|
31
|
1 <tool id="DInterproscan" name="Interproscan" version="1.0.0">
|
|
|
2 <description>Interproscan annotation for SAPP</description>
|
|
|
3 <requirements>
|
|
|
4 <container type="docker">jjkoehorst/sappdocker:INTERPROSCAN</container>
|
|
|
5 </requirements>
|
|
|
6 <command interpreter="docker">java -jar /interproscan/target/interproscanRDF-0.0.1-SNAPSHOT-jar-with-dependencies.jar
|
|
|
7 '-input' '$input' '-format' 'TURTLE'
|
|
|
8 '-applications' '$appl'
|
|
|
9 '-output'
|
|
|
10 '$outfile' -v '$version' '$disable'
|
|
|
11 </command>
|
|
|
12 <inputs>
|
|
|
13 <param format="ttl" label="genome rdf file with orf prediction" name="input" type="data"/>
|
|
|
14 <param display="checkboxes" help="Select your programm." label="Applications to run" multiple="True" name="appl" type="select">
|
|
|
15 <option selected="true" value="TIGRFAM">TIGRFAM: protein families
|
|
|
16 based on Hidden Markov Models or HMMs
|
|
|
17 </option>
|
|
|
18 <option selected="false" value="PIRSF">PIRSF: non-overlapping
|
|
|
19 clustering of UniProtKB sequences into a hierarchical order
|
|
|
20 (evolutionary relationships)
|
|
|
21 </option>
|
|
|
22 <option selected="true" value="ProDom">ProDom: set of protein domain
|
|
|
23 families generated from the UniProtKB
|
|
|
24 </option>
|
|
|
25 <option selected="true" value="SMART">SMART: identification and
|
|
|
26 analysis of domain architectures based on Hidden Markov Models or
|
|
|
27 HMMs
|
|
|
28 </option>
|
|
|
29 <option selected="false" value="PrositeProfiles">PROSITE Profiles:
|
|
|
30 protein domains, families and functional sites as well as associated
|
|
|
31 profiles to identify them
|
|
|
32 </option>
|
|
|
33 <option selected="true" value="PrositePatterns">PROSITE Pattern:
|
|
|
34 protein domains, families and functional sites as well as associated
|
|
|
35 patterns to identify them
|
|
|
36 </option>
|
|
|
37 <option selected="false" value="HAMAP">HAMAP: High-quality Automated
|
|
|
38 Annotation of Microbial Proteomes
|
|
|
39 </option>
|
|
|
40 <option selected="true" value="PfamA">PfamA: protein families, each
|
|
|
41 represented by multiple sequence alignments and hidden Markov models
|
|
|
42 </option>
|
|
|
43 <option selected="true" value="PRINTS">PRINTS: group of conserved
|
|
|
44 motifs (fingerprints) used to characterise a protein family
|
|
|
45 </option>
|
|
|
46 <option selected="true" value="SuperFamily">SUPERFAMILY: database of
|
|
|
47 structural and functional annotation
|
|
|
48 </option>
|
|
|
49 <option selected="true" value="Coils">Coils: Prediction of Coiled
|
|
|
50 Coil Regions in Proteins
|
|
|
51 </option>
|
|
|
52 <option selected="true" value="Gene3d">Gene3d: Structural assignment
|
|
|
53 for whole genes and genomes using the CATH domain structure database
|
|
|
54 </option>
|
|
|
55 </param>
|
|
|
56 <param label="Version selection" name="version" type="select">
|
|
|
57 <option value="interproscan-5.17-56.0">interproscan-5.17-56.0</option>
|
|
|
58 </param>
|
|
|
59 <param checked="false" falsevalue="-disableprecalc" help="You need to setup your own lookup server as the EBI version can differ. Look at interproscan configuration file for more info" label="Perform lookup of InterPro at defined server address" name="disable" truevalue="" type="boolean"/>
|
|
|
60 </inputs>
|
|
|
61 <outputs>
|
|
|
62 <data format="ttl" label="IPR: ${input.name}" name="outfile"/>
|
|
|
63 </outputs>
|
|
|
64 <help>Interproscan annotation suite. Select your RDF genome with
|
|
|
65 protein annotation.
