Skip to content

stack.md

Latest commit

 

History

History
59 lines (52 loc) · 2.9 KB

stack.md

File metadata and controls

59 lines (52 loc) · 2.9 KB
Raw

A view of several tracks

The code below reads e() from the quick start.

jbrowse_stack() stacks tracks as a JBrowse view does: each track's plot, or an alignments track's bands, top to bottom in the order given, on one x axis, with the keys collected.

  • jbrowse_stack() lays several loci end to end, as a view lays out several regions, and labels each tick in its own region's positions (two regions).
  • A transcript's exons, each with a few bases of flank, lay out side by side with the introns collapsed, as JBrowse's Collapse introns does, and each splice junction arcs from one exon straight to the next (GAPDH's exons).
  • Every track draws at one width: jbrowse_stack(..., width = 1200), or the widest width any track was read at. A feature or quantitative track drawn narrower than it was laid out for warns.
  • Each band is an ordinary ggplot, so patchwork's & restyles them all, as in jbrowse_stack(genes, signal, reads) & theme_classic(). A ggplot of your own stacks as a band on the same axis.
  • A track's name titles it, as the view labels a track: jbrowse_stack(Genes = genes, reads), or setNames(reads, contexts) for a list of them.
  • An alignments track read with display = list(coverage = list(show = FALSE)) stacks no coverage band.
  • A variant matrix gives each row the display's rows$step pixels, or without one fits its rows into the display's height, 200 pixels by default, as the display does.
  • A key several tracks share appears once, listing every entry any of them holds. A tumor with hard clips over a normal without them gets one "Insertions and clips" key, "Hard clip" included.
  • Files that spell a contig differently stack on one axis when each reader takes the same assembly: the readers then name the contig in every table as the region does.

Each band takes the pixels the display gives it. Reads draw at the display's pileup$mark$height (7, or 3 and 1 for its compact and super-compact settings), and jbrowse_height() gives the figure height that keeps those proportions, so a compact pileup saves as a shorter figure. jbrowse_width() gives the width at the same scale, so a track read at a wide view's width draws its labels at the display's size:

loci <- c("chr12:6,534,201-6,535,700", "chr12:6,537,201-6,538,700")
genes <- jbrowse_features(e("gapdh_gencode.gff3.gz"), loci)
reads <- jbrowse_alignments(e("gapdh_direct_rna.bam"), loci,
  display = list(pileup = list(mark = list(height = 3))))
fig <- jbrowse_stack(genes, reads)
ggsave("stack.png", fig, width = jbrowse_width(fig),
  height = jbrowse_height(fig))

From a JBrowse view

JBrowse's linear genome view has an Export R script menu item, from 5.0.0-beta.14. It writes one ggjbrowse reader per open track, then jbrowse_stack() and ggsave() at jbrowse_width(fig) by jbrowse_height(fig).