[Bioc-devel] Random access to sequences in fasta files

[Thomas Lin Pedersen]

Hi

I’m querying on whether there are any plans on supporting random access reading of fasta files in the sense that it is possible to upfront specify the indexes of sequences that should be read in.

I’m working on a package for comparative microbial genomics and it would be a huge speed improvement if it was possible to quickly read in 1000’s of sequences distributed on as many files. Currently the proper, vectorised approach requires all files to be read in at once and then subsetted, but this can result in XStringSet’s in the Gb range, just to access some sequences. The slow, un-R way would be to loop through each file (or each sequence using skip and nrec to only read in relevant sequences). I’m preferentially looking for an interface like:

readXStringSet(files, rec)

Where rec is either a vector that would index into the XStringSet as if everything from files had been read in, or a list with the same length as files, containing the indexes of interest for each file.

with best wishes

Thomas