Mercurial > repos > greg > vsnp_statistics

comparison vsnp_statistics.xml @ 8:1becb6606626 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author greg
date Mon, 02 Aug 2021 13:34:09 +0000
parents de2af65c4633
children ce1f889b3340
comparison
equal deleted inserted replaced
7:de2af65c4633 8:1becb6606626
2 <description></description> 2 <description></description>
3 <macros> 3 <macros>
4 <import>macros.xml</import> 4 <import>macros.xml</import>
5 </macros> 5 </macros>
6 <requirements> 6 <requirements>
7 <requirement type="package" version="1.79">biopython</requirement> 7 <expand macro="biopython_requirement"/>
8 <requirement type="package" version="1.21.1">numpy</requirement> 8 <expand macro="numpy_requirement"/>
9 <requirement type="package" version="3.0.7">openpyxl</requirement> 9 <expand macro="openpyxl_requirement"/>
10 <requirement type="package" version="1.3.0">pandas</requirement> 10 <expand macro="pandas_requirement"/>
11 <requirement type="package" version="2.0.1">xlrd</requirement> 11 <expand macro="xlrd_requirement"/>
12 </requirements> 12 </requirements>
13 <command detect_errors="exit_code"><![CDATA[ 13 <command detect_errors="exit_code"><![CDATA[
14 #import re 14 #import re
15 #set input_idxstats_dir = 'input_idxstats'
16 #set input_metrics_dir = 'input_metrics'
17 #set input_reads_dir = 'input_reads'
18 mkdir -p $input_idxstats_dir &&
19 mkdir -p $input_metrics_dir &&
20 mkdir -p $input_reads_dir &&
21 15
22 #if $input_type_cond.input_type == 'single_files': 16 #if $input_type_cond.input_type in ["single", "pair"]:
23 #set read1 = $input_type_cond.read_type_cond.read1 17 #set read1 = $input_type_cond.read1
24 #set read1_identifier = re.sub('[^\s\w\-]', '_', str($read1.element_identifier)) 18 #set read1_identifier = re.sub('[^\s\w\-]', '_', str($read1.element_identifier))
25 ln -s '${read1}' '${read1_identifier}' && 19 ln -s '${read1}' '${read1_identifier}' &&
26 #if $input_type_cond.read_type_cond.read_type == 'pair': 20 #if $input_type_cond.input_type == "pair":
27 #set read2 = $input_type_cond.read_type_cond.read2 21 #set read2 = $input_type_cond.read2
28 #set read2_identifier = re.sub('[^\s\w\-]', '_', str($read2.element_identifier)) 22 #set read2_identifier = re.sub('[^\s\w\-]', '_', str($read2.element_identifier))
29 ln -s '${read2}' '${read2_identifier}' && 23 ln -s '${read2}' '${read2_identifier}' &&
30 #else: 24 #else:
31 #set read2 = None 25 #set read2 = None
32 #end if 26 #end if
33 #else: 27 #else:
34 #if $input_type_cond.collection_type_cond.collection_type == 'single': 28 #set read1 = $input_type_cond.reads_collection['forward']
35 #for $i in $input_type_cond.collection_type_cond.reads_collection: 29 #set read1_identifier = re.sub('[^\s\w\-]', '_', str($read1.name))
36 #set identifier = re.sub('[^\s\w\-]', '_', str($i.element_identifier)) 30 ln -s '${read1}' '${read1_identifier}' &&
37 ln -s '${i.file_name}' '$input_reads_dir/${identifier}' && 31 #set read2 = $input_type_cond.reads_collection['reverse']
38 #end for 32 #set read2_identifier = re.sub('[^\s\w\-]', '_', str($read2.name))
39 #else: 33 ln -s '${read2}' '${read2_identifier}' &&
40 #set read1 = $input_type_cond.collection_type_cond.reads_collection['forward']
41 #set read1_identifier = re.sub('[^\s\w\-]', '_', str($read1.name))
42 ln -s '${read1}' '$input_reads_dir/${read1_identifier}' &&
43 #set read2 = $input_type_cond.collection_type_cond.reads_collection['reverse']
44 #set read2_identifier = re.sub('[^\s\w\-]', '_', str($read2.name))
45 ln -s '${read2}' '$input_reads_dir/${read2_identifier}' &&
46 #end if
47 #for $i in $input_type_cond.samtools_idxstats:
48 #set identifier = re.sub('[^\s\w\-]', '_', str($i.element_identifier))
49 ln -s '${i.file_name}' '$input_idxstats_dir/${identifier}' &&
50 #end for
51 #for $i in $input_type_cond.vsnp_azc:
52 #set identifier = re.sub('[^\s\w\-]', '_', str($i.element_identifier))
