|
0
|
1 <tool id="sequence2gpsan_rnashapes" name="Sequence to gspan" version="0.3">
|
|
|
2 <description>converter using RNAshapes</description>
|
|
|
3 <parallelism method="multi" split_inputs="infile" split_mode="to_size" split_size="10" shared_inputs="" merge_outputs="outfile"></parallelism>
|
|
|
4 <requirements>
|
|
|
5 <requirement type="package" version="2.1.6">rnashapes</requirement>
|
|
|
6 </requirements>
|
|
|
7 <command interpreter="perl">
|
|
|
8
|
|
|
9 fasta2shrep_gspan.pl
|
|
|
10 --fasta $infile
|
|
|
11
|
|
|
12 #if $annotate_infile:
|
|
|
13 -annotate $annotate_infile
|
|
|
14 #end if
|
|
|
15
|
|
|
16 -shift $shift
|
|
|
17 $cue
|
|
|
18 $stack
|
|
|
19 #if $energy_opts.energy_opts_selector == 'energy':
|
|
|
20 -e $energy_opts.e
|
|
|
21 #elif $energy_opts.energy_opts_selector == 'rel_energy':
|
|
|
22 -c $energy_opts.c
|
|
|
23 #end if
|
|
|
24
|
|
|
25 #if $structure_sampling_opts.structure_sampling_opts_selector == 'on':
|
|
|
26 -i $structure_sampling_opts.i
|
|
|
27 -sample_len $structure_sampling_opts.sample_len
|
|
|
28 #end if
|
|
|
29
|
|
|
30 $q
|
|
|
31 -Tp $Tp
|
|
|
32 $seq_graph_win
|
|
|
33 $seq_graph_t
|
|
|
34 $seq_graph_alph
|
|
|
35 $abstr
|
|
|
36 $nostr
|
|
|
37 $vp
|
|
|
38 $ignore_header
|
|
|
39 -M $M
|
|
|
40 $u
|
|
|
41 $r
|
|
|
42
|
|
|
43 -stdout
|
|
|
44
|
|
|
45 > $output
|
|
|
46
|
|
|
47 </command>
|
|
|
48 <inputs>
|
|
|
49 <param name="ifile" type="data" format="fasta" label="Nucleotide Sequence in FASTA format"/>
|
|
|
50
|
|
|
51 <!-- todo repeat tag, normally only 3 windows are specified-->
|
|
|
52 <param name="wins" type="text" value="" label="A list of window sizes to use"
|
|
|
53 help="if none are given, then the entire sequence is taken with no windows. Each window > 1 required! Example: 50,100,200. (-wins)" />
|
|
|
54 <param name="t" type="text" value="3=0" label="The shape type (RNAshapes types from 1-5)"
|
|
|
55 help="With the list format, the shape level can be changed for different window length '4=100' means that shape level 4 is used from length 100nt (window length) The first given length has to be 0! Not continuous given levels are allowed! (-t)" />
|
|
|
56
|
|
|
57
|
|
|
58 <param name="shift" type="integer" value="1" label="The shift of the window, relative to the window size given in percent"
|
|
|
59 help="Example 20 % of a window 150 would result in a step size of 30 nt. It is a relative parameter, as you can give different window sizes. Default shift of 1 nt. (-shift)">
|
|
|
60 <validator type="in_range" min="0" max="100" />
|
|
|
61 </param>
|
|
|
62 <param name="cue" type="boolean" label="Crop unpaired ends" truevalue="-cue" falsevalue="" checked="false"
|
|
|
63 help="If you give this flag, then the unpaired ends of each single structure are ignored. E.g. the structure ...(((...))).. becomes just (((...))). (-cue)" />
|
|
|
64
|
|
|
65 <param name="stack" type="boolean" label="Adds stacking information to graphs" truevalue="-stack" falsevalue="" checked="false"
|
|
|
66 help="This adds an additional vertex (type P) for each pair of stacked base-pairs and four edges (type p) from each of the involved bases to the new vertex. (-stack)" />
|
|
|
67
|
|
|
68 <conditional name="energy_opts">
|
|
|
69 <param name="energy_opts_selector" type="select" label="Energy calculation mode">
|
|
|
70 <option value="rel_energy">Relative energy range, i.e. percentage (%) of MFE energy (RNAshapes)</option>
|
|
|
71 <option value="energy">Energy range in kcal/mol (RNAshapes)</option>
|
|
|
72 <option value="">None</option>
|
|
|
73 </param>
|
|
|
74 <when value="" />
|
|
|
75 <when value="energy">
|
|
|
76 <param name="e" type="float" value="5.0" label="Energy range in kcal/mol (RNAshapes)" help="-e">
|
|
|
77 <validator type="in_range" min="0.0" />
|
|
|
78 </param>
|
|
|
79 </when>
|
|
|
80 <when value="rel_energy">
|
|
|
81 <param name="c" type="float" value="10" label="Relative energy range, i.e. percentage (%) of MFE energy (RNAshapes)" help="(-c)">
|
|
|
82 <validator type="in_range" min="0.0" />
|
|
|
83 </param>
|
|
|
84 </when>
|
|
|
85 </conditional>
|
|
|
86 <param name="M" type="integer" value="0" label="Max number of shreps that should be taken per window" help="0 means selects all shreps. (-M)">
|
|
|
87 <validator type="in_range" min="0" />
|
|
|
88 </param>
|
|
|
