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

Author: CHERRY-ui8

graphgen/models/searcher/db/ncbi_searcher.py

@@ -89,11 +89,11 @@ def _infer_molecule_type_detail(accession: Optional[str], gene_type: Optional[in
         if accession:
             if accession.startswith(("NM_", "XM_")):
                 return "mRNA"
-            elif accession.startswith(("NC_", "NT_")):
+            if accession.startswith(("NC_", "NT_")):
                 return "genomic DNA"
-            elif accession.startswith(("NR_", "XR_")):
+            if accession.startswith(("NR_", "XR_")):
                 return "RNA"
-            elif accession.startswith("NG_"):
+            if accession.startswith("NG_"):
                 return "genomic region"
         # Fallback: infer from gene type if available
         if gene_type is not None:
@@ -215,7 +215,7 @@ def _extract_metadata_from_genbank(result: dict, accession: str):
             """Extract metadata from GenBank format (title, features, organism, etc.)."""
             with Entrez.efetch(db="nuccore", id=accession, rettype="gb", retmode="text") as handle:
                 record = SeqIO.read(handle, "genbank")
-                
+
                 result["title"] = record.description
                 result["molecule_type_detail"] = (
                     "mRNA" if accession.startswith(("NM_", "XM_")) else
@@ -238,7 +238,7 @@ def _extract_metadata_from_genbank(result: dict, accession: str):
 
                 if not result.get("organism") and 'organism' in record.annotations:
                     result["organism"] = record.annotations['organism']
-            
+
             return result
 
         def _extract_sequence_from_fasta(result: dict, accession: str):
@@ -249,7 +249,10 @@ def _extract_sequence_from_fasta(result: dict, accession: str):
                     result["sequence"] = str(fasta_record.seq)
                     result["sequence_length"] = len(fasta_record.seq)
             except Exception as fasta_exc:
-                logger.warning("Failed to extract sequence from accession %s using FASTA format: %s", accession, fasta_exc)
+                logger.warning(
+                    "Failed to extract sequence from accession %s using FASTA format: %s",
+                    accession, fasta_exc
+                )
                 result["sequence"] = None
                 result["sequence_length"] = None
             return result

scripts/search/build_db/build_rna_blast_db.sh

@@ -44,12 +44,13 @@ if [ "${DB_SELECTION}" = "list" ]; then
     echo "Available RNAcentral database subsets:"
     echo ""
     echo "Fetching list from RNAcentral FTP..."
-    curl -s "${RNACENTRAL_BY_DB_URL}/" | \
-        grep -oE '<a href="[^"]*\.fasta">' | \
+    listing=$(curl -s "${RNACENTRAL_BY_DB_URL}/")
+    echo "${listing}" | \
+        grep -oE '<a href="[^\"]*\.fasta">' | \
         sed 's/<a href="//;s/">//' | \
         sort | \
         while read db; do
-            size=$(curl -s "${RNACENTRAL_BY_DB_URL}/" | grep -A 1 "${db}" | grep -oE '[0-9.]+[GMK]' | head -1 || echo "unknown")
+            size=$(echo "${listing}" | grep -A 1 "${db}" | grep -oE '[0-9.]+[GMK]' | head -1 || echo "unknown")
             echo "  - ${db%.fasta}: ${size}"
         done
     echo ""
@@ -76,16 +77,14 @@ wget -q "${RELEASE_NOTES_URL}" 2>/dev/null || {
 
 if [ -f "${RELEASE_NOTES}" ]; then
     # Try to extract version from release notes (first line usually contains version info)
-    RELEASE=$(head -1 "${RELEASE_NOTES}" | grep -oE '[0-9]+\.[0-9]+' | head -1 | tr -d '.' || date +%Y%m%d)
-    if [ -z "${RELEASE}" ] || [ "${RELEASE}" = "$(date +%Y%m%d)" ]; then
-        RELEASE=$(date +%Y%m%d)
-        echo "Using date as release identifier: ${RELEASE}"
-    else
-        echo "RNAcentral release: ${RELEASE}"
-    fi
-else
+    RELEASE=$(head -1 "${RELEASE_NOTES}" | grep -oE '[0-9]+\.[0-9]+' | head -1 | tr -d '.')
+fi
+
+if [ -z "${RELEASE}" ]; then
     RELEASE=$(date +%Y%m%d)
     echo "Using date as release identifier: ${RELEASE}"
+else
+    echo "RNAcentral release: ${RELEASE}"
 fi
 
 # Download RNAcentral FASTA file