Mercurial > repos > devteam > fastq_groomer

comparison fastq_groomer.xml @ 3:bbc72602f69f draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/galaxy_sequence_utils/fastq_groomer commit f2582539542b33240234e8ea6093e25d0aee9b6a
author devteam
date Sat, 30 Sep 2017 14:55:55 -0400
parents 6b294cefd2ae
children dafabb4c2037
comparison
equal deleted inserted replaced
2:5598a9614070 3:bbc72602f69f
1 <tool id="fastq_groomer" name="FASTQ Groomer" version="1.0.4"> 1 <tool id="fastq_groomer" name="FASTQ Groomer" version="1.1.1">
2 <description>convert between various FASTQ quality formats</description> 2 <description>convert between various FASTQ quality formats</description>
3 <requirements> 3 <requirements>
4 <requirement type="package" version="1.0.0">galaxy_sequence_utils</requirement> 4 <requirement type="package" version="1.1.1">galaxy_sequence_utils</requirement>
5 </requirements> 5 </requirements>
6 <command interpreter="python">fastq_groomer.py '$input_file' '$input_type' '$output_file' 6 <command><![CDATA[
7 #if str( $options_type['options_type_selector'] ) == 'basic': 7 gx-fastq-groomer '$input_file'
8 #if str( $input_type ) == 'cssanger': 8 #if $input_file.extension.endswith(".gz"):
9 'cssanger' 9 #set $suffix = ".gz"
10 #elif $input_file.extension.endswith(".bz2"):
11 #set $suffix = ".bz2"
10 #else: 12 #else:
11 'sanger' 13 #set $suffix = ""
12 #end if 14 #end if
13 'ascii' 'summarize_input' 15 $input_type$suffix '$output_file'
16 #if $options_type['options_type_selector'] == 'basic':
17 #if str($input_type) == 'cssanger':
18 cssanger
19 #else:
20 sanger
21 #end if
22 ascii summarize_input
14 #else: 23 #else:
15 '${options_type.output_type}' '${options_type.force_quality_encoding}' '${options_type.summarize_input}' 24 ${options_type.output_type} ${options_type.force_quality_encoding} ${options_type.summarize_input}
16 #end if 25 #end if
17 </command> 26 ]]></command>
18 <inputs> 27 <inputs>
19 <param name="input_file" type="data" format="fastq" label="File to groom" /> 28 <param name="input_file" type="data" format="fastq,fastq.gz,fastq.bz2" label="File to groom" />
20 <param name="input_type" type="select" label="Input FASTQ quality scores type"> 29 <param name="input_type" type="select" label="Input FASTQ quality scores type">
21 <option value="solexa">Solexa</option> 30 <option value="solexa">Solexa</option>
22 <option value="illumina">Illumina 1.3-1.7</option> 31 <option value="illumina">Illumina 1.3-1.7</option>
23 <option value="sanger" selected="True">Sanger &amp; Illumina 1.8+</option> 32 <option value="sanger" selected="true">Sanger &amp; Illumina 1.8+</option>
24 <option value="cssanger">Color Space Sanger</option> 33 <option value="cssanger">Color Space Sanger</option>
25 </param> 34 </param>
26 <conditional name="options_type"> 35 <conditional name="options_type">
27 <param name="options_type_selector" type="select" label="Advanced Options"> 36 <param name="options_type_selector" type="select" label="Advanced Options">
28 <option value="basic" selected="True">Hide Advanced Options</option> 37 <option value="basic" selected="true">Hide Advanced Options</option>
29 <option value="advanced">Show Advanced Options</option> 38 <option value="advanced">Show Advanced Options</option>
30 </param> 39 </param>
31 <when value="basic"> 40 <when value="basic" />
32 <!-- no options --> 41 <when value="advanced">
33 </when> 42 <param name="output_type" type="select" label="Output FASTQ quality scores type" help="Galaxy tools are designed to work with the Sanger Quality score format">
34 <when value="advanced"> 43 <option value="solexa">Solexa</option>
35 <param name="output_type" type="select" label="Output FASTQ quality scores type" help="Galaxy tools are designed to work with the Sanger Quality score format."> 44 <option value="illumina">Illumina 1.3-1.7</option>
36 <option value="solexa">Solexa</option> 45 <option value="sanger" selected="true">Sanger</option>
37 <option value="illumina">Illumina 1.3-1.7</option> 46 <option value="cssanger">Color Space Sanger</option>
38 <option value="sanger" selected="True">Sanger (recommended)</option> 47 <option value="solexa.gz">Solexa (gz compressed)</option>
39 <option value="cssanger">Color Space Sanger</option> 48 <option value="illumina.gz">Illumina 1.3-1.7 (gz compressed)</option>
40 </param> 49 <option value="sanger.gz">Sanger (gz compressed - recommended)</option>
41 <param name="force_quality_encoding" type="select" label="Force Quality Score encoding"> 50 <option value="cssanger.gz">Color Space Sanger (gz compressed)</option>
42 <option value="None">Use Source Encoding</option> 51 <option value="solexa.bz2">Solexa (bz2 compressed)</option>
43 <option value="ascii" selected="True">ASCII</option> 52 <option value="illumina.bz2">Illumina 1.3-1.7 (bz2 compressed)</option>
44 <option value="decimal">Decimal</option> 53 <option value="sanger.bz2">Sanger (bz2 compressed)</option>
45 </param> 54 <option value="cssanger.bz2">Color Space Sanger (bz2 compressed)</option>
46 <param name="summarize_input" type="select" label="Summarize input data"> 55 </param>
47 <option value="summarize_input" selected="True">Summarize Input</option> 56 <param name="force_quality_encoding" type="select" label="Force Quality Score encoding">