53 ln -s '${i.file_name}' '$input_metrics_dir/${identifier}' &&
54 #end for
55 #end if 34 #end if
56 35
57 python '$__tool_directory__/vsnp_statistics.py' 36 python '$__tool_directory__/vsnp_statistics.py'
58 #if $input_type_cond.input_type == 'single_files': 37 --read1 '${read1_identifier}'
59 --dbkey '$input_type_cond.samtools_idxstats.metadata.dbkey' 38 #if $read2 is not None
60 #if $input_type_cond.read_type_cond.read1.is_of_type('fastqsanger.gz'): 39 --read2 '${read2_identifier}'
61 --gzipped
62 #end if
63 --read1 '${read1_identifier}'
64 #if $input_type_cond.read_type_cond.read_type == 'pair':
65 --read2 '${read2_identifier}'
66 #end if
67 --samtools_idxstats '$input_type_cond.samtools_idxstats'
68 --vsnp_azc '$input_type_cond.vsnp_azc'
69 #else:
70 --dbkey '$input_type_cond.samtools_idxstats[0].metadata.dbkey'
71 #if $input_type_cond.collection_type_cond.reads_collection[0].is_of_type('fastqsanger.gz'):
72 --gzipped
73 #end if
74 #if $input_type_cond.collection_type_cond.collection_type == 'paired':
75 --list_paired
76 #end if
77 --input_idxstats_dir '$input_idxstats_dir'
78 --input_metrics_dir '$input_metrics_dir'
79 --input_reads_dir '$input_reads_dir'
80 #end if 40 #end if
41 #if $read1.is_of_type('fastqsanger.gz'):
42 --gzipped
43 #end if
44 --dbkey '$samtools_idxstats.metadata.dbkey'
45 --samtools_idxstats '$samtools_idxstats'
46 --vsnp_azc_metrics '$vsnp_azc_metrics'
81 --output '$output' 47 --output '$output'
82 ]]></command> 48 ]]></command>
83 <inputs> 49 <inputs>
84 <conditional name="input_type_cond"> 50 <conditional name="input_type_cond">
85 <param name="input_type" type="select" label="Choose the category of the files to be analyzed"> 51 <param name="input_type" type="select" label="Choose the category of the files to be analyzed">
86 <option value="single_files" selected="true">Single files</option> 52 <option value="single" selected="true">Single files</option>
87 <option value="collections">Collections of files</option> 53 <option value="paired">Paired reads</option>
54 <option value="pair">Paired reads in separate data sets</option>
88 </param> 55 </param>
89 <when value="single_files"> 56 <when value="single">
90 <conditional name="read_type_cond"> 57 <param name="read1" type="data" format="fastqsanger.gz,fastqsanger" label="Read1 fastq file"/>
91 <param name="read_type" type="select" label="Choose the read type">
92 <option value="single" selected="true">Single reads</option>
93 <option value="pair">Paired reads</option>
94 </param>
95 <when value="single">
96 <param name="read1" type="data" format="fastqsanger.gz,fastqsanger" label="Read1 fastq file"/>
97 </when>
98 <when value="pair">
99 <param name="read1" type="data" format="fastqsanger.gz,fastqsanger" label="Read1 fastq file"/>
100 <param name="read2" type="data" format="fastqsanger.gz,fastqsanger" label="Read2 fastq file"/>
101 </when>
102 </conditional>
103 <param name="samtools_idxstats" type="data" format="tabular" label="Samtools idxstats file"/>
104 <param name="vsnp_azc" type="data" format="tabular" label="vSNP: add zero coverage metrics file"/>
105 </when> 58 </when>
106 <when value="collections"> 59 <when value="paired">
107 <conditional name="collection_type_cond"> 60 <param name="reads_collection" type="data_collection" format="fastqsanger,fastqsanger.gz" collection_type="paired" label="Collection of fastqsanger paired read files"/>
108 <param name="collection_type" type="select" label="Collections of single reads or paired reads?"> 61 </when>
109 <option value="single" selected="true">Single reads</option> 62 <when value="pair">
110 <option value="paired">Paired reads in separate datasets</option> 63 <param name="read1" type="data" format="fastqsanger.gz,fastqsanger" label="Read1 fastq file"/>
111 </param> 64 <param name="read2" type="data" format="fastqsanger.gz,fastqsanger" label="Read2 fastq file"/>
