#1287 New S-Group type Query component level grouping (#1288)

epam · Sep 29, 2023 · e94605b · e94605b
1 parent 5a7bfaa
commit e94605b
Show file tree

Hide file tree

Showing 8 changed files with 298 additions and 5 deletions.
diff --git a/api/tests/integration/ref/formats/ket_with_query_components.py.out b/api/tests/integration/ref/formats/ket_with_query_components.py.out
@@ -0,0 +1,2 @@
+*** KET with query components ***
+ket_with_query_components.ket:SUCCEED
diff --git a/api/tests/integration/tests/formats/ket_with_query_components.py b/api/tests/integration/tests/formats/ket_with_query_components.py
@@ -0,0 +1,35 @@
+import difflib
+import os
+import sys
+
+
+def find_diff(a, b):
+    return "\n".join(difflib.unified_diff(a.splitlines(), b.splitlines()))
+
+
+sys.path.append(
+    os.path.normpath(
+        os.path.join(os.path.abspath(__file__), "..", "..", "..", "common")
+    )
+)
+from env_indigo import *  # noqa
+
+indigo = Indigo()
+indigo.setOption("json-saving-pretty", True)
+
+print("*** KET with query components ***")
+
+ref_path = joinPathPy("ref/", __file__)
+name = "ket_with_query_components.ket"
+filename = os.path.join(ref_path, name)
+
+mol = indigo.loadQueryMoleculeFromFile(filename)
+with open(filename, "r") as file:
+    ket_ref = file.read()
+ket = mol.json()
+diff = find_diff(ket_ref, ket)
+if not diff:
+    print(name + ":SUCCEED")
+else:
+    print(name + ":FAILED")
+    print(diff)
diff --git a/api/tests/integration/tests/formats/ref/ket_with_query_components.ket b/api/tests/integration/tests/formats/ref/ket_with_query_components.ket
@@ -0,0 +1,198 @@
+{
+    "root": {
+        "nodes": [
+            {
+                "$ref": "mol0"
+            },
+            {
+                "$ref": "mol1"
+            },
+            {
+                "$ref": "mol2"
+            }
+        ]
+    },
+    "mol0": {
+        "type": "molecule",
+        "atoms": [
+            {
+                "label": "C",
+                "location": [
+                    3.2669873237609865,
+                    -2.674999952316284,
+                    0.0
+                ]
+            },
+            {
+                "label": "C",
+                "location": [
+                    4.133012771606445,
+                    -2.174999952316284,
+                    0.0
+                ],
+                "queryProperties": {
+                    "connectivity": 2,
+                    "ringSize": 4
+                }
+            },
+            {
+                "label": "C",
+                "location": [
+                    5.016987323760986,
+                    -2.299999952316284,
+                    0.0
+                ]
+            },
+            {
+                "label": "C",
+                "location": [
+                    5.883012771606445,
+                    -1.7999999523162842,
+                    0.0
+                ]
+            }
+        ],
+        "bonds": [
+            {
+                "type": 1,
+                "atoms": [
+                    0,
+                    1
+                ]
+            },
+            {
+                "type": 1,
+                "atoms": [
+                    2,
+                    3
+                ]
+            },
+            {
+                "type": 1,
+                "atoms": [
+                    1,
+                    2
+                ]
+            }
+        ]
+    },
+    "mol1": {
+        "type": "molecule",
+        "atoms": [
+            {
+                "label": "C",
+                "location": [
+                    7.550000190734863,
+                    -3.674999952316284,
+                    0.0
+                ]
+            },
+            {
+                "label": "C",
+                "location": [
+                    8.050000190734864,
+                    -2.808974504470825,
+                    0.0
+                ]
+            },
+            {
+                "label": "C",
+                "location": [
+                    9.324999809265137,
+                    -3.299999952316284,
+                    0.0
+                ]
+            },
+            {
+                "label": "N",
+                "location": [
+                    8.61789321899414,
+                    -4.007106781005859,
+                    0.0
+                ]
+            }
+        ],
+        "bonds": [
+            {
+                "type": 1,
+                "atoms": [
+                    0,
+                    1
+                ]
+            },
+            {
+                "type": 1,
+                "atoms": [
+                    2,
+                    3
+                ]
+            }
+        ],
+        "sgroups": [
+            {
+                "type": "queryComponent",
+                "atoms": [
+                    0,
+                    1,
+                    2,
+                    3
+                ]
+            }
+        ]
+    },
+    "mol2": {
+        "type": "molecule",
+        "atoms": [
+            {
+                "label": "C",
+                "location": [
+                    3.0999999046325685,
+                    -6.375,
+                    0.0
+                ]
+            },
+            {
+                "label": "C",
+                "location": [
+                    3.8071067333221437,
+                    -7.082106590270996,
+                    0.0
+                ]
+            },
+            {
+                "label": "C",
+                "location": [
+                    2.9410812854766847,
+                    -7.582106590270996,
+                    0.0
+                ]
+            }
+        ],
+        "bonds": [
+            {
+                "type": 1,
+                "atoms": [
+                    0,
+                    1
+                ]
+            },
+            {
+                "type": 1,
+                "atoms": [
+                    1,
+                    2
+                ]
+            }
+        ],
+        "sgroups": [
+            {
+                "type": "queryComponent",
+                "atoms": [
+                    0,
+                    1,
+                    2
+                ]
+            }
+        ]
+    }
+}
diff --git a/core/indigo-core/molecule/ket_commons.h b/core/indigo-core/molecule/ket_commons.h
@@ -27,6 +27,7 @@
 #include "common/math/algebra.h"
 #include "graph/graph.h"
 #include "molecule/molecule_cip_calculator.h"
+#include "molecule/query_molecule.h"
 #include "reaction/base_reaction.h"
 
