properly connect DNA/RNA

modules/conop/src/heuristic_builder.cc

@@ -187,6 +187,7 @@ HeuristicBuilder::HeuristicBuilder():
       emap_[def_entry.abbrev]=entry;
     }
   }
+  default_nucleotide_=LookupResEntry("G").first;
   LOG_DEBUG("done importing internal tables");
 }
 
@@ -321,6 +322,10 @@ void ConnectPrevNext(HeuristicBuilder* builder,mol::ResidueHandle res0,
       res0_ret.first=builder->DefaultPeptide();
       res0_ret.second=true;
     }
+    if (res0.FindAtom("C3'") && res0.FindAtom("P")) {
+      res0_ret.first=builder->DefaultNucleotide();
+      res0_ret.second=true;
+    }
   }
 
   if(!res1_ret.second) {
@@ -329,6 +334,10 @@ void ConnectPrevNext(HeuristicBuilder* builder,mol::ResidueHandle res0,
       res1_ret.first=builder->DefaultPeptide();
       res1_ret.second=true;
     }
+    if (res1.FindAtom("C3'") && res1.FindAtom("P")) {
+      res1_ret.first=builder->DefaultNucleotide();
+      res1_ret.second=true;
+    }
   }
 
   if(res0_ret.second && res1_ret.second) {
@@ -348,16 +357,19 @@ void ConnectPrevNext(HeuristicBuilder* builder,mol::ResidueHandle res0,
       if(flag) {
         if (builder->DoesPeptideBondExist(res0_atom, res1_atom)) {
           editor.Connect(res0_atom,res1_atom);
-          res0.SetIsProtein(true);
-          res1.SetIsProtein(true);
+          if (res0_ret.first.GetChemClass().IsPeptideLinking()) {
+            res0.SetIsProtein(true);
+            res1.SetIsProtein(true);
+          }
         }
       } else {
         if (builder->DoesPeptideBondExist(res1_atom, res0_atom)) {
           editor.Connect(res1_atom, res0_atom);
-          res0.SetIsProtein(true);
-          res1.SetIsProtein(true);
+          if (res0_ret.first.GetChemClass().IsPeptideLinking()) {
+            res0.SetIsProtein(true);
+            res1.SetIsProtein(true);
+          }
         }
-
       }
     }
   } else {

modules/conop/src/heuristic_builder.hh

@@ -110,6 +110,7 @@ public:
   virtual void FillAtomProps(mol::AtomHandle atom);
 
   const detail::ConnResEntry& DefaultPeptide() const {return default_peptide_;}
+  const detail::ConnResEntry& DefaultNucleotide() const {return default_nucleotide_;}
 protected:
   void ConnectivityFromAtomNames(const mol::ResidueHandle& res,
                                  detail::ConnResEntry& centry,
@@ -117,6 +118,7 @@ protected:
 private:
   detail::ConnResEntryMap emap_;
   detail::ConnResEntry default_peptide_;
+  detail::ConnResEntry default_nucleotide_;
 };
 
 

modules/conop/src/heuristic_connect_table.hh

@@ -267,31 +267,31 @@ CONN_DEF_ENTRY def_entry_table[]={
    {{0,0,0,0,""}},1,
    {0},1
   },
-  {"Adenosin","A",'?', mol::ChemClass(mol::ChemClass::DNA_LINKING),
+  {"Adenosin","DA",'?', mol::ChemClass(mol::ChemClass::DNA_LINKING),
    {"P","O5'","C5'","C4'","O4'","C3'","O3'","C2'","C1'","O1P","O2P","N9","C8","N7","C5","C6","N6","N1","C2","N3","C4","O2'"},12,
    {{-2,1},{1,2},{2,3},{3,4},{4,5},{4,6},{6,7},{6,8},{8,9},{5,9},{1,10},{1,11},{8,22},{9,12},{7,-3}},15,
    {{0,0,0,0,""}},1,
    {0},1
   },
-  {"Cytosin","C",'?', mol::ChemClass(mol::ChemClass::DNA_LINKING),
+  {"Cytosin","DC",'?', mol::ChemClass(mol::ChemClass::DNA_LINKING),
    {"P","O5'","C5'","C4'","O4'","C3'","O3'","C2'","C1'","O1P","O2P","O2'"},12,
    {{-2,1},{1,2},{2,3},{3,4},{4,5},{4,6},{6,7},{6,8},{8,9},{5,9},{1,10},{1,11},{8,12},{7,-3}},14,
    {{0,0,0,0,""}},1,
    {0},1
   },
-  {"Guanidin","G",'?', mol::ChemClass(mol::ChemClass::DNA_LINKING),
+  {"Guanidin","DG",'?', mol::ChemClass(mol::ChemClass::DNA_LINKING),
    {"P","O5'","C5'","C4'","O4'","C3'","O3'","C2'","C1'","O1P","O2P","O2'"},12,
    {{-2,1},{1,2},{2,3},{3,4},{4,5},{4,6},{6,7},{6,8},{8,9},{5,9},{7,-3},{1,10},{1,11},{8,12}},14,
    {{0,0,0,0,""}},1,
    {0},1
   },
-  {"Thymidin","T",'?', mol::ChemClass(mol::ChemClass::DNA_LINKING),
+  {"Thymidin","DT",'?', mol::ChemClass(mol::ChemClass::DNA_LINKING),
    {"P","O5'","C5'","C4'","O4'","C3'","O3'","C2'","C1'","O1P","O2P","O2'"},12,
    {{-2,1},{1,2},{2,3},{3,4},{4,5},{4,6},{6,7},{6,8},{8,9},{5,9},{7,-3},{1,10},{8,12},{1,11}},15,
    {{0,0,0,0,""}},1,
    {0},1
   },
-  {"Uracil","U",'?', mol::ChemClass(mol::ChemClass::DNA_LINKING),
+  {"Uracil","DU",'?', mol::ChemClass(mol::ChemClass::DNA_LINKING),
    {"P","O5'","C5'","C4'","O4'","C3'","O3'","C2'","C1'","O1P","O2P","O2'"},12,
    {{-2,1},{1,2},{2,3},{3,4},{4,5},{4,6},{6,7},{6,8},{8,9},{5,9},{7,-3},{1,10},{1,11},{8,12}},14,
    {{0,0,0,0,""}},1,