Create a combined transdecoder + 6frame database

README.md

Creating a protein database from 6-frame and transdecoder sequences

Analyses in Galaxy

1. Run the TranscriptomePGMakeDatabase workflow. Input files for this include, a trinity assembly, predicted proteins from Transdecoder, gff3 coordinates corresponding to transdecoder predictions and the cRAP database of contaminants.
2. Ensure that the known_novel_crap_decoy.fasta output from the above workflow is loaded onto Mascot for searching.
3. Use the outputs from TranscriptomePGMakeDatabase to run the Transcriptome PG workflow. This workflow will be related to the Transcriptome PG workflow but should be modified to include a Mascot search for your specific organism.
4. Download the observed_peptides.gff3 file that you get from running the previous workflow step.

Use bedtools to find novel only peptides

1. First gather the following files in a directory on your laptop

   File	Description
   observed_peptides.gff	Output from the Galaxy workflow above
   transdecoder.gff3	GFF coordinates for transdecoder predictions (used as an input to the Galaxy workflow above)
   known.fasta	Output from database generation workflow above
   novel.fasta	Output from the database generation workflow above
   crap.fasta	The cRAP contaminants. Download

2. Make sure you install bedtools on your laptop

3. Extract only CDS entries from the transdecoder file

       cat transdecoder.gff3 | grep 'CDS' > transdecoder_cds.gff3

4. Compare observed peptides with transdecoder CDS and keep only those peptides that don't fully overlap with a transdecoder CDS prediction

       bedtools subtract -s -f 1.0 -a observed_peptides.gff3 -b transdecoder_cds.gff3 > really_novel.gff3

5. Downlod this filter_fasta.rb script and put it in the same directory as the files above. Make sure it has execute permissions (ie)

       chmod u+x filter_fasta.rb

6. Extract the id's from really_novel.gff and then grab their corresponding entries from the novel.fasta file

       cat really_novel.gff3 | awk -F '=' '{print $2}' | awk -F '.' '{print "lcl|"$1}' | sort -u > really_novel_ids.txt
   
       ./filter_fasta.rb novel.fasta really_novel_ids.txt > really_novel.fasta

7. Concatenate the novel proteins with known.fasta

       cat really_novel.fasta known.fasta > known_updated.fasta

Name and install the database

For MaxQuant simply rename the DB. Modify the example below for your species

    cp known_updated.fasta SAustrinumMQ_20151908.fasta

To create a database for normal TPP decoy searching and for loading on Mascot. For this you need to install protk which has the make_decoy.rb tool.

   cat crap.fasta known_updated.fasta > known_updated_crap.fasta
   make_decoy.rb -A known_updated_crap.fasta -o SAustrinum_20151908.fasta

filter_fasta.rb

#!/usr/bin/env ruby
#

require 'set'
require 'bio'

fasta_file=ARGV[0]

output_fh = $stdout

id_filter=ARGV[1]

$filter_ids = Set.new()
$filter_ids = File.readlines(id_filter).collect { |e| e.chomp }

def passes_filters(entry)

	if $filter_ids.length > 0
		# require 'byebug';byebug
		if $filter_ids.include? entry.entry_id
			return true
		end
		return false
	end

	return true

end


file = Bio::FastaFormat.open(fasta_file.chomp)
file.each do |entry|
	pass = passes_filters(entry)
	if pass
		output_fh.write entry
	end
end