Here is my solution. Please have the seqinr package installed and don't forget to chmod +x

#!/usr/bin/env Rscript

library(seqinr)

seqs <- read.fasta("a001.fasta")
locs <- read.delim("a001.coords", header=F, sep=" ") # Tabs? Put sep="\t"
outf <- file("a001-results.fasta", 'w')

doitall <- function(x) {
  seq_id <- x[[1]]; start <- x[[2]]; stop <- x[[3]]; seq <- seqs[[seq_id]]
  seqv   <- seq[start:stop]
  header <- paste(sep="", ">", attr(seq, "name"), "_", start, "_", stop, "\n")
  cat(file=outf, header)
  cat(file=outf, toupper(paste(sep="", collapse="", seqv)), "\n")
}
apply(locs, 1, doitall)
close(outf)

==> a001.coords <==

chr1 2 4
chr2 4 6
chr2 3 6

==> a001.fasta <==

>chr1
ATATATACCGA

>chr2
TGGCGTATTAT

==> a001-results.fasta <==

>chr1_2_4
TAT 
>chr2_4_6
CGT 
>chr2_3_6
GCGT

If you want to work with a plain text FASTA file, the fastaFromBed program in BEDTools will do this for you. The two inputs are your genome file in FASTA format and the coordinates in BED format. I would suggest using the unreleased version in the repository, as it uses Erik Garrison's nice FastaHack library for speed. If you already heave a .fai file from samtools, it will use that. Otherwise, it will create an index the first time, and reuse it thereafter.

fastaFromBed -fi in.fasta -bed regions.bed -fo out.regions.fa

You could try to use:

• fastacmd (NCBI): see http://www.ncbi.nlm.nih.gov/staff/tao/URLAPI/formatdb_fastacmd.html
• nibFrag or twoBitToFa: http://hgwdev.cse.ucsc.edu/~kent/src/unzipped/utils

See also: Fastest Way Of Extracting Millions Of Short Sequences From The Human Genome

Hi thanks a lot guys, I'm very new in the field, I really appreciate your help, bla bla bla

I have installed bedtool and tried fastafromBED but it looks like when I ask for positions 1 to 25, it gives me 2 to 25 instead in the output. How come?

For the other solution, I had to install R. I think there may a problem with the script because I use tab delimited file for coords and there are only three fields (BED format).

Cheers

Florianino

Ok, I'll post the BED issue right now as new question. Thanks Aleksandr, i'll try that and let you know. Thank you every one. Things are much easier when people are connected Cheers