48 <option value="dont_summarize_input">Do not Summarize Input (faster)</option> 57 <option value="None">Use Source Encoding</option>
49 </param> 58 <option value="ascii" selected="true">ASCII</option>
50 </when> 59 <option value="decimal">Decimal</option>
51 </conditional> 60 </param>
52 </inputs> 61 <param name="summarize_input" type="select" label="Summarize input data">
53 <outputs> 62 <option value="summarize_input" selected="true">Summarize Input</option>
54 <data name="output_file" format="fastqsanger"> 63 <option value="dont_summarize_input">Do not Summarize Input (faster)</option>
55 <change_format> 64 </param>
56 <when input="input_type" value="cssanger" format="fastqcssanger" /> 65 </when>
57 <when input="options_type.output_type" value="solexa" format="fastqsolexa" /> 66 </conditional>
58 <when input="options_type.output_type" value="illumina" format="fastqillumina" /> 67 </inputs>
59 <when input="options_type.output_type" value="sanger" format="fastqsanger" /> 68 <outputs>
60 <when input="options_type.output_type" value="cssanger" format="fastqcssanger" /> 69 <data name="output_file" format="fastqsanger">
61 </change_format> 70 <change_format>
62 </data> 71 <when input="input_type" value="cssanger" format="fastqcssanger" />
63 </outputs> 72 <when input="options_type.output_type" value="solexa" format="fastqsolexa" />
64 <tests> 73 <when input="options_type.output_type" value="illumina" format="fastqillumina" />
65 <!-- These tests include test files adapted from supplemental material in Cock PJ, Fields CJ, Goto N, Heuer ML, Rice PM. The Sanger FASTQ file format for sequences with quality scores, and the Solexa/Illumina FASTQ variants. Nucleic Acids Res. 2009 Dec 16. --> 74 <when input="options_type.output_type" value="sanger" format="fastqsanger" />
66 <!-- Unfortunately, cannot test for expected failures --> 75 <when input="options_type.output_type" value="cssanger" format="fastqcssanger" />
67 <!-- Test basic options --> 76 <when input="options_type.output_type" value="solexa.gz" format="fastqsolexa.gz" />
68 <test> 77 <when input="options_type.output_type" value="illumina.gz" format="fastqillumina.gz" />
69 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" /> 78 <when input="options_type.output_type" value="sanger.gz" format="fastqsanger.gz" />
70 <param name="input_type" value="sanger" /> 79 <when input="options_type.output_type" value="cssanger.gz" format="fastqcssanger.gz" />
71 <param name="options_type_selector" value="basic" /> 80 <when input="options_type.output_type" value="solexa.bz2" format="fastqsolexa.bz2" />
72 <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" /> 81 <when input="options_type.output_type" value="illumina.bz2" format="fastqillumina.bz2" />
73 </test> 82 <when input="options_type.output_type" value="sanger.bz2" format="fastqsanger.bz2" />
74 <test> 83 <when input="options_type.output_type" value="cssanger.bz2" format="fastqcssanger.bz2" />
75 <param name="input_file" value="sanger_full_range_as_cssanger.fastqcssanger" ftype="fastq" /> 84 </change_format>
76 <param name="input_type" value="cssanger" /> 85 </data>
77 <param name="options_type_selector" value="basic" /> 86 </outputs>
78 <output name="output_file" file="sanger_full_range_as_cssanger.fastqcssanger" /> 87 <tests>
79 </test> 88 <!-- These tests include test files adapted from supplemental material in Cock PJ, Fields CJ, Goto N, Heuer ML, Rice PM. The Sanger FASTQ file format for sequences with quality scores, and the Solexa/Illumina FASTQ variants. Nucleic Acids Res. 2009 Dec 16. -->
80 <test> 89 <!-- Test basic options -->
81 <param name="input_file" value="illumina_full_range_original_illumina.fastqillumina" ftype="fastq" /> 90 <test>
82 <param name="input_type" value="illumina" /> 91 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" />
83 <param name="options_type_selector" value="basic" /> 92 <param name="input_type" value="sanger" />
84 <output name="output_file" file="illumina_full_range_as_sanger.fastqsanger" /> 93 <param name="options_type_selector" value="basic" />
85 </test> 94 <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" />
86 <test> 95 </test>
87 <param name="input_file" value="solexa_full_range_original_solexa.fastqsolexa" ftype="fastq" /> 96 <test>
88 <param name="input_type" value="solexa" /> 97 <param name="input_file" value="sanger_full_range_as_cssanger.fastqcssanger" ftype="fastq" />
89 <param name="options_type_selector" value="basic" /> 98 <param name="input_type" value="cssanger" />
90 <output name="output_file" file="solexa_full_range_as_sanger.fastqsanger" /> 99 <param name="options_type_selector" value="basic" />
91 </test> 100 <output name="output_file" file="sanger_full_range_as_cssanger.fastqcssanger" />
92 <test> 101 </test>
93 <param name="input_file" value="sanger_full_range_as_illumina.fastqillumina" ftype="fastq" /> 102 <test>
94 <param name="input_type" value="sanger" /> 103 <param name="input_file" value="illumina_full_range_original_illumina.fastqillumina" ftype="fastq" />
95 <param name="options_type_selector" value="basic" /> 104 <param name="input_type" value="illumina" />
