temp

.github/workflows/buid_and_release.yml

@@ -32,9 +32,7 @@ jobs:
         title: "${{ steps.current-time.outputs.formattedTime }}"
         prerelease: false
         # todo: add back `release/medgen-disease-extract.owl`, pending https://github.com/monarch-initiative/medgen/issues/11
-        # todo: add `medgen.sssom.tsv`, pending https://github.com/monarch-initiative/medgen/issues/6
 #          output/release/medgen-disease-extract.owl
-#          output/release/medgen.sssom.tsv
         files: |
           output/release/medgen.obo
           output/release/medgen-disease-extract.obo

makefile

@@ -14,7 +14,6 @@ minimal: build-lite stage-lite clean
 # stage-lite: These commented out files are produced by `all` but not by `minimal`. Just left here for reference. See: https://github.com/monarch-initiative/medgen/issues/11
 stage-lite: | output/release/
 #	mv medgen-disease-extract.owl output/release/
-#	mv medgen.sssom.tsv output/release/
 	mv *.obo output/release/
 	mv *.robot.template.tsv output/release/
 	mv *.sssom.tsv output/release/
@@ -75,21 +74,21 @@ medgen-disease-extract.obo: x-Disease-or-Syndrome.obo x-Neoplastic-Process.obo
 medgen-disease-extract.json: medgen-disease-extract.obo
 	owltools $< -o -f json $@
 
-medgen-disease-extract.owl: medgen-disease-extract.obo
+output/medgen-disease-extract.owl: medgen-disease-extract.obo | output/
 	owltools $< -o $@
 
 # SSSOM ----------------------------------
-# todo: comemented out old pipeline: remove
-#medgen.obographs.json:
-#	robot convert -i medgen-disease-extract.owl -o $@
-#
-#medgen.sssom.tsv: medgen.obographs.json
-#	sssom parse medgen.obographs.json -I obographs-json -m config/medgen.sssom-metadata.yml -o $@
 sssom: umls-hpo.sssom.tsv
 	sssom validate umls-hpo.sssom.tsv
 	sssom validate hpo-mesh.sssom.tsv
 
-umls-hpo.sssom.tsv hpo-mesh.sssom.tsv: ftp.ncbi.nlm.nih.gov/pub/medgen/MedGenIDMappings.txt
+output/medgen.obographs.json: output/medgen-disease-extract.owl | output/
+	robot convert -i medgen-disease-extract.owl -o $@
+
+output/medgen.sssom.tsv: output/medgen.obographs.json | output/
+	sssom parse medgen.obographs.json -I obographs-json -m config/medgen.sssom-metadata.yml -o $@
+
+umls-hpo.sssom.tsv hpo-mesh.sssom.tsv output/hpo-mesh_non-matches-included.sssom.tsv: ftp.ncbi.nlm.nih.gov/pub/medgen/MedGenIDMappings.txt
 	python src/create_sssom.py --input-mappings $< --input-sssom-config config/medgen.sssom-metadata.yml
 
 # ----------------------------------------
@@ -113,16 +112,18 @@ tmp/input/mondo.sssom.tsv: | tmp/input/
 	wget http://purl.obolibrary.org/obo/mondo/mappings/mondo.sssom.tsv -O $@
 
 # creates more than just this file; that goal creates multiple files
-output/medgen_terms_mapping_status.tsv output/obsoleted_medgen_terms_in_mondo.txt: tmp/input/mondo.sssom.tsv | output/
+output/medgen_terms_mapping_status.tsv output/obsoleted_medgen_terms_in_mondo.txt: tmp/input/mondo.sssom.tsv output/medgen.sssom.tsv | output/
 	python src/mondo_mapping_status.py
 
