In jmw86069/jamba: Jam Base Methods

knitr::opts_chunk$set(
  collapse=TRUE,
  warning=FALSE,
  message=FALSE,
  comment="#>",
  fig.path="man/figures/README-"
);
options(knitr.table.format='markdown')

jamba

The goal of jamba is to provide useful custom functions for R data analysis and visualization.

Package Reference

A full online function reference is available via the pkgdown documentation:

Full jamba command reference

Functions are categorized, some examples are listed below:

Background

The R functions in jamba have been built up, used, tested, revised over several years. They are immediately useful for day-to-day work, and efficient and robust enough for production pipelines.

Many were inspired by discussion from Stackoverflow, R-help, or Bioconductor, with citations thanking principal author(s). Many thanks to the original authors! The R community is built upon the collective greatness of its contributors!

Most of the functions are designed around workflows for Bioinformatics analyses, where functions need to be efficient when operating over 10,000 to 100,000 elements. (They work quite well with millions as well.) Usually the speed gains are obvious with about 100 elements, then scale linearly (or worse) as the number increases. I and others use these functions all the time.

One example function writeOpenxlsx() is a simple wrapper around very useful openxlsx::write.xlsx(), which also applies column formatting for column types: P-values, fold changes, log2 fold changes, numeric, and integer values. Columns use conditional Excel formatting to apply color-shading to cells for each type.

Similarly, readOpenxlsx() is a wrapper function to openxlsx::read.xlsx() which reads each worksheet and returns a list of data.frame objects. It can detect multi-row column headers, for which it returns combined column names. It also applies equivalent of check.names=FALSE so column names are returned without change.

Small and large efficiencies are used wherever possible. The mixedSort() functions are based upon gtools::mixedsort(), with additional optimizations for speed and custom needs. It sorts chromosome names, gene names, micro-RNA names, etc.

Example R functions

Efficient alphanumeric sort

• mixedSort() - highly efficient alphanumeric sort, for example chr1, chr2, chr3, chr10, etc.
• mixedSortDF() - as above, applied to columns in a data.frame (or matrix, tibble, DataFrame, etc.)
• mixedSorts() - as above, applied to a list of vectors with no speed loss.

Example:

x <- sort(c(
  "miR-12","miR-1","miR-122","miR-1b",
  "miR-1a","miR-2", "miR-22",
  "ABCA2", "ABCA12"));
df1 <- data.frame(
  miRNA=x,
  sort_rank=seq_along(x),
  mixedSort_rank=order(jamba::mixedOrder(x)),
  check.names=FALSE,
  stringsAsFactors=FALSE);
df2 <- jamba::mixedSortDF(df1);
df2;

Base R plotting

These functions help with base R plots, in all those little cases when the amazing ggplot2 package is not a smooth fit.

• nullPlot() - convenient "blank" base R plot, optionally displays margins
• plotSmoothScatter() - smooth scatter plot() for point density, enhanced over smoothScatter()

require(jamba);
set.seed(123);
x <- matrix(ncol=2, data=rnorm(40000*2));
x[,2] <- x[,1] + rnorm(40000)*0.15;
x[1:3000,] <- t(t(x[1:3000,,drop=FALSE])+c(0.6,-0.7));
x[1:2000,2] <- x[1:2000,1] + rnorm(1000)*0.5;
opar <- par("mfrow"=c(1,2), "mar"=c(2.5, 2, 3, 0.5));
smoothScatter(x, main="smoothScatter()");
plotSmoothScatter(x, main="plotSmoothScatter()");
par(opar);

• plotPolygonDensity() - fast density/histogram plot for vector or matrix

plotPolygonDensity(x[,1] - x[,2]);

• imageDefault() - enhanced image() that enables raster output with consistent pixel aspect ratio.
• imageByColors() - wrapper to image() for a matrix or data.frame of colors, with optional labels

set.seed(23);
opar <- par("mar"=c(1,1,1,1));
m <- matrix(sample(colors(), 9), ncol=3)
imageByColors(m, cellnote=m);
par(opar);

• minorLogTicksAxis() - log-transformed axis labels, flexible log base, and option for properly adjusted log2(1 + x) format.
• sqrtAxis() - draw a square-root transformed axis, with proper labels.
• drawLabels() - draw square colorized text labels
• shadowText() - replacement for text() that draws shadows or outlines.

opar <- par("mar"=c(1,1,1,1));
nullPlot(fill="navy", plotAreaTitle="", doMargins=FALSE);
text <- shadowText;
drawLabels(txt="shadowText() label",
  boxColor=alpha2col("palegoldenrod", 0.6),
  labelCol=setTextContrastColor(alpha2col("palegoldenrod", 0.6),
    bg="navy"),
  x=1.6, y=1.3, labelCex=2);
rm(text);
drawLabels(txt="text() label",
  boxColor=alpha2col("palegoldenrod", 0.6),
  labelCol=setTextContrastColor(alpha2col("palegoldenrod", 0.6),
    bg="navy"),
  x=1.3, y=1.7, labelCex=2);
par(opar);

• groupedAxis() - grouped axis labels to show regions/ranges
• decideMfrow() - determine appropriate value for par("mfrow") for multipanel output in base R plotting.
• getPlotAspect() - determine visible plot aspect ratio.

Excel export

Every Bioinformatician/statistician needs to write data to Excel, the writeOpenxlsx() function is consistent and makes it look pretty. You can save numerous worksheets in a single Excel file, without having to go back and custom-format everything.

• writeOpenxlsx() - flexible Excel exporter, with categorical and conditional colors.
• applyXlsxCategoricalFormat() - apply categorical colors to Excel
• applyXlsxConditionalFormat() - apply conditional colors to Excel

Color

Almost everything uses color somewhere, especially on R console, and in every R plot.

