feat: Find compound by SMILES or InChI code/key

modules/conop/doc/compoundlib.rst

@@ -58,11 +58,17 @@ built with OST 1.5.0 or later can be loaded.
     
     Create a new compound library
     
-  .. method:: FindCompound(tlc, dialect='PDB')
+  .. method:: FindCompound(id, dialect='PDB', by="tlc")
   
-    Lookup compound by its three-letter-code, e.g ALA. If no compound with that 
-    name exists, the function returns None. Compounds are cached after they have 
-    been loaded with FindCompound. To delete the compound cache, use 
+    Lookup a compound. By default the compound is searched by its
+    three-letter-code, e.g ALA. This can be changed with the `by` argument.
+    The following keys are available: "tlc" (three-letter-code or compound ID),
+    "inchi_code", "inchi_key" and "smiles".
+
+    If no compound with that name exists, the function returns None.
+
+    Compounds are cached after they have been loaded with FindCompound.
+    To delete the compound cache, use
     :meth:`ClearCache`.
     
     :returns: The found compound

modules/conop/pymod/export_compound.cc

@@ -70,9 +70,10 @@ char get_chemtype(CompoundPtr compound)
 }
 
 CompoundPtr find_compound(CompoundLibPtr comp_lib, 
-                          const String& tlc, const String& dialect)
+                          const String& id, const String& dialect,
+                          const String& by="tlc")
 {
-  return comp_lib->FindCompound(tlc, tr_dialect(dialect));
+  return comp_lib->FindCompound(id, tr_dialect(dialect), by);
 }
 
 bool is_residue_complete(CompoundLibPtr comp_lib,
@@ -151,7 +152,7 @@ void export_Compound() {
   class_<CompoundLib>("CompoundLib", no_init)
     .def("Load", &CompoundLib::Load, arg("readonly")=true).staticmethod("Load")
     .def("FindCompound", &find_compound, 
-         (arg("tlc"), arg("dialect")="PDB"))
+         (arg("id"), arg("dialect")="PDB", arg("by")="tlc"))
     .def("IsResidueComplete", &is_residue_complete, (arg("residue"), 
                                                      arg("check_hydrogens")=false,
                                                      arg("dialect")="PDB"))

modules/conop/src/compound_lib.cc

@@ -70,8 +70,14 @@ const char* CREATE_CMD[]={
 "  inchi_key         TEXT,                                                      "
 "  smiles            TEXT                                                       "
 ");",
-" CREATE UNIQUE INDEX IF NOT EXISTS commpound_tlc_index ON chem_compounds       "
+" CREATE UNIQUE INDEX IF NOT EXISTS compound_tlc_index ON chem_compounds        "
 "                                  (tlc, dialect)",
+" CREATE INDEX IF NOT EXISTS compound_smiles_index ON chem_compounds            "
+"                                  (smiles, dialect)",
+" CREATE INDEX IF NOT EXISTS compound_inchi_code_index ON chem_compounds        "
+"                                  (inchi_code, dialect)",
+" CREATE INDEX IF NOT EXISTS compound_inchi_key_index ON chem_compounds         "
+"                                  (inchi_key, dialect)",
 "CREATE TABLE IF NOT EXISTS atoms (                                             "
 " id                 INTEGER PRIMARY KEY AUTOINCREMENT,                         "
 " compound_id        INTEGER REFERENCES chem_compounds (id) ON DELETE CASCADE,  "
@@ -498,18 +504,37 @@ void CompoundLib::LoadBondsFromDB(CompoundPtr comp, int pk) const {
 }
 
