|
1
|
1 <tool id="sam_stats" name="sam-stats" version="0.1">
|
|
|
2 <description> - Compute statistics from SAM or BAM files</description>
|
|
|
3 <requirements>
|
|
|
4 <requirement type="package" version="1.1.2-484">ea-utils</requirement>
|
|
|
5 </requirements>
|
|
|
6 <command>
|
|
|
7 sam-stats
|
|
|
8 $trackMultAlign
|
|
|
9 $reportAllChr
|
|
|
10 #if $rnaSeqStats:
|
|
|
11 -R $rnaSeqStatsFile
|
|
|
12 #end if
|
|
|
13 #if $input.extension == "bam":
|
|
|
14 -B
|
|
|
15 #end if
|
|
|
16 -S $histBinSize
|
|
|
17 $input
|
|
|
18 > $samStats
|
|
|
19 </command>
|
|
|
20 <inputs>
|
|
|
21 <param format="sam, bam" name="input" type="data" label="SAM/BAM File" />
|
|
|
22 <param name="trackMultAlign" type="boolean" value="False" truevalue="-D" falsevalue="" label="Keep track of multiple alignments (slower!)" />
|
|
|
23 <param name="reportAllChr" type="boolean" value="False" truevalue="-A" falsevalue="" label="Report all chr sigs, even if there are more than 1000" />
|
|
|
24 <!-- <param name="numReadsSubsample" type="integer" value="1000000" min="1" max="1000000" label="Number of reads to sample for per-base statistics (max 1,000,000)" /> -->
|
|
|
25 <param name="histBinSize" type="integer" value="30" min="1" label="Number of bins per chromosome for reads by chromosome "histogram"" />
|
|
|
26 <param name="rnaSeqStats" type="boolean" value="False" label="Output RNA-Seq statistics (coverage, 3 prime bias, etc.)" />
|
|
|
27 </inputs>
|
|
|
28
|
|
|
29 <outputs>
|
|
|
30 <data format="tabular" name="samStats" label="${tool.name} on ${on_string}"/>
|
|
|
31 <data format="tabular" name="rnaSeqStatsFile" label="${tool.name} on ${on_string} (RNA-Seq Stats)"> <filter>rnaSeqStats</filter>
|
|
|
32 </data>
|
|
|
33 </outputs>
|
|
|
34
|
|
|
35 <stdio>
|
|
|
36 <exit_code range="1:" level="fatal" description="Unknown error occurred" />
|
|
|
37 </stdio>
|
|
|
38
|
|
|
39 <tests>
|
|
|
40 <test>
|
|
|
41 <param name="input" value="test.sam" />
|
|
|
42 <output name="samStats" file="testout.txt" />
|
|
|
43 </test>
|
|
|
44 </tests>
|
|
|
45
|
|
|
46 <help>
|
|
|
47 Overview
|
|
|
48 --------
|
|
|
49 sam-stats computes varius statics on SAM/BAM alignment files.
|
|
|
50
|
|
|
51 Complete Stats::
|
|
|
52
|
|
|
53 <STATS> : mean, max, stdev, median, Q1 (25 percentile), Q3
|
|
|
54 reads : # of entries in the sam file, might not be # reads
|
|
|
55 phred : phred scale used
|
|
|
56 bsize : # reads used for qual stats
|
|
|
57 mapped reads : number of aligned reads (unique probe id sequences)
|
|
|
58 mapped bases : total of the lengths of the aligned reads
|
|
|
59 forward : number of forward-aligned reads
|
|
|
60 reverse : number of reverse-aligned reads
|
|
|
61 snp rate : mismatched bases / total bases
|
|
|
62 ins rate : insert bases / total bases
|
|
|
63 del rate : deleted bases / total bases
|
|
|
64 pct mismatch : percent of reads that have mismatches
|
|
|
65 len <STATS> : read length stats, ignored if fixed-length
|
|
|
66 mapq <STATS> : stats for mapping qualities
|
|
|
67 insert <STATS> : stats for insert sizes
|
|
|
68 <CHR> : percentage of mapped bases per chr, followed by a signature
|
|
|
69
|
|
|
70 Subsampled stats (1M reads max)::
|
|
|
71
|
|
|
72 base qual <STATS> : stats for base qualities
|
|
|
73 A,T,C,G : base percentages
|
|
|
74
|
|
|
75 Meaning of the per-chromosome signature:
|
|
|
76
|
|
|
77 A ascii-histogram of mapped reads by chromosome position. It is only output if the original SAM/BAM has a header. The values are the log2 of the # of mapped reads at each position + ascii '0'.
|
|
|
78
|
|
|
79 See http://code.google.com/p/ea-utils/wiki/SamStatsDetails for more information on each stat, how it's calculated and what it means.
|
|
|
80
|
|
|
81 This tool uses the sam-stats program that is part of the ea-utils suite. See http://code.google.com/p/ea-utils/wiki/SamStats for details.
|
|
|
82 </help>
|
|
|
83 </tool>
|