96 <output name="output_file" file="sanger_full_range_as_illumina.fastqillumina" /> 105 <param name="options_type_selector" value="basic" />
97 </test> 106 <output name="output_file" file="illumina_full_range_as_sanger.fastqsanger" />
98 <!-- Test grooming from illumina --> 107 </test>
99 <test> 108 <test>
100 <param name="input_file" value="illumina_full_range_original_illumina.fastqillumina" ftype="fastq" /> 109 <param name="input_file" value="solexa_full_range_original_solexa.fastqsolexa" ftype="fastq" />
101 <param name="input_type" value="illumina" /> 110 <param name="input_type" value="solexa" />
102 <param name="options_type_selector" value="advanced" /> 111 <param name="options_type_selector" value="basic" />
103 <param name="output_type" value="illumina" /> 112 <output name="output_file" file="solexa_full_range_as_sanger.fastqsanger" />
104 <param name="force_quality_encoding" value="None" /> 113 </test>
105 <param name="summarize_input" value="summarize_input" /> 114 <test>
106 <output name="output_file" file="illumina_full_range_original_illumina.fastqillumina" /> 115 <param name="input_file" value="sanger_full_range_as_illumina.fastqillumina" ftype="fastq" />
107 </test> 116 <param name="input_type" value="sanger" />
108 <test> 117 <param name="options_type_selector" value="basic" />
109 <param name="input_file" value="illumina_full_range_original_illumina.fastqillumina" ftype="fastq" /> 118 <output name="output_file" file="sanger_full_range_as_illumina.fastqillumina" />
110 <param name="input_type" value="illumina" /> 119 </test>
111 <param name="options_type_selector" value="advanced" /> 120 <!-- Test grooming from illumina -->
112 <param name="output_type" value="sanger" /> 121 <test>
113 <param name="force_quality_encoding" value="None" /> 122 <param name="input_file" value="illumina_full_range_original_illumina.fastqillumina" ftype="fastq" />
114 <param name="summarize_input" value="summarize_input" /> 123 <param name="input_type" value="illumina" />
115 <output name="output_file" file="illumina_full_range_as_sanger.fastqsanger" /> 124 <param name="options_type_selector" value="advanced" />
116 </test> 125 <param name="output_type" value="illumina" />
117 <test> 126 <param name="force_quality_encoding" value="None" />
118 <param name="input_file" value="illumina_full_range_original_illumina.fastqillumina" ftype="fastq" /> 127 <param name="summarize_input" value="summarize_input" />
119 <param name="input_type" value="illumina" /> 128 <output name="output_file" file="illumina_full_range_original_illumina.fastqillumina" />
120 <param name="options_type_selector" value="advanced" /> 129 </test>
121 <param name="output_type" value="solexa" /> 130 <test>
122 <param name="force_quality_encoding" value="None" /> 131 <param name="input_file" value="illumina_full_range_original_illumina.fastqillumina" ftype="fastq" />
123 <param name="summarize_input" value="summarize_input" /> 132 <param name="input_type" value="illumina" />
124 <output name="output_file" file="illumina_full_range_as_solexa.fastqsolexa" /> 133 <param name="options_type_selector" value="advanced" />
125 </test> 134 <param name="output_type" value="sanger" />
126 <test> 135 <param name="force_quality_encoding" value="None" />
127 <param name="input_file" value="illumina_full_range_original_illumina.fastqillumina" ftype="fastq" /> 136 <param name="summarize_input" value="summarize_input" />
128 <param name="input_type" value="illumina" /> 137 <output name="output_file" file="illumina_full_range_as_sanger.fastqsanger" />
129 <param name="options_type_selector" value="advanced" /> 138 </test>
130 <param name="output_type" value="cssanger" /> 139 <test>
131 <param name="force_quality_encoding" value="None" /> 140 <param name="input_file" value="illumina_full_range_original_illumina.fastqillumina" ftype="fastq" />
132 <param name="summarize_input" value="summarize_input" /> 141 <param name="input_type" value="illumina" />
133 <output name="output_file" file="illumina_full_range_as_cssanger.fastqcssanger" /> 142 <param name="options_type_selector" value="advanced" />
134 </test> 143 <param name="output_type" value="solexa" />
135 <!-- Test grooming from sanger --> 144 <param name="force_quality_encoding" value="None" />
136 <test> 145 <param name="summarize_input" value="summarize_input" />
137 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" /> 146 <output name="output_file" file="illumina_full_range_as_solexa.fastqsolexa" />
138 <param name="input_type" value="sanger" /> 147 </test>
139 <param name="options_type_selector" value="advanced" /> 148 <test>
140 <param name="output_type" value="sanger" /> 149 <param name="input_file" value="illumina_full_range_original_illumina.fastqillumina" ftype="fastq" />
141 <param name="force_quality_encoding" value="None" /> 150 <param name="input_type" value="illumina" />
142 <param name="summarize_input" value="summarize_input" /> 151 <param name="options_type_selector" value="advanced" />
143 <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" /> 152 <param name="output_type" value="cssanger" />
144 </test> 153 <param name="force_quality_encoding" value="None" />
145 <test> 154 <param name="summarize_input" value="summarize_input" />
