Mercurial > repos > galaxytrakr > aws_sra

annotate aws_sra.xml @ 40:04632fbfb752 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit ae8bd99f91b8f4bbf9a020c7a641425a2931106a
author galaxytrakr
date Wed, 25 Mar 2026 11:02:25 +0000
parents bb993ca44098
children d91855b2a813
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
40
04632fbfb752 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit ae8bd99f91b8f4bbf9a020c7a641425a2931106a
galaxytrakr
parents: 39
diff changeset
1 <tool id="aws_sra" name="NCBI SRA AWS Fetch" version="0.4.0+gt_1.9" profile="23.0">
23
d7f68b3cde39 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit e9adf514c1b6b341c9e5bf8cc5a41c79b738d48e
galaxytrakr
parents: 22
diff changeset
2 <description>Fetches one or more SRA runs from AWS S3 and converts them to FASTQ</description>
21
02f45c03c306 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit f72940592c22c9ba88f1dcb23ef8bb5199ce434e
galaxytrakr
parents: 20
diff changeset
3
0
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
4 <requirements>
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
5 <requirement type="package" version="3.1.1">sra-tools</requirement>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
6 <requirement type="package" version="2.8">pigz</requirement>
2
83c923bd9da6 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit a95de952e58b29875b31e55f0ef06a83a18618fc
galaxytrakr
parents: 1
diff changeset
7 <requirement type="package" version="2.34.8">awscli</requirement>
0
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
8 </requirements>
21
02f45c03c306 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit f72940592c22c9ba88f1dcb23ef8bb5199ce434e
galaxytrakr
parents: 20
diff changeset
9
13
2897d365dd62 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 619ebd7e6a24be0ec6c2728511290f43b0bad89f
galaxytrakr
parents: 12
diff changeset
10 <version_command>fasterq-dump --version</version_command>
2897d365dd62 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 619ebd7e6a24be0ec6c2728511290f43b0bad89f
galaxytrakr
parents: 12
diff changeset
11
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
12 <command detect_errors="exit_code"><![CDATA[
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
13 #if $input.input_select == "accession_number":
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
14 echo '${input.accession}' | sed -r 's/(\,|\;|__cn__)/\n/g' > accessions &&
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
15 #else:
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
16 grep '^[[:space:]]*[ESD]RR[0-9]\{1,\}[[:space:]]*$' '${input.file_list}' > accessions &&
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
17 #end if
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
18 mkdir -p output &&
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
19 mkdir -p outputOther &&
32
8b8a63786853 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 5be25e80679a5df7d0ee7526e85d727024e78920
galaxytrakr
parents: 31
diff changeset
20 mkdir -p outputSingle &&
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
21 for acc in \$(cat ./accessions);
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
22 do (
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
23 echo "Processing accession: \$acc" &&
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
24 mkdir -p sra_cache_\${acc} &&
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
25 aws s3 cp --no-sign-request "s3://sra-pub-run-odp/sra/\${acc}/\${acc}" ./sra_cache_\${acc}/\${acc} &&
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
26 fasterq-dump -e \${GALAXY_SLOTS:-4} -t . --split-3 ./sra_cache_\${acc}/\${acc} &&
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
27 rm -rf sra_cache_\${acc} &&
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
28 count="\$(ls \${acc}*.fastq 2>/dev/null | wc -l)" &&
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
29 echo "Found \$count fastq file(s) for \$acc" &&
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
30 data=(\$(ls \${acc}*.fastq 2>/dev/null)) &&
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
31 if [ "\$count" -eq 1 ]; then
40
04632fbfb752 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit ae8bd99f91b8f4bbf9a020c7a641425a2931106a
galaxytrakr
parents: 39
diff changeset
32 pigz -cqp \${GALAXY_SLOTS:-4} "\${data[0]}" > outputSingle/"\${acc}"__single.fastqsanger.gz &&
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
33 rm "\${data[0]}";
39
bb993ca44098 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 5c4ba2cc27bc718393cb3b267e6fccb0ab48038a
galaxytrakr
parents: 38
diff changeset
34 elif [ "\$count" -eq 2 ]; then
36
4632b9e5e884 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 2233d6d0b755c3cec6bfc718372d7ea8fc9192bc
galaxytrakr
parents: 35
diff changeset
35 pigz -cqp \${GALAXY_SLOTS:-4} "\${acc}"_1.fastq > output/"\${acc}"_forward.fastqsanger.gz &&
4632b9e5e884 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 2233d6d0b755c3cec6bfc718372d7ea8fc9192bc
galaxytrakr
parents: 35
diff changeset
36 pigz -cqp \${GALAXY_SLOTS:-4} "\${acc}"_2.fastq > output/"\${acc}"_reverse.fastqsanger.gz &&
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
37 rm "\${acc}"*.fastq;
39
bb993ca44098 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 5c4ba2cc27bc718393cb3b267e6fccb0ab48038a
galaxytrakr
parents: 38
diff changeset
38 elif [ -e "\${acc}".fastq ]; then
bb993ca44098 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 5c4ba2cc27bc718393cb3b267e6fccb0ab48038a
galaxytrakr
parents: 38
diff changeset
39 pigz -cqp \${GALAXY_SLOTS:-4} "\${acc}".fastq > outputSingle/"\${acc}"__single.fastqsanger.gz &&
bb993ca44098 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 5c4ba2cc27bc718393cb3b267e6fccb0ab48038a
galaxytrakr
parents: 38
diff changeset
40 pigz -cqp \${GALAXY_SLOTS:-4} "\${acc}"_forward.fastq > output/"\${acc}"_forward.fastqsanger.gz &&
bb993ca44098 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 5c4ba2cc27bc718393cb3b267e6fccb0ab48038a
