Plotting Guide

This notebook describes the types of plotting functions included with GSForge, as well as how to apply customizations to those plots.

Plot source data delineates the plotting functions.

  • gem the count array from an AnnotatedGEM or the same via a GeneSetCollection.

  • collections based on membership of GeneSet objects within a GeneSetCollection.

  • results Custom functions for specific analytical methods, e.g. volcano plots for DGE analysis.

GSForge uses the Holoviews package for creating plots. Holoviews is a common API to create plots using two popular backends, matplotlib and bokeh.

Plotting Guide Setup

A shared setup for all plotting guides.

# OS-independent path management.
from os import environ
from pathlib import Path
import numpy as np
import GSForge as gsf
import holoviews as hv
hv.extension('bokeh', 'matplotlib)

OSF_PATH = Path(environ.get("GSFORGE_DEMO_DATA", default="~/GSForge_demo_data/")).expanduser().joinpath("osfstorage", "oryza_sativa")
NORMED_GEM_PATH = OSF_PATH.joinpath("AnnotatedGEMs", "oryza_sativa_hydro_hisat2_normed.nc")
TOUR_BORUTA = OSF_PATH.joinpath("GeneSetCollections", "tour_boruta")

  File "/tmp/ipykernel_5876/1680162355.py", line 7
    hv.extension('bokeh', 'matplotlib)
                                      ^
SyntaxError: EOL while scanning string literal

Load an annotated expression matrix and a collection:

agem = gsf.AnnotatedGEM(NORMED_GEM_PATH)
agem

gsc = gsf.GeneSetCollection.from_folder(
    gem=agem, target_dir=TOUR_BORUTA, name="Boruta Results")
gsc

Select a backend, or load only the one you wish to use.

hv.output(backend="matplotlib")
gsf.plots.gem.GenewiseAggregateScatter(agem, datashade=False, apply_default_opts=False)

hv.output(backend="bokeh")
gsf.plots.gem.GenewiseAggregateScatter(agem, datashade=True)

Saving Plots

hv.save(my_plot.options(toolbar=None),  # This sometimes does not remove the toolbar.
        "my_filename.png")

Where to go for more help?

See the excellent Holoviews documentation for more on creating and styling plots.

How to return a plot with no styling applied

It is then up to you to modify the holoviews object, and to apply desired options from your selected backend.

All GSForge plotting functions have an apply_default_opts, which attempts to apply options based on the currently loaded backend extension. If you want to apply your own styling, plots can be returned with their default settings only by setting apply_default_opts=False.