annotate test-data/reproduce_test_dataset.sh @ 6:84cc0dc92b0c draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit c0f150357d52e21443197b2ec1427feb9fc7971e
author iuc
date Wed, 06 Mar 2024 14:09:08 +0000
parents c2e4127fb5bf
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
1 #!/bin/bash
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
2
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
3 # This script produces a small kraken2 database containing only a ~1kb portion each of a salmonella and ecoli genome
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
4 # It requires kraken2, and entrez-direct (available on bioconda)
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
5 kraken2-build --db test_db --download_taxonomy
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
6 mv test_db/taxonomy/nucl_gb.accession2taxid test_db/taxonomy/nucl_gb.accession2taxid_full
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
7 grep -e 'NC_003198.1' -e 'NC_011750.1' test_db/taxonomy/nucl_gb.accession2taxid_full > test_db/taxonomy/nucl_gb.accession2taxid
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
8 mv test_db/taxonomy/nodes.dmp test_db/taxonomy/nodes.dmp_full
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
9 grep -f node_patterns.txt test_db/taxonomy/nodes.dmp_full > test_db/taxonomy/nodes.dmp
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
10 mv test_db/taxonomy/names.dmp test_db/taxonomy/names.dmp_full
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
11 grep -e '^220341\s' -e '^585057\s' test_db/taxonomy/names.dmp_full > test_db/taxonomy/names.dmp
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
12 esearch -db nucleotide -query "NC_003198.1" | efetch -format fasta > NC_003198.1.fasta
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
13 esearch -db nucleotide -query "NC_011750.1" | efetch -format fasta > NC_011750.1.fasta
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
14 head -n 14 NC_003198.1.fasta > NC_003198.1_1kb.fasta
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
15 head -n 14 NC_011750.1.fasta > NC_011750.1_1kb.fasta
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
16 kraken2-build --db test_db --add-to-library NC_003198.1_1kb.fasta
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
17 kraken2-build --db test_db --add-to-library NC_011750.1_1kb.fasta
c2e4127fb5bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bracken_database/ commit 74e81c69c8806d98beb15a889741bcd702866ce3"
iuc
parents:
diff changeset
18 kraken2-build --db test_db --build