|
|
|
66 This can be either from a converted GenBank/EMBL
|
|
|
67 file or from a
|
|
|
68 Prodigal prediction.
|
|
|
69 The output will be an RDF file with
|
|
|
70 protein domain annotation from
|
|
|
71 InterPro.
|
|
|
72 </help>
|
|
|
73 <citations>
|
|
|
74 <citation type="bibtex">@article{Mitchell26112014,
|
|
|
75 author = {Mitchell,
|
|
|
76 Alex and Chang, Hsin-Yu and Daugherty, Louise and
|
|
|
77 Fraser, Matthew and
|
|
|
78 Hunter, Sarah and Lopez, Rodrigo and McAnulla,
|
|
|
79 Craig and McMenamin,
|
|
|
80 Conor and Nuka, Gift and Pesseat, Sebastien and
|
|
|
81 Sangrador-Vegas, Amaia
|
|
|
82 and Scheremetjew, Maxim and Rato, Claudia and
|
|
|
83 Yong, Siew-Yit and
|
|
|
84 Bateman, Alex and Punta, Marco and Attwood, Teresa
|
|
|
85 K. and Sigrist,
|
|
|
86 Christian J.A. and Redaschi, Nicole and Rivoire,
|
|
|
87 Catherine and
|
|
|
88 Xenarios, Ioannis and Kahn, Daniel and Guyot, Dominique
|
|
|
89 and Bork, Peer
|
|
|
90 and Letunic, Ivica and Gough, Julian and Oates, Matt
|
|
|
91 and Haft, Daniel
|
|
|
92 and Huang, Hongzhan and Natale, Darren A. and Wu,
|
|
|
93 Cathy H. and Orengo,
|
|
|
94 Christine and Sillitoe, Ian and Mi, Huaiyu and
|
|
|
95 Thomas, Paul D. and
|
|
|
96 Finn, Robert D.},
|
|
|
97 title = {The InterPro protein families database: the
|
|
|
98 classification
|
|
|
99 resource after 15 years},
|
|
|
100 year = {2014},
|
|
|
101 doi =
|
|
|
102 {10.1093/nar/gku1243},
|
|
|
103 abstract ={The InterPro database
|
|
|
104 (http://www.ebi.ac.uk/interpro/) is a freely
|
|
|
105 available resource that
|
|
|
106 can be used to classify sequences into
|
|
|
107 protein families and to predict
|
|
|
108 the presence of important domains and
|
|
|
109 sites. Central to the InterPro
|
|
|
110 database are predictive models, known
|
|
|
111 as signatures, from a range of
|
|
|
112 different protein family databases
|
|
|
113 that have different biological
|
|
|
114 focuses and use different
|
|
|
115 methodological approaches to classify
|
|
|
116 protein families and domains.
|
|
|
117 InterPro integrates these signatures,
|
|
|
118 capitalizing on the respective
|
|
|
119 strengths of the individual databases,
|
|
|
120 to produce a powerful protein
|
|
|
121 classification resource. Here, we report
|
|
|
122 on the status of InterPro as
|
|
|
123 it enters its 15th year of operation, and
|
|
|
124 give an overview of new
|
|
|
125 developments with the database and its
|
|
|
126 associated Web interfaces and
|
|
|
127 software. In particular, the new domain
|
|
|
128 architecture search tool is
|
|
|
129 described and the process of mapping of
|
|
|
130 Gene Ontology terms to
|
|
|
131 InterPro is outlined. We also discuss the
|
|
|
132 challenges faced by the
|
|
|
133 resource given the explosive growth in
|
|
|
134 sequence data in recent years.
|
|
|
135 InterPro (version 48.0) contains 36 766
|
|
|
136 member database signatures
|
|
|
137 integrated into 26 238 InterPro entries, an
|
|
|
138 increase of over 3993
|
|
|
139 entries (5081 signatures), since 2012.},
|
|
|
140 URL =
|
|
|
141 {http://nar.oxfordjournals.org/content/early/2014/11/26/nar.gku1243.abstract},
|
|
|
142 eprint =
|
|
|
143 {http://nar.oxfordjournals.org/content/early/2014/11/26/nar.gku1243.full.pdf+html},
|
|
|
144 journal = {Nucleic Acids Research}
|
|
|
145 }
|
|
|
146 </citation>
|
|
|
147 </citations>
|
|
|
148 </tool> |
