plot_linear_scatter: Make a scatter plot between two groups with a linear model...

In elsayed-lab/hpgltools: A pile of (hopefully) useful R functions

Make a scatter plot between two groups with a linear model superimposed and some supporting statistics.

Description

Make a scatter plot between two groups with a linear model superimposed and some supporting statistics.

Usage

plot_linear_scatter(
  df,
  cormethod = "pearson",
  size = 2,
  loess = FALSE,
  xcol = NULL,
  ycol = NULL,
  text_col = NULL,
  logfc = 2,
  identity = FALSE,
  z = 1.5,
  z_lines = FALSE,
  first = NULL,
  second = NULL,
  base_url = NULL,
  pretty_colors = TRUE,
  xlab = NULL,
  ylab = NULL,
  color_high = NULL,
  color_low = NULL,
  alpha = 0.4,
  ...
)

Arguments

df	Dataframe likely containing two columns.
cormethod	What type of correlation to check?
size	Size of the dots on the plot.
loess	Add a loess estimation?
xcol	Column name of x-values
ycol	Column name of y-values#'
text_col	Column containing text annotations.
logfc	Point out genes with a specific logfc.
identity	Add the identity line?
z	Use this z-score cutoff.
z_lines	Include lines defining the z-score boundaries.
first	First column to plot.
second	Second column to plot.
base_url	Base url to add to the plot.
pretty_colors	Colors!
xlab	Alternate x-axis label.
ylab	Alternate x-axis label.
color_high	Chosen color for points significantly above the mean.
color_low	Chosen color for points significantly below the mean.
alpha	Choose an alpha channel to define how see-through the dots are.
...	Extra args likely used for choosing significant genes.

Value

List including a ggplot2 scatter plot and some histograms. This plot provides a "bird's eye" view of two data sets. This plot assumes a (potential) linear correlation between the data, so it calculates the correlation between them. It then calculates and plots a robust linear model of the data using an 'SMDM' estimator (which I don't remember how to describe, just that the document I was reading said it is good). The median/mad of each axis is calculated and plotted as well. The distance from the linear model is finally used to color the dots on the plot. Histograms of each axis are plotted separately and then together under a single cdf to allow tests of distribution similarity. This will make a fun clicky googleVis graph if requested.

See Also

[robust] [stats] [ggplot2] [robust::lmRob] [stats::weights] [plot_histogram()]

Examples

## Not run: 
 plot_linear_scatter(lotsofnumbers_intwo_columns)

## End(Not run)

elsayed-lab/hpgltools documentation built on May 9, 2024, 5:02 a.m.