rename __gri_refseq to _gri_refseq

Author: mlin

bindings/python/genomicsqlite.py

@@ -130,7 +130,7 @@ class ReferenceSequence(NamedTuple):
 def get_reference_sequences_by_rid(
     con: sqlite3.Connection, assembly: Optional[str] = None, schema: Optional[str] = None
 ) -> Dict[int, ReferenceSequence]:
-    table = "__gri_refseq"
+    table = "_gri_refseq"
     if schema:
         table = f"{schema}.{table}"
     sql = (

loaders/sam_into_sqlite.cc

@@ -244,7 +244,7 @@ int main(int argc, char *argv[]) {
         // TODO: allow --append
         string ddl =
             "CREATE TABLE " + table_prefix +
-            "reads(rowid INTEGER PRIMARY KEY, flag INTEGER NOT NULL, rid INTEGER REFERENCES __gri_refseq(_gri_rid), pos INTEGER, endpos INTEGER, "
+            "reads(rowid INTEGER PRIMARY KEY, flag INTEGER NOT NULL, rid INTEGER REFERENCES _gri_refseq(_gri_rid), pos INTEGER, endpos INTEGER, "
             "mapq INTEGER, cigar TEXT, rnext INTEGER, pnext INTEGER, tlen INTEGER, "
             "rg_id INTEGER REFERENCES " +
             table_prefix + "readgroups(rg_id))";

loaders/vcf_into_sqlite.cc

@@ -49,7 +49,7 @@ string schemaDDL(const string &table_prefix, vector<map<string, string>> &info_h
 
     OStringStream out;
     out << "CREATE TABLE " << table_prefix
-        << "variants (variant_rowid INTEGER NOT NULL PRIMARY KEY, rid INTEGER NOT NULL REFERENCES __gri_refseq(_gri_rid), "
+        << "variants (variant_rowid INTEGER NOT NULL PRIMARY KEY, rid INTEGER NOT NULL REFERENCES _gri_refseq(_gri_rid), "
            "POS INTEGER NOT NULL, rlen INTEGER NOT NULL, ID_jsarray TEXT DEFAULT '[]', REF TEXT NOT NULL, "
            "ALT_jsarray TEXT NOT NULL DEFAULT '[]', QUAL REAL, FILTER_jsarray";
 

src/genomicsqlite.cc

@@ -459,12 +459,12 @@ static string gri_refseq_ddl(const string &schema) {
         schema_prefix = schema + ".";
     }
     ostringstream out;
-    out << "CREATE TABLE IF NOT EXISTS " << schema_prefix << "__gri_refseq"
+    out << "CREATE TABLE IF NOT EXISTS " << schema_prefix << "_gri_refseq"
         << "(_gri_rid INTEGER NOT NULL PRIMARY KEY, gri_refseq_name TEXT NOT NULL, gri_assembly TEXT,"
         << " gri_refget_id TEXT UNIQUE, gri_refseq_length INTEGER NOT NULL, gri_refseq_meta_json TEXT NOT NULL DEFAULT '{}', "
         << "UNIQUE(gri_assembly,gri_refseq_name))"
-        << ";\nCREATE INDEX IF NOT EXISTS " << schema_prefix << "__gri_refseq_name ON "
-        << schema_prefix << "__gri_refseq(gri_refseq_name)";
+        << ";\nCREATE INDEX IF NOT EXISTS " << schema_prefix << "_gri_refseq_name ON "
+        << schema_prefix << "_gri_refseq(gri_refseq_name)";
     return out.str();
 }
 
@@ -684,7 +684,7 @@ static void sqlfn_genomic_range_rowids_safe_sql(sqlite3_context *ctx, int argc,
 }
 
 /**************************************************************************************************
- * reference sequence metadata (__gri_refseq) helpers
+ * reference sequence metadata (_gri_refseq) helpers
  **************************************************************************************************/
 
 string PutGenomicReferenceSequenceSQL(const string &name, sqlite3_int64 length,
@@ -700,7 +700,7 @@ string PutGenomicReferenceSequenceSQL(const string &name, sqlite3_int64 length,
         out << gri_refseq_ddl(schema) << ";\n";
     }
     out << "INSERT INTO " << schema_prefix
-        << "__gri_refseq(_gri_rid,gri_refseq_name,gri_assembly,gri_refget_id,gri_refseq_length,gri_refseq_meta_json) VALUES("
+        << "_gri_refseq(_gri_rid,gri_refseq_name,gri_assembly,gri_refget_id,gri_refseq_length,gri_refseq_meta_json) VALUES("
         << (rid >= 0 ? std::to_string(rid) : "NULL") << "," << sqlquote(name) << ","
         << (assembly.empty() ? "NULL" : sqlquote(assembly)) << ","
         << (refget_id.empty() ? "NULL" : sqlquote(refget_id)) << "," << std::to_string(length)
@@ -996,7 +996,7 @@ GetGenomicReferenceSequencesByRid(sqlite3 *dbconn, const string &assembly, const
 
     string query =
         "SELECT _gri_rid, gri_refseq_name, gri_refseq_length, gri_assembly, gri_refget_id, gri_refseq_meta_json FROM " +
-        schema_prefix + "__gri_refseq";
+        schema_prefix + "_gri_refseq";
     if (!assembly.empty()) {
         query += " WHERE gri_assembly = ?";
     }

test/test_gri.py

@@ -263,7 +263,7 @@ def test_refseq():
     assert len(refseq_by_rid) == 195
 
     query = (
-        "SELECT __gri_refseq._gri_rid, rid, beg, end, id FROM exons, __gri_refseq WHERE exons.rid = gri_refseq_name AND exons._rowid_ IN "
+        "SELECT _gri_refseq._gri_rid, rid, beg, end, id FROM exons, _gri_refseq WHERE exons.rid = gri_refseq_name AND exons._rowid_ IN "
         + genomicsqlite.genomic_range_rowids_sql(con, "exons")
     )
     print("\n" + query)

test/test_vcf.py

@@ -30,7 +30,7 @@ def test_gnomad_sites_small(tmp_path):
     con = genomicsqlite.connect(dbfile, read_only=True)
 
     query = (
-        "SELECT variant_rowid, id_jsarray FROM (SELECT _gri_rid AS rid FROM __gri_refseq WHERE gri_refseq_name=?1) AS query, gnomad_variants WHERE variant_rowid IN"
+        "SELECT variant_rowid, id_jsarray FROM (SELECT _gri_rid AS rid FROM _gri_refseq WHERE gri_refseq_name=?1) AS query, gnomad_variants WHERE variant_rowid IN"
         + genomicsqlite.genomic_range_rowids_sql(con, "gnomad_variants", "query.rid")
     )
     rs671 = ("chr12", 111803912, 111804012)
@@ -57,7 +57,7 @@ def test_gnomad_sites_small(tmp_path):
     results_rowids = set(vt[0] for vt in results)
     assert next(vt for vt in results if vt[1] and "rs671" in vt[1])
 
-    control = "SELECT variant_rowid FROM gnomad_variants NATURAL JOIN __gri_refseq WHERE gri_refseq_name = ? AND NOT ((pos+rlen) < ? OR pos > ?)"
+    control = "SELECT variant_rowid FROM gnomad_variants NATURAL JOIN _gri_refseq WHERE gri_refseq_name = ? AND NOT ((pos+rlen) < ? OR pos > ?)"
     control_rowids = set(vt[0] for vt in con.execute(control, rs671))
     assert len(control_rowids) == 22
     assert results_rowids == control_rowids