steps 2 añadir debugs

Author: NIU1636242

curation-pipeline/src/components/steps/Step2GenomeTF.jsx

@@ -58,32 +58,35 @@ async function fetchNuccoreTaxInfo(acc) {
 }
 
 async function fetchTaxonomyLineageEx(taxid) {
-  const url = `${ENTREZ_BASE}/esummary.fcgi?db=taxonomy&id=${encodeURIComponent(
+  // efetch es el endpoint que devuelve LineageEx de forma consistente
+  const url = `${ENTREZ_BASE}/efetch.fcgi?db=taxonomy&id=${encodeURIComponent(
     taxid
   )}&retmode=json`;
+
   const j = await fetchJson(url, { isNcbi: true });
 
-  const node = j?.result?.[String(taxid)];
-  if (!node) return null;
+  // efetch devuelve algo tipo { result: [ { TaxId, ScientificName, Rank, LineageEx: [...] } ] }
+  const rec = Array.isArray(j?.result) ? j.result[0] : null;
+  if (!rec) return null;
 
-  const lineageEx = Array.isArray(node.lineageex) ? node.lineageex : [];
+  const lineageEx = Array.isArray(rec.LineageEx) ? rec.LineageEx : [];
 
+  // LineageEx son los ancestros; añadimos el nodo final (leaf)
   const path = lineageEx.map((x) => ({
-    taxid: x?.taxid ? String(x.taxid) : "",
-    name: x?.scientificname || "",
-    rank: x?.rank || "no rank",
+    taxid: x?.TaxId ? String(x.TaxId) : "",
+    name: x?.ScientificName || "",
+    rank: x?.Rank || "no rank",
   }));
 
-  // Afegim el node final (taxid consultat) si no hi és
   const leaf = {
-    taxid: String(taxid),
-    name: node.scientificname || "",
-    rank: node.rank || "no rank",
+    taxid: rec.TaxId ? String(rec.TaxId) : String(taxid),
+    name: rec.ScientificName || "",
+    rank: rec.Rank || "no rank",
   };
 
   if (!path.length || path[path.length - 1].taxid !== leaf.taxid) path.push(leaf);
 
-  // Neteja de duplicats i buits
+  // limpiar duplicados/basura
   const seen = new Set();
   const cleaned = [];
   for (const n of path) {
@@ -92,7 +95,7 @@ async function fetchTaxonomyLineageEx(taxid) {
     cleaned.push(n);
   }
 
-  // El format que guardem ja porta el parent_taxonomy_id per facilitar inserts
+  // devolver chain con parent_taxonomy_id encadenado
   return cleaned.map((n, i) => ({
     taxonomy_id: n.taxid,
     name: n.name,

curation-pipeline/src/components/steps/Step7CurationInfo.jsx

@@ -360,63 +360,19 @@ WHERE NOT EXISTS (
     }
 
     // Taxonomia + link a genome
-    // Inserta la cadena completa (ancestros + hoja)
-    // Pero la BD "indexa" hasta species: si el leaf es strain/subspecies, enlazamos el genoma a la species.
-
-    // DEBUG
-    console.log("STEP7 taxonomyData keys", Object.keys(taxonomyData || {}));
-    console.log("STEP7 byAcc keys", Object.keys(taxByAcc));
-
     const taxByAcc = taxonomyData?.byAccession || {};
 
-    const rankScore = (r) => {
-      const x = String(r || "no rank").toLowerCase();
-      const order = {
-        "no rank": 0,
-        superkingdom: 1,
-        kingdom: 2,
-        phylum: 3,
-        class: 4,
-        order: 5,
-        family: 6,
-        genus: 7,
-        species: 8,
-        subspecies: 9,
-        strain: 10,
-      };
-      return order[x] ?? 0;
-    };
-
     for (const acc of accessions) {
-
-      console.log("STEP7 for acc", acc, taxByAcc?.[acc]);
-
       const tInfo = taxByAcc?.[acc];
-      const chainRaw = Array.isArray(tInfo?.chain) ? tInfo.chain : [];
-      if (!chainRaw.length) continue;
-
-      // Aseguramos orden de general -> específico.
-      // Si viene al revés, invertimos (p.ej. empieza en strain y termina en phylum).
-      const chainOrdered = [...chainRaw];
-      if (chainOrdered.length >= 2) {
-        const first = rankScore(chainOrdered[0]?.rank);
-        const last = rankScore(chainOrdered[chainOrdered.length - 1]?.rank);
-        if (first > last) chainOrdered.reverse();
-      }
-
-      // Si NCBI devuelve strain/subspecies, recortamos la cadena para enlazar el genoma a species.
-      let chain = chainOrdered;
-      const idxSpecies = chainOrdered.findIndex((n) => String(n?.rank || "").toLowerCase() === "species");
-      if (idxSpecies >= 0) chain = chainOrdered.slice(0, idxSpecies + 1);
+      const chain = Array.isArray(tInfo?.chain) ? tInfo.chain : [];
+      if (!chain.length) continue;
 
-      // Insert/Update de cada nodo y su parent_id (FK a core_taxonomy.id)
       for (let i = 0; i < chain.length; i++) {
         const node = chain[i];
         const taxid = String(node.taxonomy_id || "").trim();
-        if (!taxid) continue;
-
         const name = String(node.name || "").trim();
         const rank = String(node.rank || "no rank").trim();
+        if (!taxid) continue;
 
         const parentTaxid = i > 0 ? String(chain[i - 1].taxonomy_id || "").trim() : "";
         const parentIdExpr = parentTaxid
@@ -445,7 +401,6 @@ WHERE taxonomy_id='${esc(taxid)}';
     `.trim());
       }
 
-      // Enlazamos el genoma al leaf elegido (species si existe; si no, el último del chain)
       const leafTaxid = String(chain[chain.length - 1]?.taxonomy_id || "").trim();
       if (leafTaxid) {
         sql.push(`