Merge pull request #88 from microbiomedata/69-run-lexmatch-agaisnt-nm…

…dc-schema-and-push-sssom-output-to-repo

run lexmatch agaisnt nmdc schema and push sssom output to repo

for-lexmatch.sh

@@ -0,0 +1,67 @@
+#!/bin/bash
+
+# Check if the input file exists
+if [ ! -f for-lexmatch.txt ]; then
+  echo "File for-lexmatch.txt not found!"
+  exit 1
+fi
+
+rm -rf lexmatch-errors.txt
+#rm -rf lexmatch-output/*
+
+# Loop through each line in the input file
+while IFS= read -r ontology; do
+  if [ -n "$ontology" ]; then
+
+#    echo "nmdc mixs vs $ontology"
+#    poetry run runoak \
+#      -i nmdc.db \
+#      -a sqlite:obo:$ontology \
+#      lexmatch \
+#      --add-pipeline-step CaseNormalization \
+#      --add-pipeline-step WhitespaceNormalization  \
+#      --add-pipeline-step WordOrderNormalization \
+#      --output lexmatch-output/nmdc_mixs_vs_${ontology}.SSSOM.tsv \
+#      i^mixs @ .all \
+#      2>>lexmatch-errors.txt
+#
+#    echo "nmdc nmdc vs $ontology"
+#    poetry run runoak \
+#      -i nmdc.db \
+#      -a sqlite:obo:$ontology \
+#      lexmatch \
+#      --add-pipeline-step CaseNormalization \
+#      --add-pipeline-step WhitespaceNormalization  \
+#      --add-pipeline-step WordOrderNormalization \
+#      --output lexmatch-output/nmdc_nmdc_vs_${ontology}.SSSOM.tsv \
+#      i^nmdc @ .all \
+#      2>>lexmatch-errors.txt
+#
+#    echo "submission schema vs $ontology"
+#    poetry run runoak \
+#      -i nmdc_submission_schema_no_from_schema_no_brackets.db \
+#      -a sqlite:obo:$ontology \
+#      lexmatch \
+#      --add-pipeline-step CaseNormalization \
+#      --add-pipeline-step WhitespaceNormalization  \
+#      --add-pipeline-step WordOrderNormalization \
+#      --output lexmatch-output/submission_schema_vs_${ontology}.SSSOM.tsv \
+#      i^nmdc_sub_schema @ .all \
+#      2>>lexmatch-errors.txt
+
+
+    echo "env triad PVs vs $ontology"
+    poetry run runoak \
+      -i env_triad_pvs_sheet.db \
+      -a sqlite:obo:$ontology \
+      lexmatch \
+      --add-pipeline-step CaseNormalization \
+      --add-pipeline-step WhitespaceNormalization  \
+      --add-pipeline-step WordOrderNormalization \
+      --output lexmatch-output/env_triad_pvs_vs_${ontology}.SSSOM.tsv \
+      i^nmdc_sub_schema @ .all \
+      2>>lexmatch-errors.txt
+  fi
+done < for-lexmatch.txt
+
+echo "All lexmatch operations completed."

for-lexmatch.txt

@@ -0,0 +1,25 @@
+agro
+biolink
+chmo
+cob
+comet
+ecocore
+ecosim
+ecso
+ecto
+edam
+envo
+envthes
+goldterms
+iao
+ido
+mop
+ms
+obi
+omo
+omit
+pato
+po
+schema-dot-org
+sio
+sweetAll