 CompoundPtr CompoundLib::FindCompound(const String& id, 
-                                      Compound::Dialect dialect) const {
-  CompoundMap::const_iterator i=compound_cache_.find(id);
+                                      Compound::Dialect dialect,
+                                      const String& by) const {
+
+  // Validate "by" argument
+  std::set<std::string> allowed_keys{"tlc", "inchi_code", "inchi_key"};
+  if(smiles_available_) {
+     allowed_keys.insert("smiles");
+  }
+  if (allowed_keys.find(by) == allowed_keys.end()) {
+    std::stringstream msg;
+    msg << "Invalid 'by' key: " << by;
+    throw ost::Error(msg.str());
+  }
+
+  // Check cache
+  String cache_key = by + "_" + id;
+  CompoundMap::const_iterator i=compound_cache_.find(cache_key);
   if (i!=compound_cache_.end()) {
+    LOG_DEBUG("Retrieved compound " << cache_key << " from cache");
     return i->second;
   }
+
+  // Build the query
   String query="SELECT id, tlc, olc, chem_class, dialect, formula, chem_type, name, inchi_code, inchi_key";
   if(smiles_available_) {
     query+=", smiles";
   }
-
   query+=" FROM chem_compounds"
-         " WHERE tlc=? AND dialect='"+String(1, char(dialect))+"'";
+         " WHERE " + by + "=? AND dialect='"+String(1, char(dialect))+"'";
+
+  // Run the query
   sqlite3_stmt* stmt;
   int retval=sqlite3_prepare_v2(db_->ptr, query.c_str(), 
                                 static_cast<int>(query.length()),
@@ -553,7 +578,7 @@ CompoundPtr CompoundLib::FindCompound(const String& id,
       // Load atoms and bonds      
       this->LoadAtomsFromDB(compound, pk);
       this->LoadBondsFromDB(compound, pk);
-      compound_cache_.insert(std::make_pair(compound->GetID(), compound));
+      compound_cache_.insert(std::make_pair(cache_key, compound));
       sqlite3_finalize(stmt);
       return compound;   
     }

modules/conop/src/compound_lib.hh

@@ -39,7 +39,8 @@ public:
   ~CompoundLib();
   
   virtual CompoundPtr FindCompound(const String& id, 
-                                   Compound::Dialect dialect) const;
+                                   Compound::Dialect dialect,
+                                   const String& by="tlc") const;
   void AddCompound(const CompoundPtr& compound);
   CompoundLibPtr Copy(const String& filename) const;
   void ClearCache();

modules/conop/src/compound_lib_base.hh

@@ -13,7 +13,8 @@ class DLLEXPORT_OST_CONOP CompoundLibBase {
 public:
   virtual ~CompoundLibBase() {}
   virtual CompoundPtr FindCompound(const String& id, 
-                                   Compound::Dialect dialect) const = 0;
+                                   Compound::Dialect dialect,
+                                   const String& by="tlc") const = 0;
 
   bool IsResidueComplete(const ost::mol::ResidueHandle& res, 
                          bool check_hydrogens, 

modules/conop/src/minimal_compound_lib.cc

@@ -42,8 +42,15 @@ CompoundMap MinimalCompoundLib::InitCompounds() {
 
 
 CompoundPtr MinimalCompoundLib::FindCompound(const String& id, 
-                                             Compound::Dialect dialect) const
+                                             Compound::Dialect dialect,
+                                             const String& by) const
 {
+   if (by != "tlc") {
+     // Only tlc is supported by the minimal compound lib
+     std::stringstream msg;
+     msg << "Invalid 'by' key: " << by;
+     throw ost::Error(msg.str());
+   }
    CompoundMap::const_iterator i = MinimalCompoundLib::compounds_.find(id); 
    if (i != MinimalCompoundLib::compounds_.end()) {
      return i->second;

modules/conop/src/minimal_compound_lib.hh

@@ -17,7 +17,8 @@ public:
     CompoundLibBase()
   {}
   virtual CompoundPtr FindCompound(const String& id, 
-                                   Compound::Dialect dialect) const;
+                                   Compound::Dialect dialect,
+                                   const String& by="tlc") const;
 private:
   static CompoundMap InitCompounds();
   // since this information is never going to change, it is shared