Merge pull request #1353 from microbiomedata/1342-additional-pathway-prefixes

Support pathway search with multiple prefixes

Author: naglepuff

nmdc_server/crud.py

@@ -1,3 +1,4 @@
+import re
 from datetime import datetime
 from typing import Any, Dict, List, Optional, Tuple, Type, TypeVar, cast
 from uuid import UUID
@@ -224,6 +225,14 @@ def list_omics_processing_data_objects(db: Session, id: str) -> Query:
     )
 
 
+# KEGG
+def get_pathway_prefix(term) -> Optional[str]:
+    pathway_prefixes = set(["map", "ko", "ec", "rn", "org"])
+    pathway_re = f"^({'|'.join(re.escape(p) for p in pathway_prefixes)})"
+    match = re.match(pathway_re, term)
+    return match.group(0) if match else None
+
+
 def list_ko_terms_for_module(db: Session, module: str) -> List[str]:
     q = db.query(models.KoTermToModule.term).filter(models.KoTermToModule.module.ilike(module))
     return [row[0] for row in q]
@@ -235,13 +244,24 @@ def list_ko_terms_for_pathway(db: Session, pathway: str) -> List[str]:
 
 
 def kegg_text_search(db: Session, query: str, limit: int) -> List[models.KoTermText]:
+    pathway_prefix = get_pathway_prefix(query)
+    term = query.replace(pathway_prefix, "map") if pathway_prefix else query
     q = (
         db.query(models.KoTermText)
-        .filter(models.KoTermText.text.ilike(f"%{query}%") | models.KoTermText.term.ilike(query))
+        .filter(models.KoTermText.text.ilike(f"%{term}%") | models.KoTermText.term.ilike(term))
         .order_by(models.KoTermText.term)
         .limit(limit)
     )
-    return list(q)
+    results = list(q)
+    if pathway_prefix:
+        default_pathway_prefix = "map"
+        # Transform pathway results to match given prefix. They are ingested with the
+        # 'map' prefix, but can searched for with various other prefixes.
+        for term_text in results:
+            if term_text.term.startswith(default_pathway_prefix):
+                term_text.term = term_text.term.replace(default_pathway_prefix, pathway_prefix)
+            term_text.text = term_text.text.replace(default_pathway_prefix, pathway_prefix)
+    return results
 
 
 # biosample

nmdc_server/query.py

@@ -292,10 +292,9 @@ def groups(self) -> Iterator[Tuple[str, Iterator[BaseConditionSchema]]]:
     def transform_condition(self, db, condition: BaseConditionSchema) -> List[BaseConditionSchema]:
         # Transform KEGG.(PATH|MODULE) queries into their respective ORTHOLOGY terms
         if condition.key == "Table.gene_function:id" and type(condition.value) is str:
-            if condition.value.startswith(KeggTerms.PATHWAY[0]):
-                searchable_name = condition.value.replace(
-                    KeggTerms.PATHWAY[0], KeggTerms.PATHWAY[1]
-                )
+            if any([condition.value.startswith(val) for val in KeggTerms.PATHWAY[0]]):
+                prefix = [val for val in KeggTerms.PATHWAY[0] if condition.value.startswith(val)][0]
+                searchable_name = condition.value.replace(prefix, KeggTerms.PATHWAY[1])
                 ko_terms = db.query(models.KoTermToPathway.term).filter(
                     models.KoTermToPathway.pathway.ilike(searchable_name)
                 )

nmdc_server/table.py

@@ -23,7 +23,15 @@
 
 class KeggTerms:
     ORTHOLOGY = ("KEGG.ORTHOLOGY:K", "K")
-    PATHWAY = ("KEGG.PATHWAY:MAP", "MAP")
+    PATHWAY = (
+        [
+            "KEGG.PATHWAY:MAP",
+            "KEGG.PATHWAY:EC",
+            "KEGG.PATHWAY:RN",
+            "KEGG.PATHWAY:KO",
+        ],
+        "MAP",
+    )
     MODULE = ("KEGG.MODULE:M", "M")
 
 

web/src/components/FilterKegg.vue

@@ -101,8 +101,8 @@ export default defineComponent({
         <p>
           KEGG Gene Function search filters results to
           samples that have at least one of the chosen KEGG terms.
-          Orthology, Pathway, and Module are supported.
-          Expected format: <code>K00000, M00000 or MAP00000</code>
+          Orthology, Module, and Pathway are supported.
+          Expected formats: <code>K00000, M00000, map00000, ko00000, rn00000, and ec00000</code>
         </p>
         <p class="text-subtitle-2">
           More information at <a href="https://www.genome.jp/kegg/">genome.jp/kegg</a>

web/src/encoding.ts

@@ -21,26 +21,54 @@ export interface FieldsData {
   encode?: (input: string) => string,
 }
 
-const KeggPrefix = {
+interface PrefixInfo {
+    pattern: RegExp;
+    short: Function;
+    long: Function;
+    urlBase: string;
+}
+
+const pathwayRegex = /^((map:?)|(path:?)|(ko:?)|(ec:?)|(rn:?)|(kegg.pathway:(map|path|ec|ko|rn)))(?=\d{5})/i;
+
+function pathwayPrefixShort(v: string) {
+  const match = v.match(pathwayRegex);
+  if (match) {
+    const prefix = match[8] ? match[8] : match[1].replace(/:$/, '');
+    return prefix.toLowerCase();
+  }
+  return 'map';
+}
+
+function pathwayPrefixLong(v: string) {
+  const match = v.match(pathwayRegex);
+  if (match) {
+    const prefix = match[7] ? match[7] : match[1].replace(match[1], `kegg.pathway:${match[1]}`);
+    return prefix.toUpperCase();
+  }
+  return 'KEGG.PATHWAY.MAP';
+}
+
+const KeggPrefix: Record<string, PrefixInfo> = {
   ORTHOLOGY: {
-    pattern: /^((ko?:?)|(kegg\.orthology:k))(?=\d{5})/i,
-    short: 'k',
-    long: 'KEGG.ORTHOLOGY:K',
+    pattern: /^((k:?)|(kegg\.orthology:k))(?=\d{5})/i,
+    short: () => 'k',
+    long: () => 'KEGG.ORTHOLOGY:K',
     urlBase: 'https://www.genome.jp/entry/',
   },
   PATHWAY: {
-    pattern: /^((map:?)|(path:?)|(kegg.pathway:map))(?=\d{5})/i,
-    short: 'map',
-    long: 'KEGG.PATHWAY:MAP',
+    pattern: pathwayRegex,
+    short: pathwayPrefixShort,
+    long: pathwayPrefixLong,
     urlBase: 'https://www.genome.jp/kegg-bin/show_pathway?',
   },
   MODULE: {
     pattern: /^((m:?)|(kegg.module:m))(?=\d{5})/i,
-    short: 'M',
-    long: 'KEGG.MODULE:M',
-    urlBase: 'https://www.genome.jp/brite/',
+    short: () => 'M',
+    long: () => 'KEGG.MODULE:M',
+    urlBase: 'https://www.kegg.jp/entry/',
   },
 };
+
 /**
  * Encode a string as either the long or short variant of
  * a KEGG identifier term.
@@ -51,7 +79,8 @@ function keggEncode(v: string, url = false) {
     const {
       pattern, short, long, urlBase,
     } = prefixes[i];
-    const transformed = v.replace(pattern, url ? short : long);
+    const replacement = url ? short(v) : long(v);
+    const transformed = v.replace(pattern, replacement);
     if (transformed !== v) {
       if (url) {
         return urlBase + transformed;