Merge remote-tracking branch 'origin/main' into pathfinder-newsystem

data/templateGroups.json

@@ -21,6 +21,7 @@
                ],
     "predicate": ["affects"],
     "qualifiers": {
+      "qualified_predicate": "causes",
       "object_aspect_qualifier": "activity_or_abundance",
       "object_direction_qualifier": "increased"
     }, 
@@ -40,6 +41,7 @@
                ],
     "predicate": ["affects"],
     "qualifiers": {
+      "qualified_predicate": "causes",
       "object_aspect_qualifier": "activity_or_abundance",
       "object_direction_qualifier": "decreased"
     }, 

src/edge_manager.ts

@@ -207,11 +207,8 @@ export default class QueryEdgeManager {
       const objectIDs = [record.object.original, record.object.curie, ...record.object.equivalentCuries];
 
       // there must be at least a minimal intersection
-      const subjectMatch =
-        subjectIDs.some((curie) => execSubjectCuries.includes(curie)) || execSubjectCuries.length === 0;
-      const objectMatch = objectIDs.some((curie) => execObjectCuries.includes(curie)) || execObjectCuries.length === 0;
-
-      //if both ends match then keep record
+      const subjectMatch = subjectIDs.some((curie) => execSubjectCuries.includes(curie));
+      const objectMatch = objectIDs.some((curie) => execObjectCuries.includes(curie));
 
       // Don't keep self-edges
       const selfEdge = [...subjectIDs].some((curie) => objectIDs.includes(curie));

src/graph/kg_node.ts

@@ -8,12 +8,14 @@ export interface KGNodeInfo {
   curies: string[];
   primaryCurie: string;
   qNodeID: string;
+  originalCurie?: string;
 }
 
 export default class KGNode {
   id: string;
   primaryCurie: string;
   qNodeID: string;
+  originalCurie: string;
   curies: string[];
   names: string[];
   semanticType: string[];
@@ -36,6 +38,9 @@ export default class KGNode {
     this.targetNodes = new Set();
     this.sourceQNodeIDs = new Set();
     this.targetQNodeIDs = new Set();
+
+    // store original curie to output `query_id bte#815`
+    this.originalCurie = info.originalCurie;
   }
 
   addSourceNode(kgNodeID: string): void {

src/graph/knowledge_graph.ts

@@ -14,10 +14,42 @@ import KGNode from './kg_node';
 import KGEdge from './kg_edge';
 import { BTEGraphUpdate } from './graph';
 import { APIDefinition } from '@biothings-explorer/types';
+import { Telemetry } from '@biothings-explorer/utils';
 
 const debug = Debug('bte:biothings-explorer-trapi:KnowledgeGraph');
 
-const NON_ARRAY_ATTRIBUTES = ['biolink:knowledge_level', 'biolink:agent_type'];
+const NON_ARRAY_ATTRIBUTES = ['biolink:knowledge_level', 'biolink:agent_type', 'biolink:evidence_count'];
+
+interface SpecialAttributeHandlers {
+  [attribute_type_id: string]: (value: Set<string | number>, kgEdge: KGEdge) => TrapiAttribute['value'];
+}
+
+const SPECIAL_ATTRIBUTE_HANDLERS: SpecialAttributeHandlers = {
+  'biolink:max_research_phase': (value, kgEdge) => {
+    // Special handling for max research phase
+    const phase_map = {
+      '-1.0': 'not_provided',
+      '0.5': 'pre_clinical_research_phase',
+      '1.0': 'clinical_trial_phase_1',
+      '2.0': 'clinical_trial_phase_2',
+      '3.0': 'clinical_trial_phase_3',
+      '4.0': 'clinical_trial_phase_4',
+    };
+    function map_phase(val: string) {
+      let new_val = phase_map[val];
+      if (typeof new_val !== 'undefined') return new_val;
+
+      const source = Object.values(kgEdge.sources).find((src) => typeof src.primary_knowledge_source !== 'undefined')
+        .primary_knowledge_source.resource_id;
+      const err = new Error(
+        `Unrecognized research phase (${val}) from ${source} ${kgEdge.subject} > ${kgEdge.predicate} > ${kgEdge.object}`,
+      );
+      Telemetry.captureException(err);
+      return 'not_provided';
+    }
+    return Array.from(value as Set<string>).map(map_phase);
+  },
+};
 
 export default class KnowledgeGraph {
   nodes: {
@@ -117,13 +149,19 @@ export default class KnowledgeGraph {
 
     Object.entries(kgEdge.attributes).forEach(([key, value]) => {
       if (key === 'edge-attributes') return;
-      // if (key == 'edge-attributes') return;
+
+      let formatted_value: TrapiAttribute['value'] = NON_ARRAY_ATTRIBUTES.includes(key)
+        ? Array.from(value as Set<string>).reduce((acc, val) => acc + val)
+        : Array.from(value as Set<string>);
+
+      if (key in SPECIAL_ATTRIBUTE_HANDLERS) {
+        formatted_value = SPECIAL_ATTRIBUTE_HANDLERS[key](value as Set<string | number>, kgEdge);
+      }
+
       attributes.push({
         attribute_type_id: key,
-        value: // technically works for numbers as well
-          NON_ARRAY_ATTRIBUTES.includes(key)
-            ? [...(value as Set<string>)].reduce((acc, val) => acc + val)
-            : Array.from(value as Set<string>),
+        // technically works for numbers as well
+        value: formatted_value,
         //value_type_id: 'bts:' + key,
       });
     });

