|
0
|
1 <tool id="assemblystats" name="assemblystats" version="1.0.2">
|
|
|
2 <description>Summarise an assembly (e.g. N50 metrics)</description>
|
|
|
3 <requirements>
|
|
|
4 </requirements>
|
|
|
5 <command interpreter="python">
|
|
|
6 assembly_stats_txt.py
|
|
|
7 '$input_type' '$stats.extra_files_path'
|
|
|
8 '$input_type'
|
|
|
9 '$bucket'
|
|
|
10 '$input'
|
|
|
11 '$stats'
|
|
|
12 '$sortedcontigs'
|
|
|
13 '$histogrampng'
|
|
|
14 '$summedcontigspng'
|
|
|
15 '$histogramdata'
|
|
|
16 '$summedcontigdata'
|
|
|
17
|
|
|
18 </command>
|
|
|
19 <inputs>
|
|
|
20 <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">
|
|
|
21 <option selected="yes" value="contig">Contig (if from genomic assembly)</option>
|
|
|
22 <option value="isotig">Isotig (if from transcriptomic assembly)</option>
|
|
|
23 <option value="read">Raw reads from sequencer in FASTA format (useful for 454 data)</option>
|
|
|
24 </param>
|
|
|
25 <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" />
|
|
|
26 <param format="fasta" label="Source file in FASTA format" name="input" type="data" />
|
|
|
27 <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" />
|
|
|
28 </inputs>
|
|
|
29 <outputs>
|
|
|
30 <data format="tabular" label="Assembly statistics - $input.display_name" name="stats" />
|
|
|
31 <data format="fasta" label="Sorted contigs - $input.display_name" name="sortedcontigs">
|
|
|
32 <filter>all_outputs is True</filter>
|
|
|
33 </data>
|
|
|
34 <data format="png" label="Histogram of contig sizes - $input.display_name" name="histogrampng">
|
|
|
35 <filter>all_outputs is True</filter>
|
|
|
36 </data>
|
|
|
37 <data format="png" label="Cumulative sum of contig sizes - $input.display_name" name="summedcontigspng">
|
|
|
38 <filter>all_outputs is True</filter>
|
|
|
39 </data>
|
|
|
40 <data format="tabular" label="Histogram data - $input.display_name" name="histogramdata">
|
|
|
41 <filter>all_outputs is True</filter>
|
|
|
42 </data>
|
|
|
43 <data format="tabular" label="Cumulative sum of contig size data - $input.display_name" name="summedcontigdata">
|
|
|
44 <filter>all_outputs is True</filter>
|
|
|
45 </data>
|
|
|
46 </outputs>
|
|
|
47 <help>
|
|
|
48 **Summarise assembly overview**
|
|
|
49
|
|
|
50 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.
|
|
|
51
|
|
|
52
|
|
|
53 # Gives back
|
|
|
54 # - N50
|
|
|
55 # - num of contigs > 1 kb
|
|
|
56 # - num of contigs
|
|
|
57 # - Read or Contig Histogram and graphs.
|
|
|
58 # - Summed contig length (by number of contigs, in sorted order)
|
|
|
59 </help>
|
|
|
60 </tool>
|
