|
5
|
1 <tool id="SAINT_preprocessing_v5" name="SAINT pre-processing">
|
|
|
2 <description></description>
|
|
|
3 <command interpreter="python">
|
|
60
|
4 #if (str($type_select.type) == 'Scaffold'):
|
|
62
|
5 SAINT_preprocessing.py $type_select.input $type_select.preybool $type_select.fasta_db $Inter_file $Prey_file
|
|
5
|
6 "
|
|
|
7 #for $ba in $bait
|
|
|
8 ${ba.bait1}
|
|
|
9 ${ba.assign}
|
|
|
10 ${ba.T_C}
|
|
|
11 #end for
|
|
|
12 "
|
|
62
|
13 $Bait_file \$INSTALL_RUN_PATH/ $type_select.bait_bool $type_select.bait_file_in
|
|
60
|
14 #elif (str($type_select.type) == 'MaxQuant'):
|
|
62
|
15 SAINT_preprocessing_mq_pep.py $type_select.input $type_select.preybool $type_select.fasta_db $Inter_file $Prey_file
|
|
5
|
16 "
|
|
|
17 #for $ba in $bait
|
|
|
18 ${ba.bait1}
|
|
|
19 ${ba.assign}
|
|
|
20 ${ba.T_C}
|
|
|
21 #end for
|
|
|
22 "
|
|
62
|
23 $Bait_file \$INSTALL_RUN_PATH/ $type_select.bait_bool $type_select.bait_file_in
|
|
60
|
24 #elif (str($type_select.type) == 'Peptideshaker'):
|
|
62
|
25 Protein_report_processing.py "$type_select.input" $Bait_file $type_select.use_metric $type_select.fasta_db $type_select.preybool
|
|
55
|
26 "
|
|
|
27 #for $ba in $bait
|
|
|
28 ${ba.bait1}
|
|
|
29 ${ba.assign}
|
|
|
30 ${ba.T_C}
|
|
|
31 #end for
|
|
|
32 "
|
|
62
|
33 \$INSTALL_RUN_PATH/ $type_select.bait_bool $type_select.bait_file_in $Inter_file $Prey_file
|
|
63
|
34 #elif (str($type_select.type) == 'mzIdentML'):
|
|
|
35 Protein_report_processing.py "$type_select.input" $type_select.bait_file_in $type_select.preybool $type_select.fasta_db \$INSTALL_RUN_PATH/
|
|
|
36 "
|
|
|
37 #for $ba in $bait
|
|
|
38 ${ba.bait1}
|
|
|
39 ${ba.assign}
|
|
|
40 ${ba.T_C}
|
|
|
41 #end for
|
|
|
42 "
|
|
|
43 $type_select.bait_bool $Prey_file $Bait_file $Inter_file
|
|
5
|
44 #end if
|
|
|
45 </command>
|
|
|
46 <requirements>
|
|
|
47 <requirement type="set_environment">INSTALL_RUN_PATH</requirement>
|
|
|
48 </requirements>
|
|
|
49 <inputs>
|
|
55
|
50 <conditional name="type_select">
|
|
60
|
51 <param type="select" name="type" label="Data Source">
|
|
55
|
52 <option value="MaxQuant">MaxQuant</option>
|
|
|
53 <option value="Scaffold">Scaffold</option>
|
|
|
54 <option value="Peptideshaker">Peptideshaker</option>
|
|
63
|
55 <option value="mzIdentML">mzIdentML</option>
|
|
55
|
56 </param>
|
|
|
57 <when value="MaxQuant">
|
|
|
58 <param format="dat" name="input" type="data" label="MaxQuant peptides Output"/>
|
|
|
59 <param type="boolean" name="preybool" checked="true" label="Create Prey File"/>
|
|
|
60 <param type="data" name="fasta_db" format="fasta" label="Provide Uniprot Fasta database" optional="true"/>
|
|
|
61 <param name="bait_bool" type="boolean" checked="true" label="Are You Providing Your Own bait file?"/>
|
|
|
62 <param type="data" format="dat" name="bait_file_in" label="Bait File" optional="true"/>
|
|
59
|
63 </when>
|
|
55
|
64 <when value="Scaffold">
|
|
|
65 <param format="dat" name="input" type="data" label="Scaffold Output"/>
|
|
|
66 <param type="boolean" name="preybool" checked="true" label="Create Prey File"/>
|
|
|
67 <param type="data" name="fasta_db" format="fasta" label="Provide Uniprot Fasta database" optional="true"/>
|
|
|
68 <param name="bait_bool" type="boolean" checked="true" label="Are You Providing Your Own bait file?"/>
|
|
|
69 <param type="data" format="dat" name="bait_file_in" label="Bait File" optional="true"/>
|
|
|
70 </when>
|
|
|
71 <when value="Peptideshaker">
|
|
|
72 <param format="dat" name="input" type="data" label="Peptideshaker Output" multiple="true"/>
|
|
|
73 <param type="select" name="use_metric" label="Select Report File Value for Quantification">
|
|
|
74 <option value="Validated_Peptides">#Validated Peptides</option>
