Mercurial > repos > genouest > helixer

comparison helixer.xml @ 2:7c1dc010a819 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/genouest/galaxy-tools/tree/master/tools/helixer commit 85a0295d96722d39bc249fa3be1145ce2b9801e3
author genouest
date Mon, 26 Feb 2024 09:47:53 +0000
parents 7bc75dd0f782
children e3846dc36c4d
comparison
equal deleted inserted replaced
1:7bc75dd0f782 2:7c1dc010a819
1 <?xml version="1.0"?> 1 <?xml version="1.0"?>
2 <tool id="helixer" name="Helixer" version="@TOOL_VERSION@" profile="21.05"> 2 <tool id="helixer" name="Helixer" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="21.05">
3 <description>gene calling</description> 3 <description>gene calling</description>
4 <macros> 4 <macros>
5 <import>macros.xml</import> 5 <import>macros.xml</import>
6 </macros> 6 </macros>
7 7
13 13
14 /usr/local/bin/fetch_helixer_models.py && 14 /usr/local/bin/fetch_helixer_models.py &&
15 Helixer.py 15 Helixer.py
16 --fasta-path '$input' 16 --fasta-path '$input'
17 --species '$species' 17 --species '$species'
18 --lineage $lineage.lineages 18 --lineage '$lineages'
19 --gff-output-path '$output' 19 --gff-output-path '$output'
20 20
21 --temporary-dir ./ 21 --temporary-dir ./
22 22
23 --subsequence-length $lineage.subsequence_length 23 #set default_subsequence_length = {"fungi": 21384, "land_plant": 106920, "invertebrate": 213840, "vertebrate": 213840}
24 #if str($lineage.option_overlap.use_overlap) == "true": 24 #set subsequence_len = default_subsequence_length.get(str($lineages))
25 --overlap-offset $lineage.option_overlap.overlap_offset 25 #set default_overlap_offset = {"fungi": 10692, "land_plant": 53460, "invertebrate": 106920, "vertebrate": 106920}
26 --overlap-core-length $lineage.option_overlap.overlap_core_length 26 #set overlap_off = default_overlap_offset.get(str($lineages))
27 #set default_overlap_core_length = {"fungi": 16038, "land_plant": 80190, "invertebrate": 160380, "vertebrate": 160380}
28 #set overlap_core_len = default_overlap_core_length.get(str($lineages))
29
30 #if str($subsequence_length) == "":
31 --subsequence-length '$subsequence_len'
27 #else: 32 #else:
28 --no-overlap 33 --subsequence-length '$subsequence_length'
29 #end if 34 #end if
35
36 #if str($option_overlap.use_overlap) == "true":
37 #if str($option_overlap.overlap_offset) == "":
38 --overlap-offset '$overlap_off'
39 #else:
40 --overlap-offset '$overlap_offset'
41 #end if
42
43 #if str($option_overlap.overlap_core_length) == "":
44 --overlap-core-length '$overlap_core_len'
45 #else:
46 --overlap-core-length '$overlap_core_length'
47 #end if
48 #end if
49
30 --batch-size $size 50 --batch-size $size
31 --window-size $post_processing.window_size 51 --window-size $post_processing.window_size
32 --min-coding-length $post_processing.min_coding_length 52 --min-coding-length $post_processing.min_coding_length
33 --edge-threshold $post_processing.edge_threshold 53 --edge-threshold $post_processing.edge_threshold
34 --peak-threshold $post_processing.peak_threshold 54 --peak-threshold $post_processing.peak_threshold
55
35 ]]></command> 56 ]]></command>
36 57
37 <inputs> 58 <inputs>
38 <param argument="--fasta-path" name="input" type="data" format="fasta,fasta.gz" label="Genomic sequence"></param> 59 <param argument="--fasta-path" name="input" type="data" format="fasta,fasta.gz" label="Genomic sequence"></param>
39 <conditional name="lineage"> 60
40 <param argument="--lineage" name="lineages" type="select" label="Available lineages" help="Choose the model to use for the annotation"> 61 <param argument="--lineage" name="lineages" type="select" label="Available lineages" help="Choose the model to use for the annotation">
41 <option value="land_plant">land plant</option> 62 <option value="land_plant">land plant</option>