src/index.ts

@@ -1,4 +1,4 @@
-import MetaKG from '@biothings-explorer/smartapi-kg';
+import MetaKG, { SmartAPIQueryResult } from '@biothings-explorer/smartapi-kg';
 import path from 'path';
 import QueryGraph from './query_graph';
 import KnowledgeGraph from './graph/knowledge_graph';
@@ -30,6 +30,7 @@ import { QueryHandlerOptions } from '@biothings-explorer/types';
 import BTEGraph from './graph/graph';
 import QEdge from './query_edge';
 import { Telemetry } from '@biothings-explorer/utils';
+import { enrichTrapiResultsWithPfocrFigures } from './results_assembly/pfocr';
 
 // Exports for external availability
 export * from './types';
@@ -73,14 +74,21 @@ export default class TRAPIQueryHandler {
 
   async findUnregisteredAPIs() {
     const configListAPIs = this.options.apiList['include'];
-    const smartapiRegistry = await fs.readFile(this.path, { encoding: 'utf8' });
+
+    let smartapiRegistry: SmartAPIQueryResult;
+    if (this.options.smartapi) {
+      smartapiRegistry = this.options.smartapi;
+    } else {
+      const file = await fs.readFile(this.path, "utf-8");
+      smartapiRegistry = JSON.parse(file);
+    }
 
     const smartapiIds: string[] = [];
     const inforesIds: string[] = [];
     const unregisteredAPIs: string[] = [];
 
     // TODO typing for smartapiRegistration
-    JSON.parse(smartapiRegistry).hits.forEach((smartapiRegistration) => {
+    smartapiRegistry.hits.forEach((smartapiRegistration) => {
       smartapiIds.push(smartapiRegistration._id);
       inforesIds.push(smartapiRegistration.info?.['x-translator']?.infores);
     });
@@ -96,14 +104,23 @@ export default class TRAPIQueryHandler {
     return unregisteredAPIs;
   }
 
-  _loadMetaKG(): MetaKG {
-    const metaKG = new MetaKG(this.path, this.predicatePath);
+  async _loadMetaKG(): Promise<MetaKG> {
     debug(
       `Query options are: ${JSON.stringify({
         ...this.options,
         schema: this.options.schema ? this.options.schema.info.version : 'not included',
+        metakg: "",
+        smartapi: ""
       })}`,
     );
+
+    if (this.options.metakg) {
+      const metaKG = new MetaKG(undefined, undefined, (this.options as any).metakg);
+      metaKG.filterKG(this.options);
+      return metaKG;
+    }
+
+    const metaKG = new MetaKG(this.path, this.predicatePath);
     debug(`SmartAPI Specs read from path: ${this.path}`);
     metaKG.constructMetaKGSync(this.includeReasoner, this.options);
     return metaKG;
@@ -302,6 +319,18 @@ export default class TRAPIQueryHandler {
     this.finalizedResults = fixedResults;
   }
 
+  appendOriginalCuriesToResults(results: TrapiResult[]): void {
+    results.forEach(result => {
+      Object.entries(result.node_bindings).forEach(([_, bindings]) => {
+        bindings.forEach(binding => {
+          if (this.bteGraph.nodes[binding.id].originalCurie && this.bteGraph.nodes[binding.id].originalCurie !== binding.id) {
+            binding.query_id = this.bteGraph.nodes[binding.id].originalCurie;
+          }
+        })
+      })
+    })
+  }
+
   async addQueryNodes(): Promise<void> {
     const qNodeIDsByOriginalID: Map<string, TrapiQNode> = new Map();
     const curiesToResolve = [
@@ -323,6 +352,7 @@ export default class TRAPIQueryHandler {
         this.bteGraph.nodes[resolvedEntity.primaryID] = new KGNode(resolvedEntity.primaryID, {
           primaryCurie: resolvedEntity.primaryID,
           qNodeID: qNodeIDsByOriginalID[originalCurie],
+          originalCurie: originalCurie,
           curies: resolvedEntity.equivalentIDs,
           names: resolvedEntity.labelAliases,
           semanticType: category ? [category] : ['biolink:NamedThing'],
@@ -629,7 +659,7 @@ export default class TRAPIQueryHandler {
     const span1 = Telemetry.startSpan({ description: 'loadMetaKG' });
 
     debug('Start to load metakg.');
-    const metaKG = this._loadMetaKG();
+    const metaKG = await this._loadMetaKG();
     if (!metaKG.ops.length) {
       let error: string;
       if (this.options.smartAPIID) {
@@ -710,8 +740,13 @@ export default class TRAPIQueryHandler {
     this.createSubclassSupportGraphs();
     // prune bteGraph
     this.bteGraph.prune(this.finalizedResults, this.auxGraphs);
+    // add original curies to results
+    this.appendOriginalCuriesToResults(this.finalizedResults);
     this.bteGraph.notify();
 
+    // Attempt to enrich results with PFOCR figures
+    this.logs = [...this.logs, ...(await enrichTrapiResultsWithPfocrFigures(this.getResponse()))];
+
     span3?.finish();
 
     // check primary knowledge sources