From 5e702e9ec9cc93db37a0597f2375ffad22f47937 Mon Sep 17 00:00:00 2001
From: Mouhamadou Ba <mandiayba@gmail.com>
Date: Mon, 11 Apr 2022 14:39:58 +0200
Subject: [PATCH 1/2] update path to ontologies
---
preprocess-ontology.snakefile | 42 ++++++++++++++++----------------
process_CIRM_corpus.snakefile | 12 ++++-----
process_DSMZ_corpus.snakefile | 4 +--
process_GenBank_corpus.snakefile | 4 +--
process_PubMed_corpus.snakefile | 16 ++++++------
5 files changed, 39 insertions(+), 39 deletions(-)
diff --git a/preprocess-ontology.snakefile b/preprocess-ontology.snakefile
index 82f53e34..11bce226 100644
--- a/preprocess-ontology.snakefile
+++ b/preprocess-ontology.snakefile
@@ -22,14 +22,14 @@ rule all:
'corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.json',
'corpora/florilege/alvisir/BioNLP-OST+EnovFood-Phenotype.json',
'corpora/florilege/alvisir/BioNLP-OST+EnovFood-Use.json',
- 'corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.tomap',
- 'corpora/florilege/alvisir/BioNLP-OST+EnovFood-Phenotype.tomap',
+ 'ancillaries/BioNLP-OST+EnovFood-Habitat.tomap',
+ 'ancillaries/BioNLP-OST+EnovFood-Phenotype.tomap',
'ancillaries/food-process-lexicon.txt',
'ancillaries/NCBI_taxa_ontobiotope.txt',
- 'corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo',
- 'corpora/florilege/alvisir/BioNLP-OST+EnovFood-Phenotype.obo',
- 'corpora/florilege/alvisir/BioNLP-OST+EnovFood-Use.obo',
- expand("corpora/florilege/alvisir/{ontoname}.paths", ontoname=ONTONAMES.split(' '))
+ 'ancillaries/BioNLP-OST+EnovFood-Habitat.obo',
+ 'ancillaries/BioNLP-OST+EnovFood-Phenotype.obo',
+ 'ancillaries/BioNLP-OST+EnovFood-Use.obo',
+ expand("ancillaries/{ontoname}.paths", ontoname=ONTONAMES.split(' '))
'''
@@ -51,7 +51,7 @@ rule cut_subtrees_habitat:
input:
onto='ancillaries/BioNLP-OST+EnovFood-no-obsolete.obo'
output:
- onto='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo'
+ onto='ancillaries/BioNLP-OST+EnovFood-Habitat.obo'
conda: 'softwares/envs/obo-utils-env.yaml'
shell: """
python softwares/obo-utils/obo-subtree.py \
@@ -67,7 +67,7 @@ rule cut_subtrees_phenotype:
input:
onto='ancillaries/BioNLP-OST+EnovFood-no-obsolete.obo'
output:
- onto='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Phenotype.obo'
+ onto='ancillaries/BioNLP-OST+EnovFood-Phenotype.obo'
conda: 'softwares/envs/obo-utils-env.yaml'
shell: """
python softwares/obo-utils/obo-subtree.py \
@@ -83,7 +83,7 @@ rule cut_subtrees_use:
input:
onto='ancillaries/BioNLP-OST+EnovFood-no-obsolete.obo'
output:
- onto='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Use.obo'
+ onto='ancillaries/BioNLP-OST+EnovFood-Use.obo'
conda: 'softwares/envs/obo-utils-env.yaml'
shell: """
python softwares/obo-utils/obo-subtree.py \
@@ -98,9 +98,9 @@ Analyze ontologies with tomap
'''
rule analyze_onto_Habitat:
input:
- onto='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo'
+ onto='ancillaries/BioNLP-OST+EnovFood-Habitat.obo'
output:
- tomap='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.tomap'
+ tomap='ancillaries/BioNLP-OST+EnovFood-Habitat.tomap'
params:
plan='plans/biotope_ontology_analyzer.plan'
singularity:config["SINGULARITY_IMG"]
@@ -116,9 +116,9 @@ Analyze ontologies with tomap
'''
rule analyze_onto_phenotype:
input:
- onto='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Phenotype.obo'
+ onto='ancillaries/BioNLP-OST+EnovFood-Phenotype.obo'
output:
- tomap='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Phenotype.tomap'
+ tomap='ancillaries/BioNLP-OST+EnovFood-Phenotype.tomap'
params:
plan='plans/phenotype_ontology_analyzer.plan'
singularity:config["SINGULARITY_IMG"]
@@ -135,7 +135,7 @@ Build food process lexicon
'''
rule build_food_process_lexicon:
input:
- onto='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo'
+ onto='ancillaries/BioNLP-OST+EnovFood-Habitat.obo'
output:
lexicon='ancillaries/food-process-lexicon.txt'
params:
@@ -163,9 +163,9 @@ build name lexicon
'''
rule build_ncbi_common_name_lexicon:
input:
