|
0
|
1 <tool id="kb_python" name="kb_python" version="@VERSION@+galaxy0">
|
|
|
2 <description>performs gene and feature quantification on single-cell sequencing data.</description>
|
|
|
3 <macros>
|
|
|
4 <import>macros.xml</import>
|
|
|
5 </macros>
|
|
|
6 <requirements>
|
|
|
7 <requirement type="package" version="0.25.1">kb-python</requirement>
|
|
|
8 </requirements>
|
|
|
9 <stdio>
|
|
|
10 <exit_code range="1:" />
|
|
|
11 <exit_code range=":-1" />
|
|
|
12 <regex match="Error:" />
|
|
|
13 <regex match="Exception:" />
|
|
|
14 <regex match="Exception :" />
|
|
|
15 </stdio>
|
|
|
16 <command detect_errors="exit_code"><![CDATA[
|
|
|
17 mkdir ./index
|
|
|
18 && mkdir ./kb_outs
|
|
|
19 #if $refTranscriptSource.TranscriptSource == "history":
|
|
|
20 && ln -s '${refTranscriptSource.h_index.index_file}' './index/kb_ref.idx'
|
|
|
21 && ln -s '${refTranscriptSource.h_index.t2g_file}' './index/t2g.txt'
|
|
|
22 #if $workflow == "lamanno" or $workflow == "nucleus":
|
|
|
23 && ln -s '${refTranscriptSource.h_index.history_lamanno.cdna_t2c}' './index/cdna_t2c.txt'
|
|
|
24 && ln -s '${refTranscriptSource.h_index.history_lamanno.intron_t2c}' './index/intron_t2c.txt'
|
|
|
25 #end if
|
|
|
26 #set $index_path = './index'
|
|
|
27 #else if $refTranscriptSource.TranscriptSource == "built":
|
|
|
28 && kb ref -i ./index/kb_ref.idx
|
|
|
29 -g ./index/t2g.txt
|
|
|
30 -f1 ./index/cdna.fa
|
|
|
31 --workflow $workflow
|
|
|
32 #if $workflow == "lamanno" or $workflow == "nucleus":
|
|
|
33 -f2 ./index/intron.fa
|
|
|
34 -c1 ./index/cdna_t2c.txt
|
|
|
35 -c2 ./index/intron_t2c.txt
|
|
|
36 --workflow $workflow
|
|
|
37 #end if
|
|
|
38 #if $workflow != "kite":
|
|
|
39 '${refTranscriptSource.s_index.genomic_fasta}'
|
|
|
40 '${refTranscriptSource.s_index.genomic_gtf}'
|
|
|
41 #else:
|
|
|
42 '${refTranscriptSource.s_index.kite_table}'
|
|
|
43 #end if
|
|
|
44 #set $index_path = './index'
|
|
|
45 #end if
|
|
|
46 && kb count
|
|
|
47 -t \${GALAXY_SLOTS:-1}
|
|
|
48 -i $index_path/kb_ref.idx
|
|
|
49 -g $index_path/t2g.txt
|
|
|
50 #if $workflow == "lamanno" or $workflow == "nucleus":
|
|
|
51 -c1 $index_path/cdna_t2c.txt
|
|
|
52 -c2 $index_path/intron_t2c.txt
|
|
|
53 #end if
|
|
|
54 -x $technology
|
|
|
55 #if $whitelist:
|
|
|
56 --whitelist '${optional.whitelist}'
|
|
|
57 #end if
|
|
|
58 #if $optional.multimap:
|
|
|
59 --mm
|
|
|
60 #end if
|
|
|
61 --workflow $workflow
|
|
|
62 #if $extra_dtype != "none":
|
|
|
63 $extra_dtype
|
|
|
64 #end if
|
|
|
65 -o ./kb_outs
|
|
|
66 --cellranger
|
|
|
67 $FASTQ1
|
|
|
68 $FASTQ2
|
|
|
69 ]]>
|
|
|
70 </command>
|
|
|
71 <inputs>
|
|
|
72 <param name="workflow" label="Select the workflow you want to use:" type="select" multiple="false" format="text" help="Type of workflow. Use `lamanno` for RNA velocity based on La Manno et al. 2018 logic.