42 <option value="vertebrate">vertebrate</option> 63 <option value="vertebrate">vertebrate</option>
43 <option value="invertebrate">invertebrate</option> 64 <option value="invertebrate">invertebrate</option>
44 <option value="fungi">fungi</option> 65 <option value="fungi">fungi</option>
45 </param> 66 </param>
46 <when value="land_plant"> 67
47 <expand macro="subseq" length="106920" offset="53460" offsetlen="80190" />
48 </when>
49 <when value="vertebrate">
50 <expand macro="subseq" length="213840" offset="106920" offsetlen="160380" />
51 </when>
52 <when value="invertebrate">
53 <expand macro="subseq" length="213840" offset="106920" offsetlen="160380" />
54 </when>
55 <when value="fungi">
56 <expand macro="subseq" length="21384" offset="10692" offsetlen="16038" />
57 </when>
58 </conditional>
59 <param argument="--species" type="text" optional="true" label="Species name"> 68 <param argument="--species" type="text" optional="true" label="Species name">
60 <sanitizer invalid_char=""> 69 <sanitizer invalid_char="">
61 <valid initial="string.letters,string.digits"> 70 <valid initial="string.letters,string.digits">
62 <add value="_" /> 71 <add value="_" />
63 </valid> 72 </valid>
69 <param argument="--window-size" type="integer" min="0" value="100" label="Window size" help="This determines the number of bases averaged during the sliding window approach"/> 78 <param argument="--window-size" type="integer" min="0" value="100" label="Window size" help="This determines the number of bases averaged during the sliding window approach"/>
70 <param argument="--edge-threshold" type="float" min="0" max="1" value="0.1" label="Edge threshold" help="This threshold specifies the genic score which defines the start / end boundaries of each candidate region"/> 79 <param argument="--edge-threshold" type="float" min="0" max="1" value="0.1" label="Edge threshold" help="This threshold specifies the genic score which defines the start / end boundaries of each candidate region"/>
71 <param argument="--peak-threshold" type="float" min="0" max="1" value="0.8" label="Peak threshold" help="This threshold specifies the minimum peak genic score required to accept the candidate region"/> 80 <param argument="--peak-threshold" type="float" min="0" max="1" value="0.8" label="Peak threshold" help="This threshold specifies the minimum peak genic score required to accept the candidate region"/>
72 <param argument="--min-coding-length" type="integer" min="0" value="100" label="Minimum coding length"/> 81 <param argument="--min-coding-length" type="integer" min="0" value="100" label="Minimum coding length"/>
73 </section> 82 </section>
83
84 <param name="subsequence_length" type="text" label="Subsequence length: how much of the genome the Neural Network can see at once" help="If you do not want to keep the default value, please enter the new value. Default values are 21384 for fungi, 106920 for land plant, and 213840 for vertebrates and invertebrates"/>
85
86 <conditional name="option_overlap">
87 <param name="use_overlap" type="select" label="Enable overlapping step after predictions" help="This step combines predictions made on each subsequences, to improve quality near start and end of subsequences.">
88 <option value="true" selected="true">Yes</option>
89 <option value="false">No</option>
90 </param>
91 <when value="true">
92 <param name="overlap_offset" type="text" label="Overlap offset: Smaller values may lead to better predictions but will take longer" help="If you do not want to keep the default value, please enter the new value. Default values are 10692 for fungi, 53460 for land plant, and 106920 for vertebrates and invertebrates."/>
