In jefferys/fusionExpressionPlot: Plot Expression Profiles for Fused Genes

# Set output to use #=> at start of output lines for the rest of this document.
knitr::opts_chunk$set(collapse = TRUE, comment = "#=>")

FusionExpressionPlot takes advantage of the genomicRanges package and its GRange objects to manage genomic location data. These is only a moderate difference between a GRange object and a data frame, but GRange objects have lots of useful features. Only a few of these features are used in this package, mostly wrapped in "convienience" functions. For GRange objects in all their glory, see [TODO Link to info about GRange objects].

As used in this package, a GRange object represents the exon start and stop points of one genomic object (gene, transcript, fusion end, etc). We only use genomic ranges that reside on one strand of one chromosome. More complicated features (such as a list of genes or a full two-ended fusion) require multiple GRange objects to describe, often bound into a list.

`grNew()` - Create simple GRange objects

grNew() can be used to easily create simple GRange objects from a list of exons. In keeping with the role of GRange objects in this package, only single-strand, single chromosome objects can be created with grNew().

library(FusionExpressionPlot)
gr <- grNew(
   start= c( 101, 301, 511, 513, 813 ),
   end= c( 200, 500, 511, 612, 822 ),
   chr= "chr1",
   strand= "+"
)
gr

The GRange object returned can be queried in various ways using simple functions that have been extended to apply to GRange objects. These functions are part of the normal GenomicRanges package.

library(GenomicRanges)

start(gr)
end(gr)

as.character(strand(gr))
strand(gr)

as.character(seqnames(gr))
seqnames(gr)

width(gr)
width(gaps(gr))
gaps(gr)

length(gr)

`grAddColumn()` - Add meta-data columns to GRange objects

grAddColumn() can be used to easily add a column of data to a GRanges object. This will set one value per row in the GRanges object, wrapping if needed.

gr <- grAddColumn(gr, 'exonFillColor', c('red', 'orange', 'yellow', 'green', 'blue'));
gr

gr <- grAddColumn(gr, 'exonBorderColor', c('blue', 'green', 'black', 'orange', 'red'));
gr <- grAddColumn(gr, 'exonHeight', c(50, 100, 200, 100, 50));
gr <- grAddColumn(gr, 'gene', 'forTesting');
gr

After adding, a metadata column can be retrieved from the GRange object as a vector using the $ data frame notation (like GRangeObj$colName). To subset a genomic range object (returning another genomic range object), you can use the normal [] notation (like GRangeObj[,'colName']). This behavior is part of the normal GenomicRanges package.

gr$exonFillColor
gr$exonHeight
gr$gene

gr[, 'exonFillColor']
gr[1:3, c('exonFillColor', 'exonBorderColor')]
gr[1,]

`grToLocationString()` and `grFromLocationString()` - Location strings

One format for describing a sequence of gene locations is a string. To convert between a string and a GRanges object requires a specified structure and some delimiters. It must be possible to recognize separately the chromosome, the exon set, and the strand. The order these are specified in can be different.

Location strings are parsed by grFromLocationString. To read in a location string a regular expression with named capture groups is used. The default captureRE will parse a colon-delimited string that looks like chr1:23-2343,2343-234324:+, possibly including additional blanks.

demo.makeGrForTmprss2 <- function() {
   exons <- "42836479-42838080,42839661-42839813,42840323-42840465,42842575-42842670,42843733-42843908,42845252-42845423,42848504-42848547,42851099-42851209,42852403-42852529,42860321-42860440,42861434-42861520,42866283-42866505,42870046-42870116,42879877-42879992";
gr <- grNew(
      start=  as.numeric(sub("-.+", "", unlist(strsplit(exons, ",")))),
      end=    as.numeric(sub(".+-", "", unlist(strsplit(exons, ",")))),
      chr=    'chr21',
      strand= '-'
   );
   return( gr )
}

demo.makeGrForTmprss2()