146 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" /> 155 <output name="output_file" file="illumina_full_range_as_cssanger.fastqcssanger" />
147 <param name="input_type" value="sanger" /> 156 </test>
148 <param name="options_type_selector" value="advanced" /> 157 <!-- Test grooming from sanger -->
149 <param name="output_type" value="illumina" /> 158 <test>
150 <param name="force_quality_encoding" value="None" /> 159 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" />
151 <param name="summarize_input" value="summarize_input" /> 160 <param name="input_type" value="sanger" />
152 <output name="output_file" file="sanger_full_range_as_illumina.fastqillumina" /> 161 <param name="options_type_selector" value="advanced" />
153 </test> 162 <param name="output_type" value="sanger" />
154 <test> 163 <param name="force_quality_encoding" value="None" />
155 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" /> 164 <param name="summarize_input" value="summarize_input" />
156 <param name="input_type" value="sanger" /> 165 <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" />
157 <param name="options_type_selector" value="advanced" /> 166 </test>
158 <param name="output_type" value="solexa" /> 167 <test>
159 <param name="force_quality_encoding" value="None" /> 168 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" />
160 <param name="summarize_input" value="summarize_input" /> 169 <param name="input_type" value="sanger" />
161 <output name="output_file" file="sanger_full_range_as_solexa.fastqsolexa" /> 170 <param name="options_type_selector" value="advanced" />
162 </test> 171 <param name="output_type" value="illumina" />
163 <test> 172 <param name="force_quality_encoding" value="None" />
164 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" /> 173 <param name="summarize_input" value="summarize_input" />
165 <param name="input_type" value="sanger" /> 174 <output name="output_file" file="sanger_full_range_as_illumina.fastqillumina" />
166 <param name="options_type_selector" value="advanced" /> 175 </test>
167 <param name="output_type" value="cssanger" /> 176 <test>
168 <param name="force_quality_encoding" value="None" /> 177 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" />
169 <param name="summarize_input" value="summarize_input" /> 178 <param name="input_type" value="sanger" />
170 <output name="output_file" file="sanger_full_range_as_cssanger.fastqcssanger" /> 179 <param name="options_type_selector" value="advanced" />
171 </test> 180 <param name="output_type" value="solexa" />
172 <!-- Test grooming from solexa --> 181 <param name="force_quality_encoding" value="None" />
173 <test> 182 <param name="summarize_input" value="summarize_input" />
174 <param name="input_file" value="solexa_full_range_original_solexa.fastqsolexa" ftype="fastq" /> 183 <output name="output_file" file="sanger_full_range_as_solexa.fastqsolexa" />
175 <param name="input_type" value="solexa" /> 184 </test>
176 <param name="options_type_selector" value="advanced" /> 185 <test>
177 <param name="output_type" value="solexa" /> 186 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" />
178 <param name="force_quality_encoding" value="None" /> 187 <param name="input_type" value="sanger" />
179 <param name="summarize_input" value="summarize_input" /> 188 <param name="options_type_selector" value="advanced" />
180 <output name="output_file" file="solexa_full_range_original_solexa.fastqsolexa" /> 189 <param name="output_type" value="cssanger" />
181 </test> 190 <param name="force_quality_encoding" value="None" />
182 <test> 191 <param name="summarize_input" value="summarize_input" />
183 <param name="input_file" value="solexa_full_range_original_solexa.fastqsolexa" ftype="fastq" /> 192 <output name="output_file" file="sanger_full_range_as_cssanger.fastqcssanger" />
184 <param name="input_type" value="solexa" /> 193 </test>
185 <param name="options_type_selector" value="advanced" /> 194 <!-- Test grooming from solexa -->
186 <param name="output_type" value="illumina" /> 195 <test>
187 <param name="force_quality_encoding" value="None" /> 196 <param name="input_file" value="solexa_full_range_original_solexa.fastqsolexa" ftype="fastq" />
188 <param name="summarize_input" value="summarize_input" /> 197 <param name="input_type" value="solexa" />
189 <output name="output_file" file="solexa_full_range_as_illumina.fastqillumina" /> 198 <param name="options_type_selector" value="advanced" />
190 </test> 199 <param name="output_type" value="solexa" />
191 <test> 200 <param name="force_quality_encoding" value="None" />
192 <param name="input_file" value="solexa_full_range_original_solexa.fastqsolexa" ftype="fastq" /> 201 <param name="summarize_input" value="summarize_input" />
193 <param name="input_type" value="solexa" /> 202 <output name="output_file" file="solexa_full_range_original_solexa.fastqsolexa" />
194 <param name="options_type_selector" value="advanced" /> 203 </test>
195 <param name="output_type" value="sanger" /> 204 <test>
196 <param name="force_quality_encoding" value="None" /> 205 <param name="input_file" value="solexa_full_range_original_solexa.fastqsolexa" ftype="fastq" />
