diff --git a/config.yaml b/config.yaml index 42faed3152c848b68818a20afd206e39aa52b1d9..151fab9d2ba66f172cee3ff05acc8f8f1782ae30 100644 --- a/config.yaml +++ b/config.yaml @@ -8,8 +8,7 @@ ALVISNLP: '~/code/alvisnlp/.test/alvisnlp/bin/alvisnlp' # NCBI Taxonomy files -NCBI_NODES_FILE: 'ncbi-taxonomy/nodes.dmp' -NCBI_NAMES_FILE: 'ncbi-taxonomy/names.dmp' +NCBI_DIR: 'ncbi-taxonomy_2021-03-26' # Output and working directories diff --git a/dsmz-match.snakefile b/dsmz-match.snakefile index 832e9bb3d662673adb3d2d0c68cb83d453a398f4..30dd51eefe80f9855879cf9ac1a6f31e3bbf39f4 100644 --- a/dsmz-match.snakefile +++ b/dsmz-match.snakefile @@ -14,4 +14,4 @@ rule match: match='./dsmz-match.py', strains=config['OUTDIR'] + '/' + config['DSMZ_STRAINS_DIR'] - shell: '''{input.match} {config[NCBI_NODES_FILE]} {config[NCBI_NAMES_FILE]} {input.strains} {output}''' + shell: '''{input.match} {config[NCBI_DIR]}/nodes.dmp {config[NCBI_DIR]}/names.dmp {input.strains} {output}''' diff --git a/rewrite-taxonomy.snakefile b/rewrite-taxonomy.snakefile index 8ca2f6d5e8d1e9dfaf2f746bcb8e5f68f159fd51..2fbefe57feb24d1bb5e758bb985f8e0e21170a3d 100644 --- a/rewrite-taxonomy.snakefile +++ b/rewrite-taxonomy.snakefile @@ -8,4 +8,4 @@ rule rewrite: config['OUTDIR'] + '/' + config['DSMZ_MATCH_DIR'] shell: - '''../bibliome-java-utils/test/install/bin/rewrite-taxonomy -namesFile {config[NCBI_NAMES_FILE]} -namesFile {input}/dsmz-names.dmp -prefix ncbi: -rejectionFile reject.txt -saturationFile saturate.txt -pattern '{{NAME}}\t{{TAXID}}\t{{CANONICAL}}\t{{TAXID_PATH}}\t{{POS_TAG}}\t{{RANK}}\t{{SPECIES_TAXID}}\t{{SPECIES_NAME}}\n' {config[NCBI_NODES_FILE]} {input}/dsmz-nodes.dmp >{output}''' + '''../bibliome-java-utils/test/install/bin/rewrite-taxonomy -namesFile {config[NCBI_DIR]}/names.dmp -namesFile {input}/dsmz-names.dmp -prefix ncbi: -rejectionFile reject.txt -saturationFile saturate.txt -pattern '{{NAME}}\t{{TAXID}}\t{{CANONICAL}}\t{{TAXID_PATH}}\t{{POS_TAG}}\t{{RANK}}\t{{SPECIES_TAXID}}\t{{SPECIES_NAME}}\n' {config[NCBI_DIR]}/nodes.dmp {input}/dsmz-nodes.dmp >{output}'''