <- renderJBrowseR(
  JBrowseR(assembly = hg19, tracks = my_tracks, location = "MYC")
)
observeEvent(input$browser_selected_feature, {
  print(input$browser_selected_feature$name)
})

To move the browser from R without re-running the render expression, send it the changed options with update_jbrowse(). A reactive location fed from input$browser_location would loop; an observer does not.

observeEvent(input$goto_myc, {
  update_jbrowse("browser", location = "MYC")
})