- onto='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo',
+ onto='ancillaries/BioNLP-OST+EnovFood-Habitat.obo',
common_names='ancillaries/NCBI_common_names',
- tomap='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.tomap',
+ tomap='ancillaries/BioNLP-OST+EnovFood-Habitat.tomap',
graylist='ancillaries/graylist_extended.heads',
emptywords='ancillaries/stopwords_EN.ttg'
output:
@@ -186,7 +186,7 @@ obo to json
'''
rule convert_obo2json_habitat:
input:
- obo='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo'
+ obo='ancillaries/BioNLP-OST+EnovFood-Habitat.obo'
output:
json='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.json'
conda: 'softwares/envs/obo-utils-env.yaml'
@@ -197,7 +197,7 @@ convert phenotype results to json
'''
rule convert_obo2json_phenotype:
input:
- obo='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Phenotype.obo'
+ obo='ancillaries/BioNLP-OST+EnovFood-Phenotype.obo'
output:
json='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Phenotype.json'
conda: 'softwares/envs/obo-utils-env.yaml'
@@ -209,7 +209,7 @@ convert use results to json
'''
rule convert_obo2json_use:
input:
- obo='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Use.obo'
+ obo='ancillaries/BioNLP-OST+EnovFood-Use.obo'
output:
json='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Use.json'
conda: 'softwares/envs/obo-utils-env.yaml'
@@ -222,9 +222,9 @@ generate concept paths
'''
rule generate_concept_path:
input:
- onto="corpora/florilege/alvisir/{ontoname}.obo"
+ onto="ancillaries/{ontoname}.obo"
output:
- paths="corpora/florilege/alvisir/{ontoname}.paths"
+ paths="ancillaries/{ontoname}.paths"
params:
plan="plans/get_onto_paths.plan"
singularity:config['SINGULARITY_IMG']
diff --git a/process_CIRM_corpus.snakefile b/process_CIRM_corpus.snakefile
index 902e3b40..fee3a27a 100644
--- a/process_CIRM_corpus.snakefile
+++ b/process_CIRM_corpus.snakefile
@@ -144,8 +144,8 @@ rule map_cirm_habitats:
mapped_habitats='corpora/cirm/bia/mapped_bia_habitats.txt'
params:
plan='plans/map_habitats.plan',
- onto='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo',
- tomap='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.tomap',
+ onto='ancillaries/BioNLP-OST+EnovFood-Habitat.obo',
+ tomap='ancillaries/BioNLP-OST+EnovFood-Habitat.tomap',
graylist='ancillaries/graylist_extended.heads',
emptywords='ancillaries/stopwords_EN.ttg',
outdir='corpora/cirm/bia',
@@ -170,8 +170,8 @@ rule map_cirm_yeast_habitats:
mapped_habitats='corpora/cirm/levures/mapped_yeast_habitats.txt'
params:
plan='plans/map_habitats.plan',
- onto='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo',
- tomap='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.tomap',
+ onto='ancillaries/BioNLP-OST+EnovFood-Habitat.obo',
+ tomap='ancillaries/BioNLP-OST+EnovFood-Habitat.tomap',
graylist='ancillaries/graylist_extended.heads',
emptywords='ancillaries/stopwords_EN.ttg',
outdir='corpora/cirm/levures',
@@ -196,8 +196,8 @@ rule map_cirm_cfbp_habitats:
mapped_habitats='corpora/cirm/cfbp/mapped_cfbp_habitats.txt'
params:
plan='plans/map_habitats.plan',
- onto='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo',
- tomap='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.tomap',
+ onto='ancillaries/BioNLP-OST+EnovFood-Habitat.obo',
+ tomap='ancillaries/BioNLP-OST+EnovFood-Habitat.tomap',
graylist='ancillaries/graylist_extended.heads',
emptywords='ancillaries/stopwords_EN.ttg',
outdir='corpora/cirm/cfbp',
diff --git a/process_DSMZ_corpus.snakefile b/process_DSMZ_corpus.snakefile
index 9db4c34a..7ed58a57 100644
--- a/process_DSMZ_corpus.snakefile
+++ b/process_DSMZ_corpus.snakefile
@@ -32,8 +32,8 @@ rule map_dsmz_habitats:
mapped_habitats='corpora/dsmz/mapped_habitats.txt'
params:
plan='plans/map_habitats.plan',
- onto='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo',
- tomap='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.tomap',
+ onto='ancillaries/BioNLP-OST+EnovFood-Habitat.obo',
+ tomap='ancillaries/BioNLP-OST+EnovFood-Habitat.tomap',
