Eland-to-Bed algorithm

[apfejes]

That is some serious spam above - and worse, copied from my own blog!

Anyhow, can you paste the first line of the file? There's probably something simple that's going wrong, eg, you haven't followed the work flow to remove unmapped reads.

http://sourceforge.net/apps/mediawiki/vancouvershortr/index.php?title=WorkFlows

[__sequence]

Oups, I just edited my post above. I figured out about the directory, but there is still an error:
Error: Line 1 has an invalid read:
Error: Mismatches is less than 0

[apfejes]

You need to put the log file in a directory that exists and for which you have write permissions. If the directory you've given above does not exits, then it will not be able to create the log file.

[__sequence]

The latest versions can be found as part of the Vancouver Short Read Analysis Package: http://vancouvershortr.sourceforge.net/

There should be several work flows here, depending on the starting format.

http://sourceforge.net/apps/mediawiki/vancouvershortr/index.php?title=ConvertToBed

If that doesn't work for you, please let me know, and I'll provide more explicit information when I return to the office.

Hi apfejes,

Thank you for your reply. I tried the following format: java -jar conversion_util/ConvertToBed.jar -aligner eland -input "input_dir/name" -output "output_dir" -name "name" -noprepend

As a result I got the following error:

Version: Initializing class ElandIterator $Revision: 2933 $
Error: Line 1 has an invalid read:
Error: Mismatches is less than 0

[apfejes]

Sorry for the slow reply - I'm currently away at a conference.

The latest versions can be found as part of the Vancouver Short Read Analysis Package: http://vancouvershortr.sourceforge.net/

There should be several work flows here, depending on the starting format.

http://sourceforge.net/apps/mediawiki/vancouvershortr/index.php?title=ConvertToBed

If that doesn't work for you, please let me know, and I'll provide more explicit information when I return to the office.
Cheers

[__sequence]

mgogol, Thank you for posting. I have tried this script, and it returns an empty file. May be the problem is that it requires an additional file as input?

Requires tab delim file of chromosome or contig names
# (eland fa match files) in the format:
# UCSC_chr_name chr_length eland_name

I have all chromosomes in one input file. So I still need to create this additional file with chromosome names? Not clear how.

[mgogol]

perl script export2bed.pl

Here's a script from a colleague that I've used before.

#!/usr/bin/perl
# Program to convert eland export format to BED format
# Chris Seidel, June 2009
#
# Requires tab delim file of chromosome or contig names 
# (eland fa match files) in the format:
# UCSC_chr_name chr_length eland_name
# corrects for alignments that go off the ends of the chrs
# negative bases are trimmed to 1, 
# bases > chr_length are set to chr_length
# (I know the former exist, I don't know if the latter exist)
# results are not sorted, but can be sorted in linux by:
# sort -o infile.bed -k 1,1 -k 2,2n infile.bed
# (sort in place, first column, then by second column numeric)

die("usage: $0 chrmap.txt eland_export.txt") unless(scalar(@ARGV) == 2);

# create output filename
$outfile = $ARGV[1];
$outfile =~ s/\.txt$/\.bed/;
open(FOUT, ">$outfile") || die("can't open output file: $outfile");

# get info on chromosomes
open(cmap, $ARGV[0]) || die("no chromosome name mapping file!");
%chrmap = {};
while($line = <cmap>){
    chomp($line);
    ($newval, $size, $oldval) = split(/\t/, $line);
    $chrmap{$oldval} = $newval;
    $chrsize{$oldval} = $size;
}

# open input file
open(fp, $ARGV[1]) || die("can't open eland file");

$lines = 0;
while(<fp>){
    chop;
    ++$lines;
    @bits = split(/\t/);
    # skip reads that didn't pass filtering
    next if($bits[21] eq "N");
    # get match name
    $seqname = $bits[10];
    # skip No Matches or QC failures
    # next if($seqname =~ /NM|QC/);
    # skip repeat matches
    # next if($seqname =~ /\d+:\d+:\d+/);
    # we're only interested in sequences that match our chrs
    next unless(exists($chrmap{$seqname}));

    $seqlen = length($bits[8]);
    $start = $bits[12];
    $end = $start + $seqlen - 1;
    $strand = $bits[13];

    # parse match descriptor
    $n = ($bits[14] =~ tr/[ACGTN]/[ACGTN]/);
    # skip reads beyond a certain threshold
    next if($n > 2);
    $read_code = "U".$n;

    # correct for alignments off the chromosome ends
    if( $start <= 0 ){
        print STDERR "start less than or equal to 0:   ", $start, "\n";
        print STDERR join("\t", @bits), "\n";
        $start = 1;
    }

    if($end > $chrsize{$seqname}){
        print STDERR "end greater than chr end $chrsize{$seqname}:   $end, diff: ", $end - $chrsize{$seqname}, "\n";
        print STDERR join("\t", @bits), "\n";
        $end = $chrsize{$seqname};
    }

    if($strand eq "F"){
        $strand = "+";
        $color = "0,0,255";
    }
    else{
        $strand = "-";
        $color = "255,0,0";
    }

    $score = 0;
    print FOUT join("\t", $chrmap{$seqname}, $start, $end, $read_code, $score, $strand, $start, $end, $color), "\n";

    # give some feedback
    print STDERR "$lines processed\n" if(!($lines % 100000));
}

close(FOUT);
print STDERR "output file: $outfile\n";

[__sequence]

apfejes, could you please tell, what is the last downloadable version of your program, and what is an example of a command line to use it? I need to convert a single file (whole genome, not divided into chromosomes) as provided by the Eland export, convert it to .Bed

[apfejes]

Hey - Java's not THAT bad. I like that I can get 600%+ CPU usage with it, without any explicit multi-threading.

Anyhow, I have much better translators, now, in the Vancouver Short Read Analysis Package.... but they're still java. :P

The manuals are a work in progress. If anyone would like to give them a try, I'll update that part of the manual.

[ECO]

java...*shudders*

[jlli]

ElandtoBed.jar at FindPeak package (http://www.bcgsc.ca/platform/bioinfo/software/findpeaks)

[ECO]

If you don't get any help...post a few lines of the eland file and i can try to slap something together.

[joseph]

Can anybody share an algorithm that takes eland files and make bed files out of them?
Thanks
Joseph