Trinity contig filtering

[ripeapple]

Hello, all,

I have just got my Trinity assembly, the N50 looks good. However, I have 100 contigs length extends from 4000bp to 11700bp.
Because we don't expect the contig size to be above 4000bp, so is this because of some genomic contaimination?
And can anyone suggest a program or script that can filter out the bad contigs? Thanks a lot,

[atcghelix]

I'm not sure what those contigs would be. Maybe chimeras? Have you tried blasting them?

If you're just looking to get rid of all sequences over 'x' length, then you can do something like below (if you have bioPerl installed). Usage: perl script.pl --in startingfile.fas --cutoff 4000 --out prunedfile.fas

#!/usr/bin/perl

use strict;
use warnings;
use Getopt::Long;
use Bio::SeqIO;

my $inFile;
my $cutoff;
my $outFile;

GetOptions  ("in=s"      => \$inFile,
             "cutoff=i"  => \$cutoff,
             "out=s"     => \$outFile) || die "Couldn't get parameters with Getopt::Long.\n";

my $seqIn = Bio::SeqIO->new(-file   => $inFile,
                            -format => 'fasta');
my $seqOut = Bio::SeqIO->new(-file   => ">$outFile",
                             -format => 'fasta');

while (my $seq = $seqIn->next_seq()) {
    if ($seq->length() < $cutoff) {
        $seqOut->write_seq($seq);
    }
}

[ripeapple]

Thanks, it helps

[Wallysb01]

Originally Posted by ripeapple

Hello, all,

I have just got my Trinity assembly, the N50 looks good. However, I have 100 contigs length extends from 4000bp to 11700bp.
Because we don't expect the contig size to be above 4000bp, so is this because of some genomic contaimination?
And can anyone suggest a program or script that can filter out the bad contigs? Thanks a lot,

Its possible its prespliced or incompletely spliced RNAs. However, why do you assume you shouldn't have >4000bp contigs? Ttn is >100,000bp. So certainly there are genes this big.

Also, have you tried Trinity's downstream analysis modules. They are very good at picking out protein coding orfs and blasting your dataset to identify orthologs

[ripeapple]

Yes, you're right,
I was thinking to remove the possible genomic contamination at first to set a cutoff for contig size.
Now I guess I need to map the sequence back to assembly see how it goes,
Thanks,