This solution involves formatting the reads into a FASTA file, generating a genome index, and writing to disk BAM files and importing them into R. The other proposed answer solves the problem more directly and without reading and writing of results to and from the disk.

This solution does not work because the probes do not all have the same lengths and vmatchPDict results in an error when it tries to coerce the pdict variable into a PDict object. I need to create a PDict object with tb.start and tb.end settings because of this, but if I do, vmatchPDict gives an error, because it only accepts a DNAStringSet for the pdict parameter, not PDict objects.

e.g. (artificial)

> vmatchPDict(DNAStringSet(c("GATC", "TAG")), Hsapiens)
Error in .Call2("ACtree2_build", tb, pp_exclude, base_codes, nodebuf_ptr,  :
  element 2 in Trusted Band has a different length than first element
> vmatchPDict(PDict(DNAStringSet(c("GATC", "TAG")), tb.start=1, tb.end =3), Hsapiens)
Error in .local(pdict, subject, max.mismatch, min.mismatch, fixed, algorithm,  :
  'pdict' must be a DNAStringSet object

It's bizarre that the parameter is named pdict, but it results in an error if the user provides a PDict object.

Could an example also be added to the Examples section of the documentation of how to restrict the search to only chromosomes of interest ? There doesn't seem to be any way to only search chr1, ..., chr22, chrX, chrY, chrM for my human analysis scenario.

Actually, vmatchPDict is just a code stub in BSgenome version 1.40.1. Why is it even present in the release version of BSgenome ?

> vmatchPDict(DNAStringSet("GATC", Hsapiens))
Error in .local(pdict, subject, max.mismatch, min.mismatch, with.indels,  :
  vmatchPDict() is not ready yet, sorry

Code which doesn't work should only be in the development version (and be finished before the next Bioconductor release version).