Added to post-merge to export.sh script a command that generates gene…

…ric-physical-entity-map.json (required for PC app-ui web app).

Changed premerge, Converter to use datasource's id as the xmlbase instead of pc2 base.
Started DrugbankCleaner - todo...

README.md

@@ -83,7 +83,8 @@ To see available commands and options, run:
     bash cpath2.sh
 
 In order to create a new cpath2 instance, define or update the metadata.json, 
-prepare input data archives (see below how), and run 
+prepare input data archives (see below how), also install `jq`, `gunzip`, 
+and run:
 
     bash cpath2.sh --build 2>&1 >build.log &
 

src/main/java/cpath/cleaner/DrugbankCleaner.java

@@ -0,0 +1,27 @@
+package cpath.cleaner;
+
+import cpath.service.api.Cleaner;
+
+import java.io.InputStream;
+import java.io.OutputStream;
+
+public class DrugbankCleaner implements Cleaner {
+
+  /*
+  * drugbank biopax data uses the following values in xref.db properties,
+  * which biopax validator reports as "Unknown";
+  * so we need to replace with the corresponding standard prefix/name from bioregistry.io (or identifiers.org):
+  GenBank Gene Database -> "genbank"
+  GenBank Protein Database -> "genbank" (numeric IDs, not like it's in ncbiprotein)
+  Therapeutic Targets Database ->
+  Guide to Pharmacology ->
+  HUGO Gene Nomenclature Committee (HGNC) -> "hgnc.symbol" (or "HGNC Symbol")
+  IUPHAR ->
+  Drugs Product Database (DPD) ->
+   */
+
+  @Override
+  public void clean(InputStream data, OutputStream cleanedData) {
+    //TODO: implement
+  }
+}

src/main/java/cpath/service/ConsoleApplication.java

@@ -351,9 +351,9 @@ private void postmerge() throws IOException {
     // generate the "Detailed" pathway data file:
     createDetailedBiopax(mainModel, index);
 
-    // generate the export.sh script (to run Paxtools commands for exporting the BioPAX files to other formats)
+    // Generate export.sh script (to convert the data/model to other formats)
     LOG.info("writing 'export.sh' script to convert the BioPAX models to SIF, GSEA, SBGN...");
-    final String commonPrefix = service.settings().exportArchivePrefix(); //e.g., PathwayCommons13
+    final String commonPrefix = service.settings().exportArchivePrefix();
     writer = new PrintWriter(new OutputStreamWriter(Files.newOutputStream(
       Paths.get(service.settings().exportScriptFile())), StandardCharsets.UTF_8));
     writer.println("#!/bin/sh");
@@ -364,7 +364,7 @@ private void postmerge() throws IOException {
     writer.println("# sh export.sh &");
 
     //write commands to the script file for 'All' and 'Detailed' BioPAX input files:
-//    writeScriptCommands(service.settings().biopaxFileName("Detailed"), writer, true);
+    //writeScriptCommands(service.settings().biopaxFileName("Detailed"), writer, true);
     writeScriptCommands(service.settings().biopaxFileName("All"), writer, true);
 
     //rename SIF files that were cut from corresponding extended SIF (.txt) ones
@@ -379,7 +379,11 @@ private void postmerge() throws IOException {
     writer.println(String.format("%s %s '%s' '%s' %s 2>&1 &", javaRunPaxtools, "summarize",
       service.settings().biopaxFileName("All"), "physical_entities.json", "--uri-ids"));
     writer.println("wait");
-    writer.println("gzip pathways.txt *.json");
+    writer.println("""
+        gunzip -c physical_entities.json.gz | jq -cS 'map(select(.generic)) | reduce .[] as $o ({}; . + {($o.uri): {name: $o.name, label:$o.label, synonyms:$o."hgnc.symbol"}})' > generic-physical-entity-map.json
+        """);
+    writer.println("wait");
+    writer.println("gzip pathways.txt physical_entities.json");
     writer.println("echo \"All done.\"");
     writer.close();
     LOG.info("postmerge: done.");

src/main/java/cpath/service/PreMerger.java

@@ -110,7 +110,8 @@ void premerge() {
               + "; skipping for this data source...");
             return; // skip due to the error
           }
-          converter.setXmlBase(xmlBase);
+//          converter.setXmlBase(xmlBase);
+          converter.setXmlBase(datasource.getIdentifier()+":"); //todo: test if it works/looks better than with instance's xmlBase...
         } else {
           log.info("premerge(), Converter class is not defined for " + mid);
         }

work/make_generic_pe_map.bash

@@ -0,0 +1,8 @@
+# After the PC data build is done,
+# generates generic-physical-entity-map.json (for PC app-ui webapp).
+#
+# Prerequisites:
+#  - gunzip
+#  - jq (https://stedolan.github.io/jq/; or install with npm: npm install hjson -g)
+gunzip -c "file:downloads/physical_entities.json.gz" | jq -cS 'map(select(.generic)) | reduce .[] as $o ({}; . + {($o.uri): {name: $o.name, label:$o.label, synonyms:$o."hgnc.symbol"}})' > downloads/generic-physical-entity-map.json
+echo "Generated downloads/generic-physical-entity-map.json"