R/gff.R

In SteeleCD/steeleLib: Utility functions to make life easier.

# ================================================================
# functions for reading gff files
# https://stat.ethz.ch/pipermail/bioconductor/2008-October/024669.html
# ================================================================

getAttributeField <- function (x, field, attrsep = ";") {
     s = strsplit(x, split = attrsep, fixed = TRUE)
     sapply(s, function(atts) {
         a = strsplit(atts, split = "=", fixed = TRUE)
         m = match(field, sapply(a, "[", 1))
         if (!is.na(m)) {
             rv = a[[m]][2]
         }
         else {
             rv = as.character(NA)
         }
         return(rv)
     })
}


gffRead <- function(gffFile, nrows = -1, skip = 0) {
     cat("Reading ", gffFile, ": ", sep="")
     gff = read.table(gffFile, sep="\t", as.is=TRUE, quote="",
     header=FALSE, comment.char="#", nrows = nrows, skip = skip,
     colClasses=c("character", "character", "character", "integer",  
"integer",
     "character", "character", "character", "character"))
     colnames(gff) = c("seqname", "source", "feature", "start", "end",
             "score", "strand", "frame", "attributes")
     cat("found", nrow(gff), "rows with classes:",
         paste(sapply(gff, class), collapse=", "), "\n")
     stopifnot(!any(is.na(gff$start)), !any(is.na(gff$end)))
     return(gff)
}