|
104
|
1 <tool id="ctb_osra" name="Molecule recognition" version="0.3">
|
|
|
2 <description>in Images or PDF documents (OSRA)</description>
|
|
|
3 <requirements>
|
|
|
4 <requirement type="package" version="2.0.0">osra</requirement>
|
|
|
5 <requirement type="package" version="2.3.2">openbabel</requirement>
|
|
|
6 <requirement type="package" version="1.3.18">graphicsmagick</requirement>
|
|
|
7 </requirements>
|
|
|
8 <command interpreter='python'>
|
|
|
9 ## OSRA_DATA_FILES is set during the toolshed Installation
|
|
|
10 ## if it is not set, use the standard configuration and hope the best
|
|
|
11 osra.py -f $oformat $infile
|
|
|
12 -l \$OSRA_DATA_FILES/spelling.txt -a \$OSRA_DATA_FILES/superatom.txt
|
|
|
13
|
|
|
14 ## further additions of OSRA parameter should go after -l and -a
|
|
|
15 ## because -l and -a can be removed by the python wrapper
|
|
|
16
|
|
|
17 $confidence
|
|
|
18 $adaptive
|
|
|
19 $thinning
|
|
|
20
|
|
|
21 > $outfile
|
|
|
22 </command>
|
|
|
23 <inputs>
|
|
|
24 <param name="infile" type="data" format="png,pdf" label="Image or PDF with molecules"/>
|
|
|
25 <param name="oformat" type="select" label="Output molecule format">
|
|
|
26 <option value="can">SMILES</option>
|
|
|
27 <option value="sdf">SDF</option>
|
|
|
28 </param>
|
|
|
29 <param name="confidence" type="boolean" label="Print out confidence estimate (-p)" truevalue="-p" falsevalue="" checked="true" />
|
|
|
30 <param name="adaptive" type="boolean" label="Adaptive thresholding pre-processing, useful for low light/low contrast images (-i)" truevalue="-i" falsevalue="" checked="false" />
|
|
|
31 <param name="thinning" type="boolean" label="Additional thinning/scaling down of low quality documents (-j)" truevalue="-j" falsevalue="" checked="false" />
|
|
|
32
|
|
|
33 </inputs>
|
|
|
34 <outputs>
|
|
|
35 <data name="outfile" type="data" format="sdf">
|
|
|
36 <change_format>
|
|
|
37 <when input="oformat" value="can" format="smi"/>
|
|
|
38 </change_format>
|
|
|
39 </data>
|
|
|
40 </outputs>
|
|
|
41 <tests>
|
|
|
42 <test>
|
|
|
43 <param name="infile" ftype="png" value="CID_2244.png"/>
|
|
|
44 <param name="oformat" value="sdf"/>
|
|
|
45 <output name="outfile" ftype="sdf" file="osra_on_CID2244.sdf"/>
|
|
|
46 </test>
|
|
|
47 <test>
|
|
|
48 <param name="infile" ftype="png" value="2008001635_153_chem.png"/>
|
|
|
49 <param name="oformat" value="can"/>
|
|
|
50 <output name="outfile" ftype="sdf" file="2008001635_153_chem.smi"/>
|
|
|
51 </test>
|
|
|
52
|
|
|
53 </tests>
|
|
|
54 <help>
|
|
|
55
|
|
|
56 .. class:: infomark
|
|
|
57
|
|
|
58 **What this tool does**
|
|
|
59
|
|
|
60 OSRA_ (Optical Structure Recognition Application) is a utility designed to convert graphical representations of chemical structures into SMILES or SDF. It generates the SMILES or SDF representation of any molecular structure image within a document which is parseable by GraphicMagick.
|
|
|
61
|
|
|
62 .. _OSRA: http://cactus.nci.nih.gov/osra/
|
|
|
63
|
|
|
64 -----
|
|
|
65
|
|
|
66 .. class:: infomark
|
|
|
67
|
|
|
68 **Cite**
|
|
|
69
|
|
|
70 Igor V Filippov and Marc C Nicklaus - `Optical Structure Recognition Software To Recover Chemical Information: OSRA, An Open Source Solution`_
|
|
|
71
|
|
|
72 .. _`Optical Structure Recognition Software To Recover Chemical Information: OSRA, An Open Source Solution`: http://pubs.acs.org/doi/abs/10.1021/ci800067r
|
|
|
73 </help>
|
|
|
74 </tool>
|