 # ----------------------------------------
 # Robot templates
 # ----------------------------------------
 # todo: Ideally I wanted this done at the end of the ingest, permuting from medgen.sssom.tsv, but there were some
-# problems with that file. Eventually changing to that feels like it makes more sense. Will have already been
-# pre-curated by disease. And some of the logic in this Python script is duplicative.
-medgen-xrefs.robot.template.tsv medgen-xrefs-mesh.robot.template.tsv: ftp.ncbi.nlm.nih.gov/pub/medgen/MedGenIDMappings.txt
+#  problems with that file. Eventually changing to that feels like it makes more sense. Will have already been
+#  pre-curated by disease. And some of the logic in this Python script is duplicative. Update 2024/05/19: I did include
+#  medgen.sssom.tsv as a prerequisite. the alternative build, 'make all' was failing without it. It has a different role
+#  than 'permutation' as I  mention above.
+medgen-xrefs.robot.template.tsv medgen-xrefs-mesh.robot.template.tsv: ftp.ncbi.nlm.nih.gov/pub/medgen/MedGenIDMappings.txt output/medgen.sssom.tsv
 	python src/mondo_robot_template.py -i $< \
 	--outpath-general medgen-xrefs.robot.template.tsv \
 	--outpath-mesh medgen-xrefs-mesh.robot.template.tsv

src/create_sssom.py

@@ -10,10 +10,12 @@
 PROJECT_DIR = SRC_DIR.parent
 FTP_DIR = PROJECT_DIR / "ftp.ncbi.nlm.nih.gov" / "pub" / "medgen"
 CONFIG_DIR = PROJECT_DIR / "config"
+OUTDIR = PROJECT_DIR / "output"
 INPUT_MAPPINGS = str(FTP_DIR / "MedGenIDMappings.txt")
 INPUT_CONFIG = str(CONFIG_DIR / "medgen.sssom-metadata.yml")
 OUTPUT_FILE_HPO_UMLS = str(PROJECT_DIR / "umls-hpo.sssom.tsv")
 OUTPUT_FILE_HPO_MESH = str(PROJECT_DIR / "hpo-mesh.sssom.tsv")
+OUTPUT_FILE_HPO_MESH_WITH_NON_MATCHES = str(OUTDIR / "hpo-mesh_non-matches-included.sssom.tsv")
 
 
 def _filter_and_format_cols(df: pd.DataFrame, source: str) -> pd.DataFrame:
@@ -54,8 +56,7 @@ def run(input_mappings: str = INPUT_MAPPINGS, input_sssom_config: str = INPUT_CO
     #  move the col removals below (umls) to above
     # - add mapping_justification
     df_hpo_mesh['mapping_justification'] = 'semapv:ManualMappingCuration'
-    write_sssom(df_hpo_mesh, input_sssom_config,
-                OUTPUT_FILE_HPO_MESH.replace('.sssom.tsv', '-non-matches-included.sssom.tsv'))
+    write_sssom(df_hpo_mesh, input_sssom_config, OUTPUT_FILE_HPO_MESH_WITH_NON_MATCHES)
     # -- filter non-matches & drop unneeded cols
     df_hpo_mesh = df_hpo_mesh[df_hpo_mesh['subject_id'].notna()][[
         x for x in df_hpo_mesh.columns if not x.startswith('umls')]]

src/mondo_mapping_status.py

@@ -10,7 +10,7 @@
 RELEASE_OUTDIR = OUTDIR / 'release'
 INPUT_DIR = PROJECT_DIR / 'tmp' / 'input'
 MONDO_SSSOM_TSV = INPUT_DIR / 'mondo.sssom.tsv'
-MEDGEN_SSSOM_TSV = RELEASE_OUTDIR / 'medgen.sssom.tsv'
+MEDGEN_SSSOM_TSV = OUTDIR / 'medgen.sssom.tsv'
 # MEDGEN_PREFIXES: Some of these are old, some are new, some may not be used.
 # todo: If I couldn't convert SSSOM properly with MedGen_CUI, souldn't UMLS_CUI have a problem? though i think it's just coming from previous work in mondo maybe. it's not being used in this ingest
 MEDGEN_PREFIXES = [