feat: SCHWED-6002 add SMILES to compound lib

modules/conop/pymod/export_compound.cc

@@ -126,6 +126,9 @@ void export_Compound() {
     .add_property("inchi_key",
                   make_function(&Compound::GetInchiKey, 
                                 return_value_policy<copy_const_reference>()))
+    .add_property("smiles",
+                  make_function(&Compound::GetSMILES,
+                                return_value_policy<copy_const_reference>()))
   ;
   
   class_<AtomSpec>("AtomSpec", no_init)

modules/conop/src/compound.hh

@@ -149,6 +149,7 @@ public:
     name_(),
     inchi_(),
     inchi_key_(),
+    smiles_(),
     atom_specs_(),
     bond_specs_(),
     chem_class_(),
@@ -252,6 +253,10 @@ public:
 
   const String& GetInchiKey() { return inchi_key_; }
 
+  void SetSMILES(const String& smiles) { smiles_=smiles; }
+
+  const String& GetSMILES() { return smiles_; }
+
   const BondSpecList& GetBondSpecs() const {
     return bond_specs_;
   }
@@ -281,6 +286,7 @@ private:
   String                       name_;
   String                       inchi_;
   String                       inchi_key_;
+  String                       smiles_;
   AtomSpecList                 atom_specs_;
   BondSpecList                 bond_specs_;
   mol::ChemClass               chem_class_;

modules/conop/src/compound_lib.cc

@@ -62,7 +62,8 @@ const char* CREATE_CMD[]={
 "  pdb_modified      TIMESTAMP,                                                 "
 "  name              VARCHAR(256),                                              "
 "  inchi_code        TEXT,                                                      "
-"  inchi_key         TEXT                                                       "
+"  inchi_key         TEXT,                                                      "
+"  smiles            TEXT                                                       "
 ");",
 " CREATE UNIQUE INDEX IF NOT EXISTS commpound_tlc_index ON chem_compounds       "
 "                                  (tlc, dialect)",
@@ -98,8 +99,9 @@ const char* CREATE_CMD[]={
 
 
 const char* INSERT_COMPOUND_STATEMENT="INSERT INTO chem_compounds               "
-"        (tlc, olc, dialect, chem_class, chem_type, formula, pdb_initial, pdb_modified, name, inchi_code, inchi_key) "
-" VALUES (?, ?, ?, ?, ?, ?, DATE(?), DATE(?), ?, ?, ?)";
+"        (tlc, olc, dialect, chem_class, chem_type, formula, pdb_initial,       "
+"         pdb_modified, name, inchi_code, inchi_key, smiles) "
+" VALUES (?, ?, ?, ?, ?, ?, DATE(?), DATE(?), ?, ?, ?, ?)";
 
 const char* INSERT_ATOM_STATEMENT="INSERT INTO atoms                            "
 "        (compound_id, name, alt_name, element, is_aromatic, stereo_conf,       "
@@ -262,6 +264,8 @@ void CompoundLib::AddCompound(const CompoundPtr& compound)
                       compound->GetInchi().length(), NULL);
     sqlite3_bind_text(stmt, 11, compound->GetInchiKey().c_str(),
                       compound->GetInchiKey().length(), NULL);
+    sqlite3_bind_text(stmt, 12, compound->GetSMILES().c_str(),
+                      compound->GetSMILES().length(), NULL);
   } else {
     LOG_ERROR(sqlite3_errmsg(db_->ptr));
     sqlite3_finalize(stmt);
@@ -405,6 +409,13 @@ CompoundLibPtr CompoundLib::Load(const String& database, bool readonly)
                             &stmt, NULL);
   lib->inchi_available_ = retval==SQLITE_OK;
   sqlite3_finalize(stmt);
+  // check if SMILES are available
+  aq="SELECT smiles FROM chem_compounds LIMIT 1";
+  retval=sqlite3_prepare_v2(lib->db_->ptr, aq.c_str(),
+                            static_cast<int>(aq.length()),
+                            &stmt, NULL);
+  lib->smiles_available_ = retval==SQLITE_OK;
+  sqlite3_finalize(stmt);
 
   lib->creation_date_ = lib->GetCreationDate();
   lib->ost_version_used_ = lib->GetOSTVersionUsed();
@@ -471,17 +482,24 @@ CompoundPtr CompoundLib::FindCompound(const String& id,
     return i->second;
   }
   String query="SELECT id, tlc, olc, chem_class, dialect, formula";
-  int col_offset = 0;
+  int col_offset_inchi = 0;
+  int col_offset_smiles = 0;
   if(chem_type_available_) {
     query+=", chem_type";
-    col_offset+=1;
+    col_offset_inchi+=1;
+    col_offset_smiles+=1;
     if(name_available_) {
       query+=", name";
-      col_offset+=1;
+      col_offset_inchi+=1;
+      col_offset_smiles+=1;
     }
   }
   if(inchi_available_) {
     query+=", inchi_code, inchi_key";
+    col_offset_smiles+=2;
+  }
+  if(smiles_available_) {
+    query+=", smiles";
   }
 
