efficiently read in fasta file and calculate nucleotide frequencies in R - r

Question

How can I read in a fasta file (~4 Gb) and calculate nucleotide frequencies in a window of 4 bps in length?
it takes too long to read in the fasta file using
library(ShortRead)
readFasta('myfile.fa')
I have tried to index it using (and there are many of them)
library(Rsamtools)
indexFa('myfile.fa')
fa = FaFile('myfile.fa')
however I do not know how to access the file in this format

Answer 1

I would guess that 'slow' to read in a file that size would be a minute; longer than that and something other than software is the problem. Maybe it's appropriate to ask where your file comes from, your operating system, and whether you have manipulated the files (e.g., trying to open them in a text editor) before processing.
If 'too slow' is because you are running out of memory, then reading in chunks might help. With Rsamtools
fa = "my.fasta"
## indexFa(fa) if the index does not already exist
idx = scanFaIndex(fa)
create chunks of index, e.g., into n=10 chunks
chunks = snow::splitIndices(length(idx), 10)
and then process the file
res = lapply(chunks, function(chunk, fa, idx) {
dna = scanFa(fa, idx[chunk])
## ...
}, fa, idx)
Use do.call(c, res) or similar to concatenate the final result, or perhaps use a for loop if you're accumulating a single value. Indexing the fasta file is via a call to the samtools library; using samtools on the command line is also an option, on non-Windows.
An alternative is to use Biostrings::fasta.index() to index the file, then chunk through with that
idx = fasta.index(fa, seqtype="DNA")
chunks = snow::splitIndices(nrow(fai), 10)
res = lapply(chunks, function(chunk) {
dna = readDNAStringSet(idx[chunk, ])
## ...
}, idx)
If each record consists of a single line of DNA sequence, then reading the records in to R, in (even-numbered) chunks via readLines() and processing from there is relatively easy
con = file(fa)
open(fa)
chunkSize = 10000000
while (TRUE) {
lines = readLines(fa, chunkSize)
if (length(lines) == 0)
break
dna = DNAStringSet(lines[c(FALSE, TRUE)])
## ...
}
close(fa)

Answer 2

Load the Biostrings Package and then use the readDNAStringSet() method
From example("readDNAStringSet"), slightly modified:
library(Biostrings)
# example("readDNAStringSet") #optional
filepath1 <- system.file("extdata", "someORF.fa", package="Biostrings")
head(fasta.seqlengths(filepath1, seqtype="DNA")) #
x1 <- readDNAStringSet(filepath1)
head(x1)