197 <param name="summarize_input" value="summarize_input" /> 206 <param name="input_type" value="solexa" />
198 <output name="output_file" file="solexa_full_range_as_sanger.fastqsanger" /> 207 <param name="options_type_selector" value="advanced" />
199 </test> 208 <param name="output_type" value="illumina" />
200 <test> 209 <param name="force_quality_encoding" value="None" />
201 <param name="input_file" value="solexa_full_range_original_solexa.fastqsolexa" ftype="fastq" /> 210 <param name="summarize_input" value="summarize_input" />
202 <param name="input_type" value="solexa" /> 211 <output name="output_file" file="solexa_full_range_as_illumina.fastqillumina" />
203 <param name="options_type_selector" value="advanced" /> 212 </test>
204 <param name="output_type" value="cssanger" /> 213 <test>
205 <param name="force_quality_encoding" value="None" /> 214 <param name="input_file" value="solexa_full_range_original_solexa.fastqsolexa" ftype="fastq" />
206 <param name="summarize_input" value="summarize_input" /> 215 <param name="input_type" value="solexa" />
207 <output name="output_file" file="solexa_full_range_as_cssanger.fastqcssanger" /> 216 <param name="options_type_selector" value="advanced" />
208 </test> 217 <param name="output_type" value="sanger" />
209 <!-- Test grooming from cssanger --> 218 <param name="force_quality_encoding" value="None" />
210 <test> 219 <param name="summarize_input" value="summarize_input" />
211 <param name="input_file" value="sanger_full_range_as_cssanger.fastqcssanger" ftype="fastq" /> 220 <output name="output_file" file="solexa_full_range_as_sanger.fastqsanger" />
212 <param name="input_type" value="cssanger" /> 221 </test>
213 <param name="options_type_selector" value="advanced" /> 222 <test>
214 <param name="output_type" value="cssanger" /> 223 <param name="input_file" value="solexa_full_range_original_solexa.fastqsolexa" ftype="fastq" />
215 <param name="force_quality_encoding" value="None" /> 224 <param name="input_type" value="solexa" />
216 <param name="summarize_input" value="summarize_input" /> 225 <param name="options_type_selector" value="advanced" />
217 <output name="output_file" file="sanger_full_range_as_cssanger.fastqcssanger" /> 226 <param name="output_type" value="cssanger" />
218 </test> 227 <param name="force_quality_encoding" value="None" />
219 <test> 228 <param name="summarize_input" value="summarize_input" />
220 <param name="input_file" value="sanger_full_range_as_cssanger.fastqcssanger" ftype="fastq" /> 229 <output name="output_file" file="solexa_full_range_as_cssanger.fastqcssanger" />
221 <param name="input_type" value="cssanger" /> 230 </test>
222 <param name="options_type_selector" value="advanced" /> 231 <!-- Test grooming from cssanger -->
223 <param name="output_type" value="sanger" /> 232 <test>
224 <param name="force_quality_encoding" value="None" /> 233 <param name="input_file" value="sanger_full_range_as_cssanger.fastqcssanger" ftype="fastq" />
225 <param name="summarize_input" value="summarize_input" /> 234 <param name="input_type" value="cssanger" />
226 <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" /> 235 <param name="options_type_selector" value="advanced" />
227 </test> 236 <param name="output_type" value="cssanger" />
228 <test> 237 <param name="force_quality_encoding" value="None" />
229 <param name="input_file" value="sanger_full_range_as_cssanger.fastqcssanger" ftype="fastq" /> 238 <param name="summarize_input" value="summarize_input" />
230 <param name="input_type" value="cssanger" /> 239 <output name="output_file" file="sanger_full_range_as_cssanger.fastqcssanger" />
231 <param name="options_type_selector" value="advanced" /> 240 </test>
232 <param name="output_type" value="illumina" /> 241 <test>
233 <param name="force_quality_encoding" value="None" /> 242 <param name="input_file" value="sanger_full_range_as_cssanger.fastqcssanger" ftype="fastq" />
234 <param name="summarize_input" value="summarize_input" /> 243 <param name="input_type" value="cssanger" />
235 <output name="output_file" file="sanger_full_range_as_illumina.fastqillumina" /> 244 <param name="options_type_selector" value="advanced" />
236 </test> 245 <param name="output_type" value="sanger" />
237 <test> 246 <param name="force_quality_encoding" value="None" />
238 <param name="input_file" value="sanger_full_range_as_cssanger.fastqcssanger" ftype="fastq" /> 247 <param name="summarize_input" value="summarize_input" />
239 <param name="input_type" value="cssanger" /> 248 <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" />
240 <param name="options_type_selector" value="advanced" /> 249 </test>
241 <param name="output_type" value="solexa" /> 250 <test>
242 <param name="force_quality_encoding" value="None" /> 251 <param name="input_file" value="sanger_full_range_as_cssanger.fastqcssanger" ftype="fastq" />
243 <param name="summarize_input" value="summarize_input" /> 252 <param name="input_type" value="cssanger" />
244 <output name="output_file" file="sanger_full_range_as_solexa.fastqsolexa" /> 253 <param name="options_type_selector" value="advanced" />
245 </test> 254 <param name="output_type" value="illumina" />
246 <test> 255 <param name="force_quality_encoding" value="None" />