|
|
|
73 Use `nucleus` for RNA velocity on single-nucleus RNA-seq reads. Use `kite` for feature barcoding. Use `kite:10xFB` for 10x Genomics Feature Barcoding technology.">
|
|
|
74 <option value="standard">standard</option>
|
|
|
75 <option value="lamanno">lamanno</option>
|
|
|
76 <option value="nucleus">nucleus</option>
|
|
|
77 <option value="kite">kite</option>
|
|
|
78 </param>
|
|
|
79 <expand macro="index"/>
|
|
|
80 <param name="technology" label="Select the scRNA-seq technology:" type="select" multiple="false" format="text" help="Choose the scRNA-seq technology used to generate the fastq data.">
|
|
|
81 <option value="10XV1">10XV1</option>
|
|
|
82 <option value="10XV2">10XV2</option>
|
|
|
83 <option value="10XV3">10XV3</option>
|
|
|
84 <option value="CELSEQ">CELSEQ</option>
|
|
|
85 <option value="CELSEQ2">CELSEQ2</option>
|
|
|
86 <option value="DROPSEQ">DROPSEQ</option>
|
|
|
87 <option value="INDROPSV1">INDROPSV1</option>
|
|
|
88 <option value="INDROPSV2">INDROPSV2</option>
|
|
|
89 <option value="INDROPSV3">INDROPSV3</option>
|
|
|
90 <option value="SCRUBSEQ">SCRUBSEQ</option>
|
|
|
91 <option value="SURECELL">SURECELL</option>
|
|
|
92 </param>
|
|
|
93 <param name="FASTQ1" label="Select the R1 fastq file:" type="data" format="fastqsanger.gz" multiple="false"/>
|
|
|
94 <param name="FASTQ2" label="Select the R2 fastq file:" type="data" format="fastqsanger.gz" multiple="false"/>
|
|
|
95 <param name="extra_dtype" type="select" label="Do you want any additional output data type beside CellRanger?">
|
|
|
96 <option value="none" selected = "true">No</option>
|
|
|
97 <option value="--loom">Loom</option>
|
|
|
98 <option value="--h5ad">H5ad</option>
|
|
|
99 </param>
|
|
|
100 <section name="optional" title="Optional commands" expanded="false">
|
|
|
101 <param name="whitelist" type="data" format="tsv,tabular" optional="true" label="Whitelist file" help="Whitelisted barcodes to correct to. If not provided and bustools supports the technology, a pre-packaged whitelist is used. If not, the bustools
|
|
|
102 whitelist command is used."/>
|
|
|
103 <param name="multimap" type="boolean" optional="true" label="Include multi pseudoaligned reads?" help="Do you want to include reads that pseudoalign to multiple genes?"/>
|
|
|
104 </section>
|
|
|
105 </inputs>
|
|
|
106 <outputs>
|
|
|
107 <data name="barcodes" label="cellranger barcodes" format="txt" from_work_dir="kb_outs/counts_unfiltered/cellranger/barcodes.tsv"/>
|
|
|
108 <data name="genes" label="cellranger genes" format="txt" from_work_dir="kb_outs/counts_unfiltered/cellranger/genes.tsv"/>
|
|
|
109 <data name="matrix" label="cellranger matrix" format="mtx" from_work_dir="kb_outs/counts_unfiltered/cellranger/matrix.mtx" />
|
|
|
110 <data name="inspect" label="inspect_report" format="json" from_work_dir="kb_outs/inspect.json"/>
|
|
|
111 <data name="runinfo" label="run info" format="json" from_work_dir="kb_outs/run_info.json"/>
|
|
|
112 <data name="kbinfo" label="kb info" format="json" from_work_dir="kb_outs/kb_info.json"/>
|
|
|
113 <data name="kb_ref" label="kb_ref.idx" format="kallisto.idx" from_work_dir="index/kb_ref.idx"/>
|
|
|
114 <data name="t2g" label="t2g.txt" format="txt" from_work_dir="index/t2g.txt"/>
|
|
|
115 <data name="cdna_t2c" label="-c1 cdna_t2c.txt" format="txt" from_work_dir="index/cdna_t2c.txt">
|
|
|
116 <filter>workflow == "lamanno"</filter>
|
|
|
117 </data>
|
|
|
118 <data name="intron_t2c" label="-c2 intron_t2c.txt" format="txt" from_work_dir="index/intron_t2c.txt">
|
|
|
119 <filter>workflow == "lamanno"</filter>
|
|
|
120 </data>
|
|
|
121 <data name="adata" label="adata.h5ad" format="h5ad" from_work_dir="kb_outs/counts_unfiltered/adata.h5ad">
|
|
|
122 <filter >extra_dtype == "--h5ad"</filter>
|
|
|
123 </data>
|
|
|
124 <data name="adata" label="adata.loom" format="loom" from_work_dir="kb_outs/counts_unfiltered/adata.loom">
|
|
|
125 <filter>extra_dtype == "--loom"</filter>
|
|
|
126 </data>
|
|
|
127 </outputs>
|
|
|
128 <tests>
|
|
|
129 <expand macro="tests"/>
|
|
|
130 </tests>
|
|
|
131 <help><![CDATA[
|
|
|
132 This is a galaxy wrapper for kallisto-bustools (kb) for quantification of different types of single-cell sequencing data. The main kallisto-bustools homepage can be found under:
|
|
|
133 'kallisto-bustools<https://www.kallistobus.tools/>'.
|
|
|
134
|
|
|
135 The kb count command runs the kallisto and bustools programs. It can be used for pre-processing of data from a variety of single-cell RNA-seq technologies,
|
|
|
136 and for a number of different workflows (e.g. production of gene count matrices, RNA velocity analyses, etc.).
|
|
|
137
|
|
|
138 The kb package is developed and maintained by the Pachterlab under the MIT License.
|
|
|
139 ]]></help>
|
|
|
140 <citations>
|
|
|
141 <citation type="doi">https://doi.org/10.1101/673285</citation>
|
|
|
142 </citations>
|
|
|
143 </tool> |