• getColorRamp() - flexible to create or retrieve color gradients
• warpRamp() - "bend" a color gradient to enhance the visual range
• color2gradient() - convert a color to gradient of n colors; or do the same for a vector
• makeColorDarker() - adjust darkness and saturation
• showColors() - display a vector or list of colors
• fixYellow() - adjust the weird green-yellow, by personal preference
• printDebug() - pretty colorized text output using crayon package.

opar=par("mar"=c(1,1,1,1));
rainbowv <- c("red","yellow","green","cyan","blue","magenta");
colorlist <- list(
  viridis_lens5=rep(each=2, getColorRamp("viridis", n=12, lens=5)),
  viridis=rep(each=2, getColorRamp("viridis", n=12)),
  `viridis_lens-5`=rep(each=2, getColorRamp("viridis", n=12, lens=-5)),
  RdBu_r_lens5=rep(each=2, getColorRamp("RdBu_r", n=11, lens=5)),
  RdBu_r=rep(each=2, getColorRamp("RdBu_r", n=11)),
  `RdBu_r_lens-5`=rep(each=2, getColorRamp("RdBu_r", n=11, lens=-5)),
  rainbow=rep(rainbowv, each=4),
  rainbow_gradient2=rep(each=2, color2gradient(rainbowv, n=2, gradientWtFactor=1/3)),
  rainbow_gradient4=color2gradient(rainbowv, n=4, gradientWtFactor=1/3)
)
showColors(colorlist, xaxt="n", labelCells=FALSE);
par(opar);

List

Cool methods to operate on super-long lists in one call, to avoid looping through the list either with for() loops, lapply() or map() functions.

• cPaste() - highly efficient paste() over a large list of vectors
• cPasteS() - as above but using mixedSort() before paste().
• cPasteU() - as above but using uniques() before paste().
• cPasteSU() - as above but using mixedSort() and uniques() before paste().
• uniques() - efficient unique() over a list of vectors
• sclass() - runs class() on a list
• sdim(), ssdim() - dimensions of list objects, or nested list of lists
• rbindList() - efficient do.call(rbind, ...) to bind rows into a matrix or data.frame, useful when following strsplit().
• mergeAllXY() - merge a list of data.frame objects
• rmNULL() - remove NULL or empty elements from a list, with optional replacement

Names

We use R names as an additional method to make sure everything is kept in the proper order. Many R functions return results using input names, so it helps to have a really solid naming strategy. For the R functions that remove names -- I highly recommend adding them back yourself!

• makeNames() - make unique names, using flexible logic
• nameVector() - add names to a vector, using its own value, or supplied names
• nameVectorN() - make named vector using the names of a vector (useful inside lapply()) or any function that returns data using names of the input vector.

data.frame/matrix/tibble

• pasteByRow() - fast, flexible row-paste with delimiters, optionally remove blanks
• pasteByRowOrdered() - as above but returns ordered factor, using existing factor orders from each column when present
• rowGroupMeans(), rowRmMadOutliers() - efficient grouped row functions
• mergeAllXY() - merge a list of data.frame into one
• renameColumn() - rename columns from and to.
• kable_coloring() - flexible colorized data.frame output in Rmarkdown.

String / grep

• gsubOrdered() - gsub that returns ordered factor, maintians the previous factor order
• grepls() - grep the environment (including attached packages) for object names
• vgrep(), vigrep() - value-grep shortcut
• unvgrep(), unvigrep() - un-grep -- remove matched results from the output.
• provigrep() - progressive grep, searches each pattern in order, returning results in that order
• igrepHas() - rapid case-insensitive grep presence/absense test
• ucfirst() - upper-case the first letter of each word.
• padString(), padInteger() - produce strings from numeric values with consistent leading zeros.

Numeric

• noiseFloor() - apply noise floor (and ceiling) with flexible replacement values
• warpAroundZero() - warp a numeric vector symmetrically around zero
• rowGroupMeans(), rowRmMadOutliers() - efficient grouped row functions
• deg2rad(), rad2deg() - convert degrees to radians
• rmNA() - remove NA values, with optional replacement
• rmInfinite() - remove infinite values, with optional replacement.
• formatInt() - convenient format() for integer output, with comma-delimiter by default

Practical / helpful

• jargs() - pretty function arguments, optional pattern search argument name

jargs(plotSmoothScatter)

• sdim(), ssdim() - dimensions of list objects, or nested list of lists
• sdima() - runs sdim() on the attributes of an object.
• isTRUEV(), isFALSEV() - vectorized test for TRUE or FALSE values, since isTRUE() only operates on single values, and does not allow NA.

R console

• printDebug() - pretty colorized text output using crayon package.
• setPrompt() - pretty colorized R console prompt with project name and R version
• newestFile() - most recently modified file from a vector of files

Other related Jam packages

• jamma -- MA-plots (also known as "mean-variance", "Bland-Altman", or "mean-difference" plots), relies upon jamba::plotSmoothScatter(); centerGeneData() to apply flexible row-centering with optional groups and control samples; jammanorm() - normalize data based upon MA-plot output
• colorjam -- colorjam::rainbowJam() for scalable categorical colors using alternating luminance and chroma values.
• genejam -- fast, consistent conversion of gene symbols to the most current gene nomenclature
• splicejam -- Sashimi plots for RNA-seq data
• multienrichjam -- multiple gene set enrichment analysis and visualization
• platjam -- platform technology functions, importers for NanoString

jmw86069/jamba documentation built on March 26, 2024, 5:26 a.m.