247 <param name="input_file" value="sanger_full_range_as_cssanger_adapter_base_with_quality_score.fastqcssanger_fake_score" ftype="fastq" /> 256 <param name="summarize_input" value="summarize_input" />
248 <param name="input_type" value="cssanger" /> 257 <output name="output_file" file="sanger_full_range_as_illumina.fastqillumina" />
249 <param name="options_type_selector" value="advanced" /> 258 </test>
250 <param name="output_type" value="cssanger" /> 259 <test>
251 <param name="force_quality_encoding" value="None" /> 260 <param name="input_file" value="sanger_full_range_as_cssanger.fastqcssanger" ftype="fastq" />
252 <param name="summarize_input" value="summarize_input" /> 261 <param name="input_type" value="cssanger" />
253 <output name="output_file" file="sanger_full_range_as_cssanger.fastqcssanger" /> 262 <param name="options_type_selector" value="advanced" />
254 </test> 263 <param name="output_type" value="solexa" />
255 <!-- Test fastq with line wrapping --> 264 <param name="force_quality_encoding" value="None" />
256 <test> 265 <param name="summarize_input" value="summarize_input" />
257 <param name="input_file" value="wrapping_original_sanger.fastqsanger" ftype="fastq" /> 266 <output name="output_file" file="sanger_full_range_as_solexa.fastqsolexa" />
258 <param name="input_type" value="sanger" /> 267 </test>
259 <param name="options_type_selector" value="advanced" /> 268 <test>
260 <param name="output_type" value="sanger" /> 269 <param name="input_file" value="sanger_full_range_as_cssanger_adapter_base_with_quality_score.fastqcssanger_fake_score" ftype="fastq" />
261 <param name="force_quality_encoding" value="None" /> 270 <param name="input_type" value="cssanger" />
262 <param name="summarize_input" value="summarize_input" /> 271 <param name="options_type_selector" value="advanced" />
263 <output name="output_file" file="wrapping_as_sanger.fastqsanger" /> 272 <param name="output_type" value="cssanger" />
264 </test> 273 <param name="force_quality_encoding" value="None" />
265 <test> 274 <param name="summarize_input" value="summarize_input" />
266 <param name="input_file" value="wrapping_original_sanger.fastqsanger" ftype="fastq" /> 275 <output name="output_file" file="sanger_full_range_as_cssanger.fastqcssanger" />
267 <param name="input_type" value="sanger" /> 276 </test>
268 <param name="options_type_selector" value="advanced" /> 277 <!-- Test fastq with line wrapping -->
269 <param name="output_type" value="illumina" /> 278 <test>
270 <param name="force_quality_encoding" value="None" /> 279 <param name="input_file" value="wrapping_original_sanger.fastqsanger" ftype="fastq" />
271 <param name="summarize_input" value="summarize_input" /> 280 <param name="input_type" value="sanger" />
272 <output name="output_file" file="wrapping_as_illumina.fastqillumina" /> 281 <param name="options_type_selector" value="advanced" />
273 </test> 282 <param name="output_type" value="sanger" />
274 <test> 283 <param name="force_quality_encoding" value="None" />
275 <param name="input_file" value="wrapping_original_sanger.fastqsanger" ftype="fastq" /> 284 <param name="summarize_input" value="summarize_input" />
276 <param name="input_type" value="sanger" /> 285 <output name="output_file" file="wrapping_as_sanger.fastqsanger" />
277 <param name="options_type_selector" value="advanced" /> 286 </test>
278 <param name="output_type" value="solexa" /> 287 <test>
279 <param name="force_quality_encoding" value="None" /> 288 <param name="input_file" value="wrapping_original_sanger.fastqsanger" ftype="fastq" />
280 <param name="summarize_input" value="summarize_input" /> 289 <param name="input_type" value="sanger" />
281 <output name="output_file" file="wrapping_as_solexa.fastqsolexa" /> 290 <param name="options_type_selector" value="advanced" />
282 </test> 291 <param name="output_type" value="illumina" />
283 <!-- Test forcing quality score encoding --> 292 <param name="force_quality_encoding" value="None" />
284 <!-- Sanger, range 0 - 93 --> 293 <param name="summarize_input" value="summarize_input" />
285 <test> 294 <output name="output_file" file="wrapping_as_illumina.fastqillumina" />
286 <param name="input_file" value="sanger_full_range_as_decimal_sanger.fastqsanger" ftype="fastq" /> 295 </test>
287 <param name="input_type" value="sanger" /> 296 <test>
288 <param name="options_type_selector" value="advanced" /> 297 <param name="input_file" value="wrapping_original_sanger.fastqsanger" ftype="fastq" />
289 <param name="output_type" value="sanger" /> 298 <param name="input_type" value="sanger" />
290 <param name="force_quality_encoding" value="ascii" /> 299 <param name="options_type_selector" value="advanced" />
291 <param name="summarize_input" value="summarize_input" /> 300 <param name="output_type" value="solexa" />
292 <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" /> 301 <param name="force_quality_encoding" value="None" />
293 </test> 302 <param name="summarize_input" value="summarize_input" />
294 <test> 303 <output name="output_file" file="wrapping_as_solexa.fastqsolexa" />
295 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" /> 304 </test>