112 <when value="single">
113 <param name="reads_collection" type="data_collection" format="fastqsanger,fastqsanger.gz" collection_type="list" label="Collection of fastqsanger files"/>
114 </when>
115 <when value="paired">
116 <param name="reads_collection" type="data_collection" format="fastqsanger,fastqsanger.gz" collection_type="paired" label="Collection of fastqsanger paired read files"/>
117 </when>
118 </conditional>
119 <param name="samtools_idxstats" type="data_collection" format="tabular" collection_type="list" label="Collection of samtools idxstats files"/>
120 <param name="vsnp_azc" type="data_collection" format="tabular" collection_type="list" label="Collection of vSNP: add zero coverage metrics files"/>
121 </when> 65 </when>
122 </conditional> 66 </conditional>
67 <param name="samtools_idxstats" type="data" format="tabular" label="Samtools idxstats file"/>
68 <param name="vsnp_azc_metrics" type="data" format="tabular" label="vSNP: add zero coverage metrics file"/>
123 </inputs> 69 </inputs>
124 <outputs> 70 <outputs>
125 <data name="output" format="tabular"/> 71 <data name="output" format="tabular"/>
126 </outputs> 72 </outputs>
127 <tests> 73 <tests>
128 <!-- A single fastq file --> 74 <!-- A single fastq file -->
129 <test expect_num_outputs="1"> 75 <test expect_num_outputs="1">
130 <param name="input_type" value="single_files"/> 76 <param name="input_type" value="single"/>
131 <param name="read_type" value="single"/>
132 <param name="read1" value="Mcap_Deer_DE_SRR650221.fastq.gz" ftype="fastqsanger.gz" dbkey="89"/> 77 <param name="read1" value="Mcap_Deer_DE_SRR650221.fastq.gz" ftype="fastqsanger.gz" dbkey="89"/>
133 <param name="samtools_idxstats" value="samtools_idxstats1.tabular" ftype="tabular" dbkey="89"/> 78 <param name="samtools_idxstats" value="samtools_idxstats1.tabular" ftype="tabular" dbkey="89"/>
134 <param name="vsnp_azc" value="add_zc_metrics1.tabular" ftype="tabular" dbkey="89"/> 79 <param name="vsnp_azc_metrics" value="add_zc_metrics1.tabular" ftype="tabular" dbkey="89"/>
135 <output name="output" file="vsnp_statistics1.tabular" ftype="tabular"/> 80 <output name="output" file="vsnp_statistics1.tabular" ftype="tabular"/>
136 </test> 81 </test>
137 <!-- A set of paired fastq files --> 82 <!-- A set of paired fastq files -->
138 <test expect_num_outputs="1"> 83 <test expect_num_outputs="1">
139 <param name="input_type" value="single_files"/> 84 <param name="input_type" value="pair"/>
140 <param name="read_type" value="pair"/>
141 <param name="read1" value="13-1941-6_S4_L001_R1_600000.fastq.gz" ftype="fastqsanger.gz" dbkey="89"/> 85 <param name="read1" value="13-1941-6_S4_L001_R1_600000.fastq.gz" ftype="fastqsanger.gz" dbkey="89"/>
142 <param name="read2" value="13-1941-6_S4_L001_R2_600000.fastq.gz" ftype="fastqsanger.gz" dbkey="89"/> 86 <param name="read2" value="13-1941-6_S4_L001_R2_600000.fastq.gz" ftype="fastqsanger.gz" dbkey="89"/>
143 <param name="samtools_idxstats" value="samtools_idxstats2.tabular" ftype="tabular" dbkey="89"/> 87 <param name="samtools_idxstats" value="samtools_idxstats2.tabular" ftype="tabular" dbkey="89"/>
144 <param name="vsnp_azc" value="add_zc_metrics2.tabular" ftype="tabular" dbkey="89"/> 88 <param name="vsnp_azc_metrics" value="add_zc_metrics2.tabular" ftype="tabular" dbkey="89"/>
145 <output name="output" file="vsnp_statistics2.tabular" ftype="tabular"/> 89 <output name="output" file="vsnp_statistics2.tabular" ftype="tabular"/>
146 </test> 90 </test>
147 <!-- A collection of SE fastq files --> 91 <!-- A collection of paired fastq files -->
148 <test expect_num_outputs="1"> 92 <test expect_num_outputs="1">
149 <param name="input_type" value="collections"/> 93 <param name="input_type" value="paired"/>
150 <param name="read_type" value="single"/>
151 <param name="reads_collection">
152 <collection type="list">