89 <param name="u" type="boolean" label="Ignore unstable RNAshapes" truevalue="-u" falsevalue="" checked="false"
|
|
|
90 help="This option filters out closed structures with positive free energy. (-u)" />
|
|
|
91
|
|
|
92 <param name="r" type="boolean" label="Calculate structure probabilities for RNAshapes" help="(-r)" truevalue="-r" falsevalue="" checked="false" />
|
|
|
93
|
|
|
94
|
|
|
95 <conditional name="structure_sampling_opts">
|
|
|
96 <param name="structure_sampling_opts_selector" type="select" label="Structure Sampling">
|
|
|
97 <option value="on">On</option>
|
|
|
98 <option value="off">Off</option>
|
|
|
99 </param>
|
|
|
100 <when value="on">
|
|
|
101 <param name="i" type="integer" value="1" label="Number of sampling iterations" help="(-i)">
|
|
|
102 <validator type="in_range" min="1" />
|
|
|
103 </param>
|
|
|
104 <param name="smaple_len" type="integer" value="0" label="Sampling is only used for seqs/windows >= given length" help="Default 0 (sample all length). (-sample-len)">
|
|
|
105 <validator type="in_range" min="0" />
|
|
|
106 </param>
|
|
|
107 </when>
|
|
|
108 <when value="off" />
|
|
|
109 </conditional>
|
|
|
110
|
|
|
111 <param name="q" type="boolean" label="Turn on shape probabilities for RNAshapes, no sampling mode allowed" help="(-q)" truevalue="-q" falsevalue="" checked="false" />
|
|
|
112 <param name="Tp" type="float" value="0.001" label="Filter cutoff for shape probabilities, applied before -M filter!" help="(-Tp)">
|
|
|
113 <validator type="in_range" min="0.0" />
|
|
|
114 </param>
|
|
|
115
|
|
|
116 <param name="seq_graph_win" type="boolean" label="Add for each window a graph which contains no structure"
|
|
|
117 help="(-seq-graph-win)"
|
|
|
118 truevalue="-seq-graph-win" falsevalue="" checked="false" />
|
|
|
119 <param name="seq_graph_t" type="boolean" label="add for each 't #' a graph which contains no structure"
|
|
|
120 help="(-seq-graph-t)"
|
|
|
121 truevalue="-seq-graph-t" falsevalue="" checked="false" />
|
|
|
122 <param name="seq_graph_alph" type="boolean" label="Change the alphabet of unstructured graphs"
|
|
|
123 help="(-seq-graph-alph)"
|
|
|
124 truevalue="-seq-graph-alph" falsevalue="" checked="false" />
|
|
|
125
|
|
|
126 <param format="tabular" name="annotate_infile" type="data" optional="True" label="A file with annotations to be added as abstract graphs"
|
|
|
127 help="The format is has the following TAB-delimited columns: SEQID, START, END, NAMESPACE#LABEL. Labels with the same name-space and SEQID form connected components, which is a sequence of label vertices ordered by the START position in the sequence."/>
|
|
|
128
|
|
|
129 <param name="abstr" type="boolean" label="Add abstract structure graphs to the single shrep graph instances"
|
|
|
130 help="(-abstr)"
|
|
|
131 truevalue="-abstr" falsevalue="" checked="false" />
|
|
|
132 <param name="nostr" type="boolean" label="Calculate no structures, only add sequence information"
|
|
|
133 help="-seq-graph-win AND/OR -seq-graph-t are required (-nostr)"
|
|
|
134 truevalue="-nostr" falsevalue="" checked="false" />
|
|
|
135
|
|
|
136 <!--
|
|
|
137 -match-shape SHAPE
|
|
|
138 all seqs/windows will be constraint folded into that shape via
|
|
|
139 RNAshapes (if structure is given in another way this struct will be kept),
|
|
|
140 if this shape is not possible within given energy range, produce a
|
|
|
141 specific t graph with only one vertex 'X'. By this the instance
|
|
|
142 becomes very unsimilar to all other graphs (for knn)
|
|
|
143 -->
|
|
|
144 <param name="vp" type="boolean" label="Enable graph computation with viewpoints"
|
|
|
145 help="svmsgdnspdk will center on those nucleotides that are given via capital letters and ignore those given as lowercase letters (-vp)"
|
|
|
146 truevalue="-vp" falsevalue="" checked="false" />
|
|
|
147
|
|
|
148 <param name="ignore_header" type="boolean" label="Don't write fasta id part after first space to gspan" help="(-ignore-header)" truevalue="-ignore-header" falsevalue="" checked="false" />
|
|
|
149
|
|
|
150 </inputs>
|
|
|
151 <outputs>
|
|
|
152 <data format="gspan" name="output" label="${tool.name} on ${on_string}" />
|
|
|
153 </outputs>
|
|
|
154 <tests>
|
|
|
155 </tests>
|
|
|
156 <help>
|
|
|
157
|
|
|
158 **What it does**
|
|
|
159
|
|
|
160
|
|
|
161
|
|
|
162 **References**
|
|
|
163
|
|
|
164 Sita et al...
|
|
|
165
|
|
|
166 </help>
|
|
|
167 </tool>
|
