This is a brief R script to generate histograms of genomic locations broken out by chromosome, starting with a BED-like file containing chromosomal locations and intensity values. It sets axis limits equal to the sizes of the chromosomes (detecting human or mouse samples automatically based on the number of chromosomes), and graphs the results using ggplot2 with facet wrapping. For human data, it plots a vertical line at the location of the centromere for each chromosome. No vertical lines are plotted for mouse data, because all the chromosomes are telocentric so the centromeres are at the end.
The intensity values could be peak height, fold change between samples, or any other measure of intensity at a given genomic location. In particular, a version of this script was used to generate Fig. S7 (Bush et al.).
This is a graph of H1 human embryonic stem cell peaks for the transcription factor Ctcf, using data downloaded from Encode.
These graphs are useful for detecting large-scale trends in genomic data, such as:
- sequencing/mapping errors that lead to large portions of a chromosome having no reads mapped to it
- Chip-seq peaks preferentially located near telomeres or centromeres
- gross chromosomal abnormalities leading to no reads mapping to a region
- overall changes in distribution of reads or peaks between two samples
You'll need R installed, along with the libraries ggplot2 and gtools.
drawHist.sh: One-line shell script to generate graphs of all files in the current directory with a "bed" extension, not using any score data, saving the graphs as PNG files with the same name appended with "-graph.png". Call with no arguments.
drawHist.r: Call it like this: Rscript drawHist.r filename startcol endcol scorecol, where "filename" is the name of a file of genomic locations (in a format such as BED or GFF). The three col numbers indicate which columns contain the relevant pieces of information. Enter 0 for endcol and/or scorecol to ignore those columns in the analysis.