296 <param name="input_type" value="sanger" /> 305 <!-- Test forcing quality score encoding -->
297 <param name="options_type_selector" value="advanced" /> 306 <!-- Sanger, range 0 - 93 -->
298 <param name="output_type" value="sanger" /> 307 <test>
299 <param name="force_quality_encoding" value="decimal" /> 308 <param name="input_file" value="sanger_full_range_as_decimal_sanger.fastqsanger" ftype="fastq" />
300 <param name="summarize_input" value="summarize_input" /> 309 <param name="input_type" value="sanger" />
301 <output name="output_file" file="sanger_full_range_as_decimal_sanger.fastqsanger" /> 310 <param name="options_type_selector" value="advanced" />
302 </test> 311 <param name="output_type" value="sanger" />
303 <test> 312 <param name="force_quality_encoding" value="ascii" />
304 <param name="input_file" value="sanger_full_range_as_tab_decimal_sanger.fastqsanger" ftype="fastq" /> 313 <param name="summarize_input" value="summarize_input" />
305 <param name="input_type" value="sanger" /> 314 <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" />
306 <param name="options_type_selector" value="advanced" /> 315 </test>
307 <param name="output_type" value="sanger" /> 316 <test>
308 <param name="force_quality_encoding" value="ascii" /> 317 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" />
309 <param name="summarize_input" value="summarize_input" /> 318 <param name="input_type" value="sanger" />
310 <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" /> 319 <param name="options_type_selector" value="advanced" />
311 </test> 320 <param name="output_type" value="sanger" />
312 <!-- Solexa, range -5 - 62 --> 321 <param name="force_quality_encoding" value="decimal" />
313 <test> 322 <param name="summarize_input" value="summarize_input" />
314 <param name="input_file" value="solexa_full_range_as_decimal_solexa.fastqsolexa" ftype="fastq" /> 323 <output name="output_file" file="sanger_full_range_as_decimal_sanger.fastqsanger" />
315 <param name="input_type" value="solexa" /> 324 </test>
316 <param name="options_type_selector" value="advanced" /> 325 <test>
317 <param name="output_type" value="solexa" /> 326 <param name="input_file" value="sanger_full_range_as_tab_decimal_sanger.fastqsanger" ftype="fastq" />
318 <param name="force_quality_encoding" value="ascii" /> 327 <param name="input_type" value="sanger" />
319 <param name="summarize_input" value="summarize_input" /> 328 <param name="options_type_selector" value="advanced" />
320 <output name="output_file" file="solexa_full_range_original_solexa.fastqsolexa" /> 329 <param name="output_type" value="sanger" />
321 </test> 330 <param name="force_quality_encoding" value="ascii" />
322 <test> 331 <param name="summarize_input" value="summarize_input" />
323 <param name="input_file" value="solexa_full_range_original_solexa.fastqsolexa" ftype="fastq" /> 332 <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" />
324 <param name="input_type" value="solexa" /> 333 </test>
325 <param name="options_type_selector" value="advanced" /> 334 <!-- Solexa, range -5 - 62 -->
326 <param name="output_type" value="solexa" /> 335 <test>
327 <param name="force_quality_encoding" value="decimal" /> 336 <param name="input_file" value="solexa_full_range_as_decimal_solexa.fastqsolexa" ftype="fastq" />
328 <param name="summarize_input" value="summarize_input" /> 337 <param name="input_type" value="solexa" />
329 <output name="output_file" file="solexa_full_range_as_decimal_solexa.fastqsolexa" /> 338 <param name="options_type_selector" value="advanced" />
330 </test> 339 <param name="output_type" value="solexa" />
331 </tests> 340 <param name="force_quality_encoding" value="ascii" />
332 <help> 341 <param name="summarize_input" value="summarize_input" />
342 <output name="output_file" file="solexa_full_range_original_solexa.fastqsolexa" />
343 </test>
344 <test>
345 <param name="input_file" value="solexa_full_range_original_solexa.fastqsolexa" ftype="fastq" />
346 <param name="input_type" value="solexa" />
347 <param name="options_type_selector" value="advanced" />
348 <param name="output_type" value="solexa" />
349 <param name="force_quality_encoding" value="decimal" />
350 <param name="summarize_input" value="summarize_input" />
351 <output name="output_file" file="solexa_full_range_as_decimal_solexa.fastqsolexa" />
352 </test>
353 <!-- compressed formats -->
354 <test>
355 <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger.gz" ftype="fastq.gz" />
356 <param name="input_type" value="sanger" />
357 <param name="options_type_selector" value="basic" />
358 <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" decompress="true" />
359 </test>
360 <test>
361 <param name="input_file" value="sanger_full_range_as_cssanger.fastqcssanger.bz2" ftype="fastq.bz2" />
362 <param name="input_type" value="cssanger" />
363 <param name="options_type_selector" value="basic" />
364 <output name="output_file" file="sanger_full_range_as_cssanger.fastqcssanger" decompress="true" />
365 </test>
366 </tests>
367 <help><![CDATA[
333 **What it does** 368 **What it does**
334 369
335 This tool offers several conversions options relating to the FASTQ format. 370 This tool offers several conversions options relating to the FASTQ format.