galaxytrakr
parents: 38
diff changeset
41 pigz -cqp \${GALAXY_SLOTS:-4} "\${acc}"_reverse.fastq > output/"\${acc}"_reverse.fastqsanger.gz &&
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
42 rm "\${acc}"*.fastq;
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
43 else
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
44 for file in \${data[*]}; do
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
45 pigz -cqp \${GALAXY_SLOTS:-4} "\$file" > outputOther/"\$file"sanger.gz &&
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
46 rm "\$file";
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
47 done;
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
48 fi
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
49 ); done;
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
50 echo "Done with all accessions."
0
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
51 ]]></command>
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
52
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
53 <inputs>
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
54 <conditional name="input">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
55 <param name="input_select" type="select" label="Select input type">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
56 <option value="accession_number">SRA Accession number(s)</option>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
57 <option value="file_list">File containing accession list</option>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
58 </param>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
59 <when value="accession_number">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
60 <param name="accession" type="text" label="SRA Accession(s)"
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
61 help="One or more SRA run accessions (SRR, ERR, or DRR), separated by commas or spaces.">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
62 <validator type="empty_field" message="At least one SRA accession is required."/>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
63 <sanitizer>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
64 <valid initial="string.printable">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
65 <remove value="&apos;"/>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
66 </valid>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
67 <mapping initial="none">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
68 <add source="&apos;" target="&apos;&quot;&apos;&quot;&apos;"/>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
69 </mapping>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
70 </sanitizer>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
71 </param>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
72 </when>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
73 <when value="file_list">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
74 <param name="file_list" type="data" format="txt,tabular" label="Accession list file"
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
75 help="A text file with one SRA accession (SRR, ERR, or DRR) per line."/>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
76 </when>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
77 </conditional>
0
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
78 </inputs>
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
79
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
80 <outputs>
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
81 <collection name="list_paired" type="list:paired" label="Paired-end FASTQ (aws_sra)">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
82 <discover_datasets
39
bb993ca44098 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 5c4ba2cc27bc718393cb3b267e6fccb0ab48038a
galaxytrakr
parents: 38
diff changeset
83 pattern="(?P&lt;identifier_0&gt;[^_]+)_(?P&lt;identifier_1&gt;[^_]+)\.fastqsanger.gz"
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
84 directory="output"
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
85 ext="fastqsanger.gz"/>
23
d7f68b3cde39 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit e9adf514c1b6b341c9e5bf8cc5a41c79b738d48e
galaxytrakr
parents: 22
diff changeset
86 </collection>
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
87 <collection name="output_single" type="list" label="Single-end FASTQ (aws_sra)">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
88 <discover_datasets
39
bb993ca44098 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 5c4ba2cc27bc718393cb3b267e6fccb0ab48038a
galaxytrakr
parents: 38
diff changeset
89 pattern="(?P&lt;designation&gt;.+)__single\.fastqsanger.gz"
32
8b8a63786853 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 5be25e80679a5df7d0ee7526e85d727024e78920
galaxytrakr
parents: 31
diff changeset
90 directory="outputSingle"
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
91 ext="fastqsanger.gz"/>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
92 </collection>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
93 <collection name="output_other" type="list" label="Other FASTQ (aws_sra)">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
94 <discover_datasets
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
95 pattern="(?P&lt;designation&gt;.+)\.fastqsanger\.gz"
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
96 directory="outputOther"
39
bb993ca44098 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 5c4ba2cc27bc718393cb3b267e6fccb0ab48038a
galaxytrakr
parents: 38
diff changeset
97 format="fastqsanger.gz"/>
23
d7f68b3cde39 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit e9adf514c1b6b341c9e5bf8cc5a41c79b738d48e
galaxytrakr
parents: 22
diff changeset
98 </collection>
21
02f45c03c306 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit f72940592c22c9ba88f1dcb23ef8bb5199ce434e
galaxytrakr
parents: 20
diff changeset
99 </outputs>
17
9fb80e0392ce planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 9707fa5e3ca6db5b58f271d133484d078cf65390
