How to convert genoma.fa to chr*.fa

[krafiq]

I am working with a genome file which is not on the USC genome browser. I have my own Genome.fa file and I want to separate the genome fasta file into chromosomes, so that there is one chr*.fa file for each chromosome.

Could someone please guide me as to how? I am just beginning to use Linux!

Thanks so much!

[SNPsaurus]

Hi krafiq,

Here is some Perl to do that. You just need to save it (as split.pl) to the folder you have your fasta file, then in the command line, write "perl split.pl genome.fa" where genome.fa is the name of your file. It will save a file for whatever is used as the header (I'm assuming it looks like ">chr1".

#!/usr/bin/perl

$f = $ARGV[0]; #get the file name

 open (INFILE, "<$f")
	or die "Can't open: $f $!";
		
while (<INFILE>) {
	$line = $_;
	chomp $line;
	if ($line =~ /\>/) { #if has fasta >
		close OUTFILE;
		$new_file = substr($line,1);
		$new_file .= ".fa";
		open (OUTFILE, ">$new_file")
			or die "Can't open: $new_file $!";
	}
	print OUTFILE "$line\n";
}
close OUTFILE;

[krafiq]

Dear SNPsaurus,

Thanks for your response! Howevr, my genome.fa file contains scaffold files and not chr files. So if I use your code above, it would create one file per scaffold. How do I get to one chr*.fa file for each chromosome from here?

[dpryan]

Do you mean to say that there are no ">chr..." lines in the Genome.fa file, only ">scaffold..." or something like that? If that's the case then there's no way to do it without assembling the scaffolds into chromosomes (which would take more sequencing and likely some wet-bench experiments).

If you mean to say that you have a mixture of ">chr..." and ">scaffold..." entries, then you can just modify the perl script above (just add a variable to store the context that you're in and only write output if you're in a ">chr..." context).

[krafiq]

Dear dpryan,

I'm trying to use the hotspot software and I'm trying to run their enumerateUniquelyMappableSpace script on my custom genome file. However, we don't have chromosome information for my genome. The genome.fa file only has >scaffolds. Is there a way to run hotspot without separating the genome fasta file into chromosomes/obtaining chr*.fa for every chromosome?

[EricHaugen]

Two options:

1. Change "chr" to "scaffold" in the enumerateUniquelyMappableSpace bash wrapper script, to list the individual fasta files.

2. Just run the whole genome fasta file, after building a bowtie index, with:

enumerateUniquelyMappableSpace.pl read_length bowtie_index_prefix genome.fa | sort-bed - | bedops -m - > genome.read_length.mappable_only.bed

If "sort-bed" runs out of memory here, the BEDOPS suite includes a "bbms" script that can be used in place of sort-bed.

[fengqi]

Did you try
'samtools faidx genome.fasta chrX > chrX.fasta'

[krafiq]

Thanks all!!

EricHaugen: I'm trying option 1 for now. I'm trying to run the script again with bowtie and bowtie-build in the same folder as the script. But it's giving me the following error:
./enumerateUniquelyMappableSpace: line 30: bowtie-build: command not found

And then it goes on to give the following error multiple times:
Failed to find bowtie index file Genome.1.ebwt

Does anyone know why and what I should change?

Thanks!

[EricHaugen]

It looks like "bowtie-build" isn't in your PATH, so the shell couldn't find it.

Try adding a line near the top of "enumerateUniquelyMappableSpace" like:

export PATH=$PATH:/location/of/this/script/folder

Then it should be able to find bowtie-build, and the Perl script it calls later will be able to find your bowtie executable there also.

[sivasubramani]

In HHblits package, there is a script which does the way you want.

HHblits_src/scripts/splitfasta.pl

[krafiq]

My enumerateUniquelyMappableSpace script is calling this line:
bowtie-build $chromosomeFiles $genome

The chromosome files variable in this case is a list of 30,000 file names. So when I run the script, it gives me the following error:

Argument list too long

is there a way around this?

[dpryan]

Try concatenating the files together first (you'll have to do it in a couple batches, since the command will be too long for "cat" too) and just use the multifasta file.

[krafiq]

dpryan: I had to split the genome.fa file to get the individual files in the first place to use the hotspot software. should i still cat them? won't that bring it back to genome.fa?

[dpryan]

Well, if you read through the perl scripts, it'll become pretty apparent that they only designed hotspot around human/mouse/etc. genomes (rather than your situation with scaffolds), so you're probably going to have to just edit the script. It's just trying to run bowtie-build, which will effectively concatentate everything together anyway (it looks like they normally use individual chromosome files so things can more easily be split to later run on a cluster). The script is pretty simple, so go ahead and change it to suite your usage needs.

[krafiq]

dpryan: I'm sorry-could you please clarify a bit as to what exactly I should do?
Also, is there a way to get the source code for bowtie?