lexmatch-output/env_triad_pvs_vs_cob.SSSOM.tsv

@@ -0,0 +1,16 @@
+# curie_map:
+#   COB: http://purl.obolibrary.org/obo/COB_
+#   ENVO: http://purl.obolibrary.org/obo/ENVO_
+#   RO: http://purl.obolibrary.org/obo/RO_
+#   owl: http://www.w3.org/2002/07/owl#
+#   rdf: http://www.w3.org/1999/02/22-rdf-syntax-ns#
+#   rdfs: http://www.w3.org/2000/01/rdf-schema#
+#   semapv: https://w3id.org/semapv/vocab/
+#   skos: http://www.w3.org/2004/02/skos/core#
+#   sssom: https://w3id.org/sssom/
+# license: https://w3id.org/sssom/license/unspecified
+# mapping_set_id: https://w3id.org/sssom/mappings/f0aa1cf4-ee40-4933-be0d-903f7d8ed0cc
+subject_id	subject_label	predicate_id	object_id	object_label	mapping_justification	mapping_tool	confidence	subject_match_field	object_match_field	match_string
+COB:0000502	characteristic	skos:closeMatch	RO:0000052	characteristic of	semapv:LexicalMatching	oaklib	0.5	rdfs:label	rdfs:label	characteristic
+ENVO:00002042	surface water	skos:closeMatch	ENVO:01001191	water surface	semapv:LexicalMatching	oaklib	0.5	rdfs:label	rdfs:label	surface water
+ENVO:00003096	tap water	skos:closeMatch	ENVO:03600052	water tap	semapv:LexicalMatching	oaklib	0.5	rdfs:label	rdfs:label	tap water

lexmatch-output/env_triad_pvs_vs_comet.SSSOM.tsv

@@ -0,0 +1,20 @@
+# curie_map:
+#   ENVO: http://purl.obolibrary.org/obo/ENVO_
+#   comet: http://w3id.org/sssom/unknown_prefix/comet/
+#   owl: http://www.w3.org/2002/07/owl#
+#   rdf: http://www.w3.org/1999/02/22-rdf-syntax-ns#
+#   rdfs: http://www.w3.org/2000/01/rdf-schema#
+#   semapv: https://w3id.org/semapv/vocab/
+#   skos: http://www.w3.org/2004/02/skos/core#
+#   sssom: https://w3id.org/sssom/
+# license: https://w3id.org/sssom/license/unspecified
+# mapping_set_id: https://w3id.org/sssom/mappings/7bfe79fb-0c2c-4195-b83b-9c1dc0db925c
+subject_id	subject_label	predicate_id	object_id	object_label	mapping_justification	mapping_tool	confidence	subject_match_field	object_match_field	match_string
+ENVO:00000141	bank	skos:closeMatch	comet:bank	bank	semapv:LexicalMatching	oaklib	0.5	rdfs:label	rdfs:label	bank
+ENVO:00002042	surface water	skos:closeMatch	ENVO:01001191	water surface	semapv:LexicalMatching	oaklib	0.5	rdfs:label	rdfs:label	surface water
+ENVO:00003096	tap water	skos:closeMatch	ENVO:03600052	water tap	semapv:LexicalMatching	oaklib	0.5	rdfs:label	rdfs:label	tap water
+comet:ExtractiveIndustryProductType#COAL	COAL	skos:closeMatch	comet:FossilFuelType#COAL	COAL	semapv:LexicalMatching	oaklib	0.5	rdfs:label	rdfs:label	coal
+comet:ExtractiveIndustryProductType#GAS	GAS	skos:closeMatch	comet:WellType#GAS	GAS	semapv:LexicalMatching	oaklib	0.5	rdfs:label	rdfs:label	gas
+comet:ExtractiveIndustryProductType#OIL	OIL	skos:closeMatch	comet:WellType#OIL	OIL	semapv:LexicalMatching	oaklib	0.5	rdfs:label	rdfs:label	oil
+comet:HealthcareSite	HealthcareSite	skos:closeMatch	comet:PostalAddress	PostalAddress	semapv:LexicalMatching	oaklib	0.5	skos:exactMatch	skos:exactMatch	ex:omop/care_site
+comet:Location	Location	skos:closeMatch	comet:location	location	semapv:LexicalMatching	oaklib	0.5	rdfs:label	rdfs:label	location