galaxytrakr
parents: 16
diff changeset
100
21
02f45c03c306 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit f72940592c22c9ba88f1dcb23ef8bb5199ce434e
galaxytrakr
parents: 20
diff changeset
101 <tests>
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
102 <test expect_num_outputs="3">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
103 <conditional name="input">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
104 <param name="input_select" value="accession_number"/>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
105 <param name="accession" value="SRR13333333"/>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
106 </conditional>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
107 <output_collection name="list_paired" type="list:paired" count="1">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
108 <element name="SRR13333333">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
109 <element name="forward" ftype="fastqsanger.gz">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
110 <assert_contents>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
111 <has_text text="@SRR13333333"/>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
112 </assert_contents>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
113 </element>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
114 <element name="reverse" ftype="fastqsanger.gz">
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
115 <assert_contents>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
116 <has_text text="@SRR13333333"/>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
117 </assert_contents>
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
118 </element>
24
7dbb60c48056 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit 29ed254c4aaab320b5482db410bd236d00f1b70e
galaxytrakr
parents: 23
diff changeset
119 </element>
23
d7f68b3cde39 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit e9adf514c1b6b341c9e5bf8cc5a41c79b738d48e
galaxytrakr
parents: 22
diff changeset
120 </output_collection>
21
02f45c03c306 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit f72940592c22c9ba88f1dcb23ef8bb5199ce434e
galaxytrakr
parents: 20
diff changeset
121 </test>
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
122
21
02f45c03c306 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit f72940592c22c9ba88f1dcb23ef8bb5199ce434e
galaxytrakr
parents: 20
diff changeset
123 </tests>
02f45c03c306 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit f72940592c22c9ba88f1dcb23ef8bb5199ce434e
galaxytrakr
parents: 20
diff changeset
124
0
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
125 <help><![CDATA[
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
126 **NCBI SRA AWS Fetch**
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
127
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
128 Fetches one or more SRA runs from the public ``sra-pub-run-odp`` S3 bucket and converts them to
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
129 gzip-compressed FASTQ using ``fasterq-dump``. Downloads use ``aws s3 cp`` with no credentials
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
130 required (public bucket).
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
131
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
132 **Inputs**
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
133
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
134 - **SRA Accession number(s)**: Type one or more accessions (SRR, ERR, or DRR) directly, comma- or space-separated.
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
135 - **File containing accession list**: A text file with one accession per line. This option is connectable as a workflow input.
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
136
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
137 **Outputs**
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
138
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
139 Three collections are always created (some may be empty depending on the data):
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
140
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
141 - **Paired-end FASTQ**: A nested ``list:paired`` collection — each accession contains a forward/reverse pair.
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
142 - **Single-end FASTQ**: A flat list for reads confirmed single-end by ``--split-3``.
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
143 - **Other FASTQ**: Reads that could not be cleanly classified.
0
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
144 ]]></help>
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
145
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
146 <citations>
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
147 <citation type="bibtex">
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
148 @misc{ncbi_sra_aws,
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
149 title = {{NCBI} {SRA} on {AWS} Open Data},
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
150 author = {{National Center for Biotechnology Information}},
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
151 howpublished = {\url{https://registry.opendata.aws/ncbi-sra/}},
0
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
152 note = {Accessed via AWS S3 without credentials}
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
153 }
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
154 </citation>
30
73ee30eb273a planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit d0c08749588099d40db3c23bfd554800ac307a30
galaxytrakr
parents: 29
diff changeset
155 <citation type="doi">10.1093/nar/gkq1019</citation>
0
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
156 </citations>
a4afe551dfc9 planemo upload for repository https://github.com/CFSAN-Biostatistics/galaxytrakr-tools commit c55c06b92c0ee0429047bcff1992bf2ec293284a
galaxytrakr
parents:
diff changeset
157 </tool>