view assembly_stats_txt.xml @ 0:44c401ebc424 draft

Uploaded
author aaronpetkau
date Sat, 04 Jul 2015 08:57:35 -0400
parents
children
line wrap: on
line source

<tool id="assemblystats" name="assemblystats" version="1.0.2">
    <description>Summarise an assembly (e.g. N50 metrics)</description>
    <requirements>
		</requirements>
    <command interpreter="python">
	  assembly_stats_txt.py 
           '$input_type' '$stats.extra_files_path'
           '$input_type'
           '$bucket'
	   '$input'
   	    '$stats'
    '$sortedcontigs'
    '$histogrampng' 
    '$summedcontigspng'
    '$histogramdata' 
    '$summedcontigdata' 

	</command>
    <inputs>
        <param help="Is this from an genomic (contig) or transcriptomic assembly (isotig) or are these raw reads (read)" label="Type of read" name="input_type" type="select">
            <option selected="yes" value="contig">Contig (if from genomic assembly)</option>
            <option value="isotig">Isotig (if from transcriptomic assembly)</option>
            <option value="read">Raw reads from sequencer in FASTA format (useful for 454 data)</option>
        </param>
        <param falsevalue="" help="Use this to specify whether or not bin sizes of 1 should be used when plotting histograms" label="Output histogram with bin sizes=1" name="bucket" truevalue="-b" type="boolean" />
        <param format="fasta" label="Source file in FASTA format" name="input" type="data" />
        <param checked="false" help="If checked, all output files will be displayed. If not checked, only the file 'Assembly Statistics' will be provided." label="Return all output files" name="all_outputs" type="boolean" />
    </inputs>
    <outputs>
        <data format="tabular" label="Assembly statistics - $input.display_name" name="stats" />
        <data format="fasta" label="Sorted contigs - $input.display_name" name="sortedcontigs">
            <filter>all_outputs is True</filter>
        </data>
        <data format="png" label="Histogram of contig sizes - $input.display_name" name="histogrampng">
            <filter>all_outputs is True</filter>
        </data>
        <data format="png" label="Cumulative sum of contig sizes - $input.display_name" name="summedcontigspng">
            <filter>all_outputs is True</filter>
        </data>
        <data format="tabular" label="Histogram data - $input.display_name" name="histogramdata">
            <filter>all_outputs is True</filter>
        </data>
        <data format="tabular" label="Cumulative sum of contig size data - $input.display_name" name="summedcontigdata">
            <filter>all_outputs is True</filter>
        </data>
    </outputs>
    <help>
**Summarise assembly overview** 

This script is used to give summary statistics of an assembly or set of reads. Typically this is run after an assembly to evaluate gross features.


# Gives back
# - N50
# - num of contigs &gt; 1 kb
# - num of contigs
# - Read or Contig Histogram and graphs.
# - Summed contig length (by number of contigs, in sorted order)
	</help>
</tool>