336 371
337 When using *Basic* options, the output will be *sanger* formatted or *cssanger* formatted (when the input is Color Space Sanger). 372 When using *Basic* options, the output will be *sanger* formatted or *cssanger* formatted (when the input is Color Space Sanger).
338 373
339 When converting, if a quality score falls outside of the target score range, it will be coerced to the closest available value (i.e. the minimum or maximum). 374 When converting, if a quality score falls outside of the target score range, it will be coerced to the closest available value (i.e. the minimum or maximum).
340 375
341 When converting between Solexa and the other formats, quality scores are mapped between Solexa and PHRED scales using the equations found in `Cock PJ, Fields CJ, Goto N, Heuer ML, Rice PM. The Sanger FASTQ file format for sequences with quality scores, and the Solexa/Illumina FASTQ variants. Nucleic Acids Res. 2009 Dec 16.`_ 376 When converting between Solexa and the other formats, quality scores are mapped between Solexa and PHRED scales using the equations found in `Cock PJ, Fields CJ, Goto N, Heuer ML, Rice PM. The Sanger FASTQ file format for sequences with quality scores, and the Solexa/Illumina FASTQ variants. Nucleic Acids Res. 2009 Dec 16.`_
342 377
343 When converting between color space (csSanger) and base/sequence space (Sanger, Illumina, Solexa) formats, adapter bases are lost or gained; if gained, the base 'G' is used as the adapter. You cannot convert a color space read to base space if there is no adapter present in the color space sequence. Any masked or ambiguous nucleotides in base space will be converted to 'N's when determining color space encoding. 378 When converting between color space (csSanger) and base/sequence space (Sanger, Illumina, Solexa) formats, adapter bases are lost or gained; if gained, the base 'G' is used as the adapter. You cannot convert a color space read to base space if there is no adapter present in the color space sequence. Any masked or ambiguous nucleotides in base space will be converted to 'N's when determining color space encoding.
344 379
349 :: 384 ::
350 385
351 SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS 386 SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS
352 ...............................IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII 387 ...............................IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
353 ..........................XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX 388 ..........................XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
354 !"#$%&amp;'()*+,-./0123456789:;&lt;=&gt;?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~ 389 !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~
355 | | | | | | 390 | | | | | |
356 33 59 64 73 104 126 391 33 59 64 73 104 126
357 392
358 S - Sanger Phred+33, 93 values (0, 93) (0 to 60 expected in raw reads) 393 S - Sanger Phred+33, 93 values (0, 93) (0 to 60 expected in raw reads)
359 I - Illumina 1.3 Phred+64, 62 values (0, 62) (0 to 40 expected in raw reads) 394 I - Illumina 1.3 Phred+64, 62 values (0, 62) (0 to 40 expected in raw reads)
360 X - Solexa Solexa+64, 67 values (-5, 62) (-5 to 40 expected in raw reads) 395 X - Solexa Solexa+64, 67 values (-5, 62) (-5 to 40 expected in raw reads)
361 396
362 Diagram adapted from http://en.wikipedia.org/wiki/FASTQ_format 397 Diagram adapted from http://en.wikipedia.org/wiki/FASTQ_format
365 400
366 Output from Illumina 1.8+ pipelines are Sanger encoded. 401 Output from Illumina 1.8+ pipelines are Sanger encoded.
367 402
368 ------ 403 ------
369 404
370 405 .. _Cock PJ, Fields CJ, Goto N, Heuer ML, Rice PM. The Sanger FASTQ file format for sequences with quality scores, and the Solexa/Illumina FASTQ variants. Nucleic Acids Res. 2009 Dec 16.: https://doi.org/10.1093/nar/gkp1137
371 .. _Cock PJ, Fields CJ, Goto N, Heuer ML, Rice PM. The Sanger FASTQ file format for sequences with quality scores, and the Solexa/Illumina FASTQ variants. Nucleic Acids Res. 2009 Dec 16.: http://www.ncbi.nlm.nih.gov/pubmed/20015970 406 ]]></help>
372 407 <citations>
373 </help> 408 <citation type="doi">10.1093/bioinformatics/btq281</citation>
374 <citations> 409 </citations>
375 <citation type="doi">10.1093/bioinformatics/btq281</citation>
376 </citations>
377
378 </tool> 410 </tool>