|
19
|
1 <?xml version="1.0"?>
|
|
|
2 <tool name="BloomTree Manager - Create" id="btman_create" version="1.0.0">
|
|
|
3 <description>a Sequence Bloom Tree</description>
|
|
|
4 <macros>
|
|
|
5 <import>macros.xml</import>
|
|
|
6 </macros>
|
|
|
7 <expand macro="requirements" />
|
|
|
8 <command detect_errors="exit_code">
|
|
|
9 <![CDATA[
|
|
|
10 python '$__tool_directory__/create.py'
|
|
|
11
|
|
|
12 #set formats = ''
|
|
|
13 #set filepaths = ''
|
|
|
14 #set filenames = ''
|
|
|
15 #set compressed = ''
|
|
|
16 #set minab = ''
|
|
|
17 #set qthres = ''
|
|
|
18 #for $i, $exp in enumerate( $experiments ):
|
|
|
19 #set formats += str( $exp.conditional_format.format ) + '|'
|
|
|
20 #if $exp.conditional_format.format == 'accessions':
|
|
|
21 #set filepaths += str( $exp.conditional_format.accession_numbers ) + '|'
|
|
|
22 #set filenames += str( $exp.conditional_format.accession_numbers.name ) + '|'
|
|
|
23 #set compressed += '0|'
|
|
|
24 #else:
|
|
|
25 #if $exp.conditional_format.format == 'fasta':
|
|
|
26 #set compressed += str( $exp.conditional_format.conditional_fasta_compressed.fasta_compressed ) + '|'
|
|
|
27 #if $exp.conditional_format.conditional_fasta_compressed.fasta_compressed == 0:
|
|
|
28 #set filepaths += ','.join( [ str( $f ) for $f in $exp.conditional_format.conditional_fasta_compressed.fastafiles ] ) + '|'
|
|
|
29 #set filenames += ','.join( [ str( $f.name ) for $f in $exp.conditional_format.conditional_fasta_compressed.fastafiles ] ) + '|'
|
|
|
30 #else:
|
|
|
31 #set filepaths += ','.join( [ str( $f ) for $f in $exp.conditional_format.conditional_fasta_compressed.fastagzfiles ] ) + '|'
|
|
|
32 #set filenames += ','.join( [ str( $f.name ) for $f in $exp.conditional_format.conditional_fasta_compressed.fastagzfiles ] ) + '|'
|
|
|
33 #end if
|
|
|
34 #elif $exp.conditional_format.format == 'fastq':
|
|
|
35 #set compressed += str( $exp.conditional_format.conditional_fastq_compressed.fastq_compressed ) + '|'
|
|
|
36 #if $exp.conditional_format.conditional_fastq_compressed.fastq_compressed == 0:
|
|
|
37 #set filepaths += ','.join( [ str( $f ) for $f in $exp.conditional_format.conditional_fastq_compressed.fastqfiles ] ) + '|'
|
|
|
38 #set filenames += ','.join( [ str( $f.name ) for $f in $exp.conditional_format.conditional_fastq_compressed.fastqfiles ] ) + '|'
|
|
|
39 #else:
|
|
|
40 #set filepaths += ','.join( [ str( $f ) for $f in $exp.conditional_format.conditional_fastq_compressed.fastqgzfiles ] ) + '|'
|
|
|
41 #set filenames += ','.join( [ str( $f.name ) for $f in $exp.conditional_format.conditional_fastq_compressed.fastqgzfiles ] ) + '|'
|
|
|
42 #end if
|
|
|
43 #end if
|
|
|
44 #end if
|
|
|
45 #set minab += str( $exp.min_abundance ) + '|'
|
|
|
46 #if $exp.conditional_quality.quality_control == '1':
|
|
|
47 #set qthres += str( $exp.conditional_quality.quality_threshold ) + '|'
|
|
|
48 #else:
|
|
|
49 #set qthres += '-1.0|'
|
|
|
50 #end if
|
|
|
51 #end for
|
|
|
52 #set klen = $kmer_len
|
|
|
53 #set bfsize = -1
|
|
|
54 #if $bloomsize_condition.bloomsize_control == '0':
|
|
|
55 #set bfsize = $bloomsize_condition.bloom_filter_size
|
|
|
56 #end if
|
|
|
57
|
|
|
58 --formats '${formats}'
|
|
|
59 --filepaths '${filepaths}'
|
|
|
60 --filenames '${filenames}'
|
|
|
61 --compressed '${compressed}'
|
|
|
62 --minabundances '${minab}'
|
|
|
63 --qualitythresholds '${qthres}'
|
|
|
64
|
|
|
65 --klen ${klen}
|
|
|
66 --bfsize ${bfsize}
|
|
|
67
|
|
|
68 --outfile '${resulttxt}'
|
|
|
69 --outdir 'sbt'
|
|
|
70 --tooldir '$__tool_directory__'
|
|
|
71 ]]>
|
|
|
72 </command>
|
|
|
73 <inputs>
|
|
|
74 <repeat name="experiments" title="Select a list of experiments" help="Select a set of experiments on which the Sequence Bloom Tree will be built." min="1">
|
|
|
75 <conditional name="conditional_format">
|
|
|
76 <param name="format" type="select" label="Select the experiment format" help="FASTA and FASTQ are the supported formats">
|
|
|
77 <option value="fasta">FASTA Experiments</option>
|
|
|
78 <option value="fastq">FASTQ Experiments</option>
|
|
|
79 <option value="accessions">SRA Accession Numbers</option>
|
|
|
80 </param>
|
|
|
81 <when value="fasta">
|
|
|
82 <conditional name="conditional_fasta_compressed">
|
|
|
83 <param name="fasta_compressed" type="boolean" checked="true" truevalue="1" falsevalue="0" label="Are your experiments compressed?" />
|
|
|
84 <when value="0">
|
|
|
85 <param format="fasta" name="fastafiles" multiple="true" type="data" label="Select one or more FASTA experiments" />
|
|
|
86 </when>
|
|
|
87 <when value="1">
|
|
|
