Help with FASTA parsing code.

[bigmac3000lbs]

Hi guys!

I'm working on a project where I am reading in values from a text file and then using them as search terms in a FASTA file. Ideally every time the program finds encounters the search term in the header section of the FASTA file, it should copy the header section of the FASTA file, along with the associated sequence into an output file. I've included examples of the kinds of data I am working with, along with the code I'm currently working on so you can get a better sense of what I am talking about.

Currently the problem I've been having with my code is that when I run it, it only outputs the results of the search conducted with the final term in the text file. It's that the program ran through the entire text file without performing any searches on the file to be searched, before stopping on the last item in the text file and performing a search. I should warn you that I just started programming in perl recently, so it is quite possible that my code has a few glaring errors that I've missed.

#Data from the text file I am using for search terms. Note these sequence ID names represent a subset of the sequences found in the FASTA file I am searching through. The actual text file is much larger, containing 1000s of terms.
BF01013B2E03.f1
BF01010B1A11.f1
BF01028B1E03.f1
BF01029A2C12.f1
BF01028B2D11.f1

#Data from input FASTA file. I am searching through this file using th +e items from the text file shown above. The actual input FASTA file contains 1000s more items
> BF01013B2E03.f1 735 1 735 ABI CAATCCAAGAACATTTTGAAGAAAAAATCTCTTAAAAAAAAGAAATCAAAACAAGTGATCAAAAATGAAATGAATGGTCA
> BF01010B1A11.f1 782 1 782 ABI AACGGACNANNCGGCAACCAGGAGGCCTTCCAAGCTGAACTGGGAGAGTGGATCAAGAAG
> BF01028B2D11.f1 674 1 674 ABI CCAGCACNNNNTNAGATATTAGCCTAGCCTCTATGTCGTATTTGTATTTCNNCTAGTTTTTCATCCGACTTTTTTGGATC

#output file note the output file has only one sequence in it. This is clearly not what I want. Something is wrong, I just can't put my finger on it. > BF01028B2D11.f1 674 1 674 ABI CCAGCACNNNNTNAGATATTAGCCTAGCCTCTATGTCGTATTTGTATTTCNNCTAGTTTTTCATCCGACTTTTTTGGATC #perl program
<code>
#!/usr/bin/perl
use strict;
use File::Basename;
my $database;
my $data = shift(@ARGV);
my $input_file = shift(@ARGV);
my $infile;
my $match; my $output_file;
my $ESTs_W_SNPs;
$output_file = "ESTsWsnpsANDgoodCoverage.fasta";
open ($database,'<', $data) or die "Cannot open $data\n";
while ($match = <$database>)
{
chomp $match;
open (IN, $input_file) || die "Can't open input file. Please provide a valid input filename.\n";
open ($ESTs_W_SNPs, '>', $output_file) or die "Cannot write to $output_file\n";
my ($seq, $prevhead) = (0, "", '');
while(<IN>)
{
my $line = $_;
$line =~ s/\r\n/\n/;
chomp $line;
$seq.= uc($line) if(eof(IN));
if (/\>(.*)/ || eof(IN))
{
my $head=$1;
printf $ESTs_W_SNPs ">$prevhead\n$seq\n" if($prevhead =~ /$ma +tch/); $prevhead = $head; $seq='';
}
else
{
$seq.=$line;
}
}
close (IN);
close ($ESTs_W_SNPs);
}

[kmcarr]

You are re-opening the output file every time you grab a new input from your database file (the outermost while loop) which basically reinitializes the file. You need to move the open/close statements for the output file outside your main program loop.

Also, you approach is not the most efficient. You are scanning through your entire FASTA file for each record you want to save. A better approach is to read in your search list and store it in a hash. Then go through your FASTA file just once, comparing each record to your list (hash). If the ID is in your hash then write the sequence to your output file. If the ID is not in your list then move on to the next record.

[chalex]

So, uh, wouldn't you just do

grep -F -f file1 -A 1 file2

where file1 is your search terms and file2 is the full FASTA file?

See the man page for GNU grep for more info.

[kmcarr]

So, uh, wouldn't you just do

grep -F -f file1 -A 1 file2

where file1 is your search terms and file2 is the full FASTA file?

See the man page for GNU grep for more info.

You're method only copies the first line of sequence after the definition line. There is no way to tell how many lines of sequence are follow a given definition line so you need some way to examine the lines following your matched definition.

[greigite]

You're method only copies the first line of sequence after the definition line. There is no way to tell how many lines of sequence are follow a given definition line so you need some way to examine the lines following your matched definition.

Funny, I was just working on this problem today. I quality filtered some Illumina PE data (reads 1 and 2 in separate files) and after that they didn't have exactly matched reads in them anymore. I needed to produce new files for each read containing only those reads with a partner in the other quality filtered file. Eventually I figured out that grep does work for this if you tell it to pull only the 4 lines after the read ID match:

     my $lineone = `grep -n $readID $read_file_name]`;
     my ($linenum) = $lineone =~/([0-9]{1,}):.*/;
     $linenum = $linenum +3;
     my @printone = `head -$linenum $read_file_name | tail -4`;

[tomc]

tools do exist

Is there a reason not to use the pattern of,
indexing the sequence fasta file with whatever blast like tool
you are use to, then feeding the set of identifiers to the command
that extracts the fasta records from the indexed sequence?
especially if you may want to partition out other subset of sequence.
something like ...

classic ncbi blast

formatdb  -p F -o T -i sequence.fasta -n blastdb
fastacmd  -d blastdb  -i accession.list  -o isolated.sequences

wublast

xdformat -n -I -o blastdb sequence.fasta  
xdget -n -f  blastdb  accession.list > isolated.sequences

new blast+ (have never used yet)

makeblastdb -in sequence.fasta -dbtype nucl  -hash_index  - parse_seqids  -out blastdb
blastdbcmd  -entry_batch accession.list -db blastdb -dbtype nucl

nor I have been using the newer tools, bowtie, bwa ...
long enough to to know if they have a similar mode
but it would not surprise me.

[Seqasaurus]

pyfasta http://pypi.python.org/pypi/pyfasta/

or flatten the fasta so that each sequence only occupies a single line then use grep?