graylist='ancillaries/graylist_extended.heads',
emptywords='ancillaries/stopwords_EN.ttg',
outdir='corpora/dsmz',
diff --git a/process_GenBank_corpus.snakefile b/process_GenBank_corpus.snakefile
index 89d075f5..c71dc069 100644
--- a/process_GenBank_corpus.snakefile
+++ b/process_GenBank_corpus.snakefile
@@ -75,8 +75,8 @@ rule map_genbank_habitats:
mapped_habitats='corpora/genbank/mapped_habitats.txt'
params:
plan='plans/map_habitats.plan',
- onto='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo',
- tomap='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.tomap',
+ onto='ancillaries/BioNLP-OST+EnovFood-Habitat.obo',
+ tomap='ancillaries/BioNLP-OST+EnovFood-Habitat.tomap',
graylist='ancillaries/graylist_extended.heads',
emptywords='ancillaries/stopwords_EN.ttg',
inhibitSyntax='inhibit-syntax',
diff --git a/process_PubMed_corpus.snakefile b/process_PubMed_corpus.snakefile
index cc072812..dff513c7 100644
--- a/process_PubMed_corpus.snakefile
+++ b/process_PubMed_corpus.snakefile
@@ -45,13 +45,13 @@ rule run_pubmed_entities:
batch="{B}",
corpus='pubmed',
inhibitSyntax='inhibit-syntax',
- onto_habitat='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo',
- tomap_habitat='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.tomap',
- onto_pheno='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Phenotype.obo',
- tomap_pheno='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Phenotype.tomap',
+ onto_habitat='ancillaries/BioNLP-OST+EnovFood-Habitat.obo',
+ tomap_habitat='ancillaries/BioNLP-OST+EnovFood-Habitat.tomap',
+ onto_pheno='ancillaries/BioNLP-OST+EnovFood-Phenotype.obo',
+ tomap_pheno='ancillaries/BioNLP-OST+EnovFood-Phenotype.tomap',
graylist='ancillaries/graylist_extended.heads',
emptywords='ancillaries/stopwords_EN.ttg',
- ontobiotopeUse='corpora/florilege/alvisir/BioNLP-OST+EnovFood-Use.obo',
+ ontobiotopeUse='ancillaries/BioNLP-OST+EnovFood-Use.obo',
plan='plans/entities.plan',
dir='corpora/pubmed/batches/{B}/',
taxid_microorganisms='ancillaries/extended-microorganisms-taxonomy/taxid_microorganisms.txt',
@@ -130,9 +130,9 @@ rule create_pubmed_expander:
input:
expander="ancillaries/expander.xml",
taxa_id_microorganisms="ancillaries/extended-microorganisms-taxonomy/taxa+id_microorganisms.txt",
- onto_habitat="corpora/florilege/alvisir/BioNLP-OST+EnovFood-Habitat.obo",
- onto_phenotype="corpora/florilege/alvisir/BioNLP-OST+EnovFood-Phenotype.obo",
- onto_use="corpora/florilege/alvisir/BioNLP-OST+EnovFood-Use.obo"
+ onto_habitat="ancillaries/BioNLP-OST+EnovFood-Habitat.obo",
+ onto_phenotype="ancillaries/BioNLP-OST+EnovFood-Phenotype.obo",
+ onto_use="ancillaries/BioNLP-OST+EnovFood-Use.obo"
output:
expander_folder=directory("corpora/florilege/alvisir/expander")
params:
--
GitLab
From e088375326494d4a168dfd16e6daca22fd446b28 Mon Sep 17 00:00:00 2001
From: Mouhamadou Ba <mandiayba@gmail.com>
Date: Mon, 11 Apr 2022 14:44:56 +0200
Subject: [PATCH 2/2] Deleted generate_concept_path.snakefile
---
generate_concept_path.snakefile | 34 ---------------------------------
1 file changed, 34 deletions(-)
delete mode 100644 generate_concept_path.snakefile
diff --git a/generate_concept_path.snakefile b/generate_concept_path.snakefile
deleted file mode 100644
index ca1ba773..00000000
--- a/generate_concept_path.snakefile
+++ /dev/null
@@ -1,34 +0,0 @@
-## config file
-configfile: "config/config.yaml"
-
-
-
-ONTONAMES = 'BioNLP-OST+EnovFood-Habitat BioNLP-OST+EnovFood-Phenotype BioNLP-OST+EnovFood-Use'
-
-
-
-'''
-all
-'''
-rule all:
- input:
- expand("ancillaries/{ontoname}.paths", ontoname=ONTONAMES.split(' '))
-
-
-
-'''
-generate concept paths
-'''
-rule generate_concept_path:
- input:
- onto="ancillaries/{ontoname}.obo"
- output:
- paths="ancillaries/{ontoname}.paths"
- params:
- plan="plans/get_onto_paths.plan"
- singularity:config['SINGULARITY_IMG']
- shell: """alvisnlp -cleanTmp -verbose \
- -alias input {input.onto} \
- -alias output {output.paths} \
- {params.plan}
- """
--
GitLab