88 <param format="fastagz" name="fastagzfiles" multiple="true" type="data" label="Select one or more FASTA .gz experiments" />
|
|
|
89 </when>
|
|
|
90 </conditional>
|
|
|
91 </when>
|
|
|
92 <when value="fastq">
|
|
|
93 <conditional name="conditional_fastq_compressed">
|
|
|
94 <param name="fastq_compressed" type="boolean" checked="true" truevalue="1" falsevalue="0" label="Are youe experiments compressed?" />
|
|
|
95 <when value="0">
|
|
|
96 <param format="fastq" name="fastqfiles" multiple="true" type="data" label="Select one or more FASTQ experiments" />
|
|
|
97 </when>
|
|
|
98 <when value="1">
|
|
|
99 <param format="fastqgz" name="fastqgzfiles" multiple="true" type="data" label="Select one or more FASTQ .gz experiments" />
|
|
|
100 </when>
|
|
|
101 </conditional>
|
|
|
102 </when>
|
|
|
103 <when value="accessions">
|
|
|
104 <param name="accession_numbers" type="data" format="tabular" label="Select a list of SRA Accession Numbers" help="Select a tabular file with a list of accession numbers in the first column." />
|
|
|
105 </when>
|
|
|
106 </conditional>
|
|
|
107
|
|
|
108 <param name="min_abundance" type="integer" value="2" min="0" label="Insert a Bloom filter minimum abundance" help="This value is the minimum abundance cutoff for the creation of the Bloom filter." />
|
|
|
109
|
|
|
110 <conditional name="conditional_quality">
|
|
|
111 <param name="quality_control" type="boolean" checked="false" truevalue="1" falsevalue="0" label="Apply a quality control procedure" />
|
|
|
112 <when value="1">
|
|
|
113 <param name="quality_threshold" size="1" type="float" value="0.8" min="0.0" max="1.0" label="Quality threshold" help="If the number of sequences flagged as poor quality on the total number of sequences in a file is less than this threshold, the whole experiment will be excluded." />
|
|
|
114 </when>
|
|
|
115 </conditional>
|
|
|
116 </repeat>
|
|
|
117
|
|
|
118 <param name="kmer_len" type="integer" value="21" min="0" label="K-mer length" />
|
|
|
119
|
|
|
120 <conditional name="bloomsize_condition">
|
|
|
121 <param name="bloomsize_control" type="boolean" checked="true" truevalue="1" falsevalue="0" label="Automatically estimate the Bloom filter size" />
|
|
|
122 <when value="0">
|
|
|
123 <param name="bloom_filter_size" size="1" type="integer" value="1" min="1" label="Bloom Filter size" help="Disable this field to let the tool estimate an appropriate Bloom filter size." />
|
|
|
124 </when>
|
|
|
125 </conditional>
|
|
|
126 </inputs>
|
|
|
127 <outputs>
|
|
|
128 <collection name="list_output" type="list" label="${tool.name} SBT Collection">
|
|
|
129 <discover_datasets pattern="(?P<identifier_0>.*(?=\.)).(?P<ext>[^\.]*$)" ext="auto" directory="sbt" />
|
|
|
130 </collection>
|
|
|
131 <data format="txt" name="resulttxt" label="${tool.name} SBT: Result" from_work_dir="sbtres.txt" />
|
|
|
132 </outputs>
|
|
|
133
|
|
|
134 <help><![CDATA[
|
|
|
135 This tool allows to create Sequence Bloom Trees starting from a set of FASTA or FASTQ files.
|
|
|
136 It also allows to control the quality of the input dataset and exclude the files that do not reach a specified quality level.
|
|
|
137
|
|
|
138 -----
|
|
|
139
|
|
|
140 **Input file**
|
|
|
141
|
|
|
142 The input of this tool is a set of FASTA or FASTQ experiments, additionally to a set of SRA accession numbers.
|
|
|
143 For each of the selected experiments, the minimum abundance for the corresponding Bloom filter is required.
|
|
|
144 Additionally, a quality control procedure could be applied to guarantee that the quality of every experiment always exceed a
|
|
|
145 specified treshold. Otherwise, experiments with low quality level will be discarded.
|
|
|
146
|
|
|
147 The k-mer length must also be specified, additionally to the Bloom filter size. This last field is optional and it will be
|
|
|
148 automatically estimated if not provided.
|
|
|
149
|
|
|
150 -----
|
|
|
151
|
|
|
152 **Output**
|
|
|
153
|
|
|
154 This tool returns a collection containing the Sequence Bloom Tree nodes and a file representing the organization of the tree.
|
|
|
155
|
|
|
156 Take a look at the Query tool documentation for a detailed description about how
|
|
|
157 to query a Sequence Bloom Tree.
|
|
|
158
|
|
|
159 -----
|
|
|
160
|
|
|
161 .. class:: infomark
|
|
|
162
|
|
|
163 **Notes**
|
|
|
164
|
|
|
165 This Galaxy tool has been developed by Fabio Cumbo.
|
|
|
166
|
|
|
167 Please visit this GithHub_repository_ for more information about the BloomTree Manager
|
|
|
168
|
|
|
169 .. _GithHub_repository: https://github.com/fabio-cumbo/bloomtree-manager
|
|
|
170 ]]></help>
|
|
|
171
|
|
|
172 <expand macro="citations" />
|
|
|
173 </tool>
|