|
|
|
75 <option value="Peptides">#Peptides</option>
|
|
|
76 <option value="Unique">#Unique</option>
|
|
|
77 <option value="Validated_PSMs">#Validated PSMs</option>
|
|
|
78 <option value="PSMs">#PSMs</option>
|
|
|
79 </param>
|
|
|
80 <param type="boolean" name="preybool" checked="true" label="Create Prey File"/>
|
|
|
81 <param type="data" name="fasta_db" format="fasta" label="Provide Uniprot Fasta database" optional="true"/>
|
|
|
82 <param name="bait_bool" type="boolean" checked="true" label="Are You Providing Your Own bait file?"/>
|
|
|
83 <param type="data" format="dat" name="bait_file_in" label="Bait File" optional="true"/>
|
|
|
84 </when>
|
|
63
|
85 <when value="mzIdentML">
|
|
|
86 <param format="dat" name="input" type="data" label="mzIdentML Files" multiple="true"/>
|
|
|
87 <param type="boolean" name="preybool" checked="true" label="Create Prey File"/>
|
|
|
88 <param type="data" name="fasta_db" format="fasta" label="Provide Uniprot Fasta database" optional="true"/>
|
|
|
89 <param name="bait_bool" type="boolean" checked="true" label="Are You Providing Your Own bait file?"/>
|
|
|
90 <param type="data" format="dat" name="bait_file_in" label="Bait File" optional="true"/>
|
|
|
91 </when>
|
|
55
|
92 </conditional>
|
|
61
|
93 <repeat name="bait" title="Bait Create">
|
|
|
94 <param name="bait1" type="text" size="100" label="Bait"/>
|
|
|
95 <param name="assign" type="text" size="100" label="Group Assignment"/>
|
|
|
96 <param name="T_C" type="boolean" checked="true" label="Is this a Control?"/>
|
|
|
97 </repeat>
|
|
5
|
98 </inputs>
|
|
|
99 <outputs>
|
|
|
100 <data format="txt" name="Inter_file" label="Inter File"/>
|
|
|
101 <data format="txt" name="Prey_file" label="Prey File" />
|
|
|
102 <data format="txt" name="Bait_file" label="Bait File" />
|
|
|
103 </outputs>
|
|
|
104 <stdio>
|
|
52
|
105 <regex match="Error: bad bait"
|
|
|
106 source="stdout"
|
|
|
107 level="fatal"
|
|
|
108 description="Error: bad bait"/>
|
|
|
109 <regex match="IOError: [Errno 2] No such file or directory: './tukeys_output.txt'"
|
|
|
110 source="stderr"
|
|
|
111 level="fatal"
|
|
|
112 description="Error: Scaffold/MaxQuant mismatch. Check job settings."/>
|
|
|
113 <regex match="bait_temp_file = open(sys.argv[10], 'r')"
|
|
|
114 source="stderr"
|
|
|
115 level="fatal"
|
|
|
116 description="Error: Bait create settings mismatch. Check job settings."/>
|
|
5
|
117 <regex match="Error|error"
|
|
|
118 source="stdout"
|
|
|
119 level="fatal"
|
|
|
120 description="Unknown error"/>
|
|
|
121 <regex match="Error|error"
|
|
|
122 source="stderr"
|
|
|
123 level="fatal"
|
|
|
124 description="Unknown error"/>
|
|
|
125 </stdio>
|
|
|
126
|
|
|
127 <tests>
|
|
|
128 <test>
|
|
|
129 <param name="input" value="fa_gc_content_input.fa"/>
|
|
|
130 <output name="out_file1" file="fa_gc_content_output.txt"/>
|
|
|
131 </test>
|
|
|
132 </tests>
|
|
|
133 <help>
|
|
21
|
134 Pre-Processing
|
|
25
|
135 ^^^^^^^^^^^^^^
|
|
20
|
136
|
|
21
|
137 This tool will read in a Scaffold *Samples Report* file (tab-delimited
|
|
55
|
138 txt file), multiple Peptideshaker *Protein Report* files or a MaxQuant
|
|
|
139 *peptides.txt* file and process them to generate a *Bait File, Prey
|
|
|
140 File,* and *Inter File* for SAINTexpress analysis.
|
|
20
|
141
|
|
25
|
142 --------------
|
|
|
143
|
|
27
|
144 **1) MaxQuant or Scaffold**
|
|
20
|
145
|
|
21
|
146 APOSTL is able to recognize either a Scaffold *Samples Report* file
|
|
|
147 (tab-delimited txt file) or the *peptides.txt* file output in the
|
|
|
148 MaxQuant *txt* output folder. No modifications should be made to these
|
|
|
149 files. Please designate which one is being provided.