93 <param name="overlap_core_length" type="text" label="Overlap core length: Predicted subsequences will be cut to this length to increase prediction quality. Smaller values may lead to better predictions but will take longer" help="If you do not want to keep the default value, please enter the new value. Default values are 16038 for fungi, 80190 for land plant, and 160380 for vertebrates and invertebrates."/>
94 </when>
95 <when value="false"/>
96 </conditional>
74 </inputs> 97 </inputs>
75 98
76 <outputs> 99 <outputs>
77 <data name="output" format="gff3"> 100 <data name="output" format="gff3">
78 </data> 101 </data>
79 </outputs> 102 </outputs>
80 <tests> 103 <tests>
81 <test expect_num_outputs="1"> 104 <test expect_num_outputs="1">
82 <!-- Test for species and land_plant--> 105 <!-- Test for species and land_plant-->
83 <param name="input" value="sequence.fasta"/> 106 <param name="input" value="sequence.fasta"/>
107 <param name="lineages" value="land_plant"/>
84 <param name="species" value="Arabidopsis"/> 108 <param name="species" value="Arabidopsis"/>
85 <conditional name="lineage">
86 <param name="lineages" value="land_plant"/>
87 </conditional>
88 <param name="size" value="8"/> 109 <param name="size" value="8"/>
110 <param name="subsequence_length" value=""/>
111 <section name="post_processing">
112 <param name="window_size" value="100"/>
113 <param name="edge_threshold" value="0.1"/>
114 <param name="peak_threshold" value="0.8"/>
115 <param name="min_coding_length" value="100"/>
116 </section>
117 <conditional name="option_overlap">
118 <param name="use_overlap" value="true"/>
119 <param name="overlap_offset" value=""/>
120 <param name="overlap_core_length" value=""/>
121 </conditional>
89 <output name="output" value="ouput_species.gff3" ftype="gff3" compare="sim_size" delta="100"/> 122 <output name="output" value="ouput_species.gff3" ftype="gff3" compare="sim_size" delta="100"/>
90 </test> 123 </test>
124
91 <test expect_num_outputs="1"> 125 <test expect_num_outputs="1">
92 <!-- Test for vertebrates--> 126 <!-- Test for vertebrates-->
93 <param name="input" value="sequence.fasta"/> 127 <param name="input" value="sequence.fasta"/>
94 <conditional name="lineage"> 128 <param name="lineages" value="vertebrate"/>
95 <param name="lineages" value="vertebrate"/>
96 </conditional>
97 <param name="size" value="8"/> 129 <param name="size" value="8"/>
130 <param name="size" value="8"/>
131 <param name="subsequence_length" value=""/>
132 <section name="post_processing">
133 <param name="window_size" value="100"/>
134 <param name="edge_threshold" value="0.1"/>
135 <param name="peak_threshold" value="0.8"/>
136 <param name="min_coding_length" value="100"/>
137 </section>
138 <conditional name="option_overlap">
139 <param name="use_overlap" value="true"/>
140 <param name="overlap_offset" value=""/>
141 <param name="overlap_core_length" value=""/>
142 </conditional>
98 <output name="output" value="vertebrate.gff3" ftype="gff3" lines_diff="2"/> 143 <output name="output" value="vertebrate.gff3" ftype="gff3" lines_diff="2"/>
99 </test>
100 <test expect_num_outputs="1">
101 <!-- Test for invertebrates-->
102 <param name="input" value="sequence.fasta"/>
103 <conditional name="lineage">
104 <param name="lineages" value="invertebrate"/>
105 </conditional>
106 <param name="size" value="8"/>
107 <output name="output" value="invertebrate.gff3" ftype="gff3" lines_diff="2"/>
108 </test>
109 <test expect_num_outputs="1">
110 <!-- Test for fungi-->
111 <param name="input" value="sequence.fasta"/>
112 <conditional name="lineage">
113 <param name="lineages" value="fungi"/>
114 </conditional>
115 <param name="size" value="8"/>
116 <output name="output" value="fungi.gff3" ftype="gff3" lines_diff="2"/>
117 </test> 144 </test>
118 </tests> 145 </tests>
119 146
120 <help><![CDATA[ 147 <help><![CDATA[
121 Helixer_: Gene calling with Deep Neural Networks. 148 Helixer_: Gene calling with Deep Neural Networks.