From cfe43383f5ff7883d4e9d977ae2c7a8fdef42760 Mon Sep 17 00:00:00 2001
From: Josef Hardi <josef.hardi@gmail.com>
Date: Sat, 3 Aug 2024 15:41:39 -0700
Subject: [PATCH] Combine together CL and PCL to simplify the process

---
 scripts/download-ontologies.sh  | 15 +++++++++------
 src/enrichment/enrich-2d-ftu.js | 13 -------------
 src/enrichment/enrich-asct-b.js | 13 -------------
 src/utils/get-cl-terms.sparql   |  6 ++++--
 src/utils/get-pcl-terms.sparql  | 16 ----------------
 5 files changed, 13 insertions(+), 50 deletions(-)
 delete mode 100644 src/utils/get-pcl-terms.sparql

diff --git a/scripts/download-ontologies.sh b/scripts/download-ontologies.sh
index d76960c..b523d3e 100755
--- a/scripts/download-ontologies.sh
+++ b/scripts/download-ontologies.sh
@@ -36,21 +36,24 @@ rm -rf $MIRROR_DIR/fma.zip
 
 echo "Downloading the latest CL ontology..."
 curl -L "$OBO_BASE_URL/cl/cl-base.owl" \
-     --create-dirs -o "$MIRROR_DIR/cl.owl" \
+     --create-dirs -o "$MIRROR_DIR/cl-base.owl" \
      --retry 4 \
      --max-time 200 && \
-robot convert -i "$MIRROR_DIR/cl.owl" \
+robot convert -i "$MIRROR_DIR/cl-base.owl" \
      --format owl \
-     -o "$MIRROR_DIR/cl.owl"
+     -o "$MIRROR_DIR/cl-base.owl"
 
 echo "Downloading the latest PCL ontology..."
 curl -L "$OBO_BASE_URL/pcl/pcl-base.owl" \
-     --create-dirs -o "$MIRROR_DIR/pcl.owl" \
+     --create-dirs -o "$MIRROR_DIR/pcl-base.owl" \
      --retry 4 \
      --max-time 200 && \
-robot convert -i "$MIRROR_DIR/pcl.owl" \
+robot convert -i "$MIRROR_DIR/pcl-base.owl" \
      --format owl \
-     -o "$MIRROR_DIR/pcl.owl"
+     -o "$MIRROR_DIR/pcl-base.owl"
+
+echo "Merging CL and PCL ontologies to create a full CL ontology..."
+robot merge --input "$MIRROR_DIR/cl-base.owl" --input "$MIRROR_DIR/pcl-base.owl" -o "$MIRROR_DIR/cl.owl"
 
 echo "Downloading the latest LHMA ontology..."
 curl -L "https://data.bioontology.org/ontologies/LUNGMAP_H_CELL/submissions/6/download?apikey=8b5b7825-538d-40e0-9e9e-5ab9274a9aeb" \
diff --git a/src/enrichment/enrich-2d-ftu.js b/src/enrichment/enrich-2d-ftu.js
index 89987a7..5075e93 100644
--- a/src/enrichment/enrich-2d-ftu.js
+++ b/src/enrichment/enrich-2d-ftu.js
@@ -66,19 +66,6 @@ export function enrich2dFtuData(context) {
       push(inputPaths, clExtractPath);
     }
 
-    const pclEntitiesPath = collectEntities(context, 'pcl', baseInputPath);
-    if (!isFileEmpty(pclEntitiesPath)) {
-      info('Extracting PCL.');
-      const pclExtractPath = extractClassHierarchy(
-        context,
-        'pcl',
-        'http://purl.obolibrary.org/obo/CL_0000000',
-        pclEntitiesPath
-      );
-      logOutput(pclExtractPath);
-      push(inputPaths, pclExtractPath);
-    }
-
     info('Merging files:');
     for (const inputPath of inputPaths) {
       more(` -> ${inputPath}`);
diff --git a/src/enrichment/enrich-asct-b.js b/src/enrichment/enrich-asct-b.js
index 2c32b24..c77b21f 100644
--- a/src/enrichment/enrich-asct-b.js
+++ b/src/enrichment/enrich-asct-b.js
@@ -93,19 +93,6 @@ export function enrichAsctbData(context) {
       inputPaths.push(clExtractPath);
     }
 
-    const pclEntitiesPath = collectEntities(context, 'pcl', enrichedWithValidationPath);
-    if (!isFileEmpty(pclEntitiesPath)) {
-      info('Extracting PCL.');
-      const pclExtractPath = extractClassHierarchy(
-        context,
-        'pcl',
-        'http://purl.obolibrary.org/obo/CL_0000000',
-        pclEntitiesPath
-      );
-      logOutput(pclExtractPath);
-      inputPaths.push(pclExtractPath);
-    }
-
     const lmhaEntitiesPath = collectEntities(context, 'lmha', enrichedWithValidationPath);
     if (!isFileEmpty(lmhaEntitiesPath)) {
       info('Extracting LMHA.');
diff --git a/src/utils/get-cl-terms.sparql b/src/utils/get-cl-terms.sparql
index 9d3951d..f32cc28 100644
--- a/src/utils/get-cl-terms.sparql
+++ b/src/utils/get-cl-terms.sparql
@@ -3,14 +3,16 @@ WHERE {
    {
       SELECT (?s AS ?entity) WHERE { 
          ?s ?p ?o .
-         FILTER(STRSTARTS(STR(?s),"http://purl.obolibrary.org/obo/CL_"))
+         FILTER(STRSTARTS(STR(?s),"http://purl.obolibrary.org/obo/CL_") 
+            || STRSTARTS(STR(?s),"http://purl.obolibrary.org/obo/PCL_"))
       }
    }
    UNION
    {
       SELECT (?o AS ?entity) WHERE {
          ?s ?p ?o .
-         FILTER(STRSTARTS(STR(?o),"http://purl.obolibrary.org/obo/CL_"))
+         FILTER(STRSTARTS(STR(?o),"http://purl.obolibrary.org/obo/CL_") 
+            || STRSTARTS(STR(?o),"http://purl.obolibrary.org/obo/PCL_"))
       }
    }
 }
\ No newline at end of file
diff --git a/src/utils/get-pcl-terms.sparql b/src/utils/get-pcl-terms.sparql
deleted file mode 100644
index ce915be..0000000
--- a/src/utils/get-pcl-terms.sparql
+++ /dev/null
@@ -1,16 +0,0 @@
-SELECT DISTINCT ?entity
-WHERE {
-   {
-      SELECT (?s AS ?entity) WHERE { 
-         ?s ?p ?o .
-         FILTER(STRSTARTS(STR(?s),"http://purl.obolibrary.org/obo/PCL_"))
-      }
-   }
-   UNION
-   {
-      SELECT (?o AS ?entity) WHERE {
-         ?s ?p ?o .
-         FILTER(STRSTARTS(STR(?o),"http://purl.obolibrary.org/obo/PCL_"))
-      }
-   }
-}
\ No newline at end of file