|
|
20
|
150
|
|
27
|
151 **2) Scaffold or MaxQuant File Input**
|
|
20
|
152
|
|
21
|
153 Select the corresponding *Samples Report* or *peptides.txt* here.
|
|
|
154
|
|
27
|
155 **3) Create Prey File**
|
|
5
|
156
|
|
20
|
157 Select whether or not you would like APOSTL to generate a prey file.
|
|
|
158
|
|
21
|
159 When making a prey file, APOSTL queries a user provided FASTA database
|
|
|
160 (see below) in order to extract protein amino acid lengths and gene
|
|
|
161 names. This may take several minutes depending on your computer and if
|
|
|
162 your Galaxy distribution is cluster enabled. Some users may want to run
|
|
|
163 SAINTexpress using the same data set while changing which baits are
|
|
|
164 considered test or control. It is useful to toggle **Create Prey File**
|
|
|
165 off in order to save time by circumventing this step as the same prey
|
|
|
166 file can be used for both SAINTexpress runs.
|
|
20
|
167
|
|
27
|
168 **4) Provide Uniprot FASTA database**
|
|
20
|
169
|
|
21
|
170 Please specify a FASTA file containing the protein sequence. You can
|
|
|
171 download the latest release of Uniprot's reviewed sequences (Swiss-Prot)
|
|
24
|
172 http://www.uniprot.org/downloads. For computing efficiency,
|
|
21
|
173 please limit your FASTA file to your species of interest. It is
|
|
|
174 generally advised to use the same FASTA database that was used for your
|
|
|
175 database searches during peptide and protein ID.
|
|
5
|
176
|
|
27
|
177 **5) Are You Providing Your Own Bait File?**
|
|
20
|
178
|
|
21
|
179 Users have two options for providing a *Bait* file. If you would like to
|
|
|
180 create your own *Bait* file, select Yes and designate the file below. If
|
|
|
181 you would like to use the **Bait Create** tool to create a *Bait* file
|
|
|
182 within Galaxy, please select No and proceed to the **Bait Create** tool.
|
|
20
|
183
|
|
27
|
184 **6) Bait File**
|
|
21
|
185
|
|
|
186 If you selected Yes above, please specify the *Bait* file you would like
|
|
|
187 to provide. If you selected No above, ignore this argument and proceed
|
|
|
188 to the **Bait Create** tool.
|
|
5
|
189
|
|
21
|
190 **Note:** Individual bait names must match to the sample names
|
|
|
191 within your Scaffold or MaxQuant output exactly and must contain no
|
|
|
192 whitespace characters (e.g. spaces) or dashes.
|
|
59
|
193
|
|
55
|
194 **Note:** When using Peptideshaker output the baits in the baitfile
|
|
|
195 need to be in the same order as the files selected.
|
|
5
|
196
|
|
27
|
197 **7) Bait Create**
|
|
21
|
198
|
|
|
199 Using the **Bait Create** tool, you can create your *bait.txt* file.
|
|
20
|
200
|
|
21
|
201 **Note:** Individual bait names must match to the sample names
|
|
|
202 within your Scaffold or MaxQuant output exactly and must contain no
|
|
|
203 whitespace characters (e.g. spaces) or dashes.
|
|
5
|
204
|
|
21
|
205 APOSTL uses this bait file to find the user's baits of interest within a
|
|
|
206 *Samples Report* or *peptides.txt* file and when preparing the
|
|
|
207 *inter.txt* file.
|
|
20
|
208
|
|
21
|
209 --------------
|
|
20
|
210
|
|
21
|
211 Once your parameters have been finalized, please press the Execute
|
|
|
212 button to start processing. This may take a few minutes. Once your
|
|
|
213 process has been completed, you will see your *Bait File, Prey File,*
|
|
|
214 and *Inter File* on the right hand side of your panel highlighted in
|
|
|
215 green.
|
|
5
|
216 </help>
|
|
|
217 </tool>
|