153 <element name="Mcap_Deer_DE_SRR650221.fastq.gz" value="Mcap_Deer_DE_SRR650221.fastq.gz" dbkey="89"/>
154 <element name="13-1941-6_S4_L001_R1_600000.fastq.gz" value="13-1941-6_S4_L001_R1_600000.fastq.gz" dbkey="89"/>
155 </collection>
156 </param>
157 <param name="samtools_idxstats">
158 <collection type="list">
159 <element name="13-1941-6_S4_L001_R1_600000.fastq.gz" value="samtools_idxstats3.tabular" dbkey="89"/>
160 <element name="Mcap_Deer_DE_SRR650221.fastq.gz" value="samtools_idxstats4.tabular" dbkey="89"/>
161 </collection>
162 </param>
163 <param name="vsnp_azc">
164 <collection type="list">
165 <element name="13-1941-6_S4_L001_R1_600000.fastq.gz" value="add_zc_metrics3.tabular" dbkey="89"/>
166 <element name="Mcap_Deer_DE_SRR650221.fastq.gz" value="add_zc_metrics4.tabular" dbkey="89"/>
167 </collection>
168 </param>
169 <output name="output" file="vsnp_statistics3.tabular" ftype="tabular"/>
170 </test>
171 <!-- A collection of PE fastq files -->
172 <test expect_num_outputs="1">
173 <param name="input_type" value="collections"/>
174 <param name="collection_type" value="paired"/>
175 <param name="reads_collection"> 94 <param name="reads_collection">
176 <collection type="paired"> 95 <collection type="paired">
177 <element name="forward" value="13-1941-6_S4_L001_R1_600000.fastq.gz" ftype="fastqsanger.gz"/> 96 <element name="forward" value="13-1941-6_S4_L001_R1_600000.fastq.gz" ftype="fastqsanger.gz"/>
178 <element name="reverse" value="13-1941-6_S4_L001_R2_600000.fastq.gz" ftype="fastqsanger.gz"/> 97 <element name="reverse" value="13-1941-6_S4_L001_R2_600000.fastq.gz" ftype="fastqsanger.gz"/>
179 </collection> 98 </collection>
180 </param> 99 </param>
181 <param name="samtools_idxstats"> 100 <param name="samtools_idxstats" value="samtools_idxstats5.tabular" ftype="tabular" dbkey="89"/>
182 <collection type="list"> 101 <param name="vsnp_azc_metrics" value="add_zc_metrics5.tabular" ftype="tabular" dbkey="89"/>
183 <element name="13-1941-6_S4_L001_R1_600000.fastq" value="samtools_idxstats5.tabular" dbkey="89"/>
184 </collection>
185 </param>
186 <param name="vsnp_azc">
187 <collection type="list">
188 <element name="13-1941-6_S4_L001_R1_600000.fastq" value="add_zc_metrics5.tabular" dbkey="89"/>
189 </collection>
190 </param>
191 <output name="output" file="vsnp_statistics4.tabular" ftype="tabular"/> 102 <output name="output" file="vsnp_statistics4.tabular" ftype="tabular"/>
192 </test> 103 </test>
193 </tests> 104 </tests>
194 <help> 105 <help>
195 **What it does** 106 **What it does**
196 107
197 Accepts associated fastq files, SAMtools idxstats files and **vSNP: add zero coverage** metrics files and extracts information from them 108 Accepts associated fastq files, SAMtools idxstats files and **vSNP: add zero coverage** metrics files and extracts information from them
198 to produce an Excel spreadsheet containing statistics for each sample. The samples can be single or paired reads, and all associated inputs 109 to produce an Excel spreadsheet containing statistics for each sample. The samples can be a single read, a single set of paired reads in
199 can be either single files or collections of files. The output statistics include reference, file size, mean read length, mean read quality, 110 separate datasets or collection of paired reads. The output statistics include reference, file size, mean read length, mean read quality,
200 reads passing Q30, total reads, all mapped reads, unmapped reads, unmapped reads percentage of total, reference with coverage, average depth 111 reads passing Q30, total reads, all mapped reads, unmapped reads, unmapped reads percentage of total, reference with coverage, average depth
201 of coverage and good SNP count. 112 of coverage and good SNP count.
202 </help> 113 </help>
203 <expand macro="citations"/> 114 <expand macro="citations"/>
204 </tool> 115 </tool>