Let’s look at my gff track headers again. Why not do it with plyr instead?

d_ply splits the data frame by the feature column and applies a nameless function that writes subsets to the file (and returns nothing, hence the ”_” in the name). This isn’t shorter or necessarily better, but it appeals to me.

library(plyr)
connection <- file("separate_tracks_2.gff", "w")
d_ply(gff, "feature", function(x) {
  writeLines(paste("track name=", x$feature[1], sep=""), connection)
  write.table(x, sep="\t", row.names=F, col.names=F,
              quote=F, file=connection)
})
close(connection)