   query+=" FROM chem_compounds"
@@ -515,15 +533,22 @@ CompoundPtr CompoundLib::FindCompound(const String& id,
         }
       }
       if (inchi_available_) {
-        const char* inchi_code=reinterpret_cast<const char*>(sqlite3_column_text(stmt, 6+col_offset));
+        const char* inchi_code=reinterpret_cast<const char*>(sqlite3_column_text(stmt, 6+col_offset_inchi));
         if (inchi_code) {
           compound->SetInchi(inchi_code);
         }
-        const char* inchi_key=reinterpret_cast<const char*>(sqlite3_column_text(stmt, 6+col_offset+1));
+        const char* inchi_key=reinterpret_cast<const char*>(sqlite3_column_text(stmt, 6+col_offset_inchi+1));
         if (inchi_key) {
           compound->SetInchiKey(inchi_key);
         }
       }
+      if (smiles_available_) {
+        const char* smiles=reinterpret_cast<const char*>(sqlite3_column_text(stmt, 6+col_offset_smiles));
+        if (smiles) {
+          compound->SetSMILES(smiles);
+        }
+      }
+
       // Load atoms and bonds      
       this->LoadAtomsFromDB(compound, pk);
       this->LoadBondsFromDB(compound, pk);
@@ -548,6 +573,7 @@ CompoundLib::CompoundLib():
   chem_type_available_(false),
   name_available_(),
   inchi_available_(),
+  smiles_available_(),
   creation_date_(),
   ost_version_used_() { }
 

modules/conop/src/compound_lib.hh

@@ -58,6 +58,7 @@ private:
   bool                      chem_type_available_; // wether pdbx_type is available in db
   bool                      name_available_; // wether name is available in db
   bool                      inchi_available_; //whether inchi is available in db
+  bool                      smiles_available_; //whether smiles are available in db
   Date                      creation_date_;
   String                    ost_version_used_;
 };

modules/conop/tests/test_complib.py

@@ -7,8 +7,8 @@ import tempfile
 
 class TestCompLib(unittest.TestCase):
 
-    def test_three_vs_five_letter_code(self):
-
+    @classmethod
+    def setUpClass(cls):
         prefix_path = ost.GetPrefixPath()
         chemdict_tool_path = os.path.join(prefix_path, "bin", "chemdict_tool")
         if not os.path.exists(chemdict_tool_path):
@@ -18,8 +18,11 @@ class TestCompLib(unittest.TestCase):
         complib_path = os.path.join(tmp_dir.name, "test_complib.dat")
         cmd = [chemdict_tool_path, "create", compounds_path, complib_path]
         subprocess.run(cmd)
+        cls.complib = conop.CompoundLib.Load(complib_path)
+        tmp_dir.cleanup()
 
-        complib = conop.CompoundLib.Load(complib_path)
+    def test_three_vs_five_letter_code(self):
+        complib = self.complib
 
         comp_001 = complib.FindCompound("001")
         comp_hello = complib.FindCompound("hello")
@@ -29,6 +32,12 @@ class TestCompLib(unittest.TestCase):
         self.assertFalse(comp_hello is None)
         self.assertTrue(comp_yolo is None)
 
+    def test_smiles(self):
+        complib = self.complib
+        comp_001 = complib.FindCompound("001")
+        self.assertTrue(comp_001.smiles == "COc1cc(cc(c1OC)OC)C(C(=O)N2CCCC[C@H]2C(=O)O[C@@H](CCCc3ccccc3)CCCc4cccnc4)(F)F")
+
+
 if __name__ == "__main__":
     from ost import testutils
     testutils.RunTests()

modules/conop/tests/test_compound.py

@@ -21,6 +21,7 @@ class TestCompound(unittest.TestCase):
         self.assertEqual(compound.inchi,
                         "1S/C3H7NO2/c1-2(4)3(5)6/h2H,4H2,1H3,(H,5,6)/t2-/m0/s1")
         self.assertEqual(compound.inchi_key, "QNAYBMKLOCPYGJ-REOHCLBHSA-N")
+        self.assertEqual(compound.smiles, "C[C@@H](C(=O)O)N"  )
 
      
 if __name__=='__main__':

modules/io/src/mol/chemdict_parser.cc

@@ -85,6 +85,9 @@ void ChemdictParser::OnDataRow(const StarLoopDesc& header,
       compound_->SetInchi(columns[indices_[DESC]].substr(6).str());
     } else if (columns[indices_[DESC_TYPE]] == StringRef("InChIKey", 8)) {
       compound_->SetInchiKey(columns[indices_[DESC]].str());
+    } else if (columns[indices_[DESC_TYPE]] == StringRef("SMILES_CANONICAL", 16) &&
+              columns[indices_[PROGRAM]] == StringRef("OpenEye OEToolkits", 18)) {
+      compound_->SetSMILES(columns[indices_[DESC]].str());
     }
   }
 }

modules/io/src/mol/chemdict_parser.hh

@@ -80,7 +80,8 @@ private:
     ATOM_ID2=1,
     BOND_ORDER=2,
     DESC_TYPE=0,
-    DESC=1
+    DESC=1,
+    PROGRAM=2,
   } PropIndex;
   char                                    last_;
   int                                     indices_[10];