 namespace indigo
@@ -111,6 +112,11 @@ namespace indigo
             for (auto atom_idx : sgroup.atoms)
                 sg_set.insert(atom_idx);
         }
+        if (mol.isQueryMolecule())
+        {
+            QueryMolecule& qmol = static_cast<QueryMolecule&>(mol);
+            qmol.getComponentNeighbors(neighbors);
+        }
     }
 
     class KETSimpleObject : public MetaObject

diff --git a/core/indigo-core/molecule/molecule_json_loader.h b/core/indigo-core/molecule/molecule_json_loader.h
@@ -110,6 +110,7 @@ namespace indigo
         Molecule* _pmol;
         QueryMolecule* _pqmol;
         std::vector<EnhancedStereoCenter> _stereo_centers;
+        unsigned int components_count;
     };
 
 } // namespace indigo

diff --git a/core/indigo-core/molecule/src/molecule_json_loader.cpp b/core/indigo-core/molecule/src/molecule_json_loader.cpp
@@ -20,7 +20,8 @@ using namespace std;
 IMPL_ERROR(MoleculeJsonLoader, "molecule json loader");
 
 MoleculeJsonLoader::MoleculeJsonLoader(Document& ket)
-    : _mol_array(kArrayType), _mol_nodes(_mol_array), _meta_objects(kArrayType), _pmol(0), _pqmol(0), ignore_noncritical_query_features(false)
+    : _mol_array(kArrayType), _mol_nodes(_mol_array), _meta_objects(kArrayType), _pmol(0), _pqmol(0), ignore_noncritical_query_features(false),
+      components_count(0)
 {
     Value& root = ket["root"];
     Value& nodes = root["nodes"];
@@ -56,7 +57,7 @@ MoleculeJsonLoader::MoleculeJsonLoader(Document& ket)
 
 MoleculeJsonLoader::MoleculeJsonLoader(Value& mol_nodes)
     : _mol_nodes(mol_nodes), _meta_objects(kArrayType), _pmol(0), _pqmol(0), ignore_noncritical_query_features(false), ignore_no_chiral_flag(false),
-      skip_3d_chirality(false), treat_x_as_pseudoatom(false), treat_stereo_as(0)
+      skip_3d_chirality(false), treat_x_as_pseudoatom(false), treat_stereo_as(0), components_count(0)
 {
 }
 
@@ -672,6 +673,12 @@ void MoleculeJsonLoader::parseBonds(const rapidjson::Value& bonds, BaseMolecule&
                 }
             }
 
+            if (b.HasMember("customQuery"))
+            {
+                std::string customQuery = b["customQuery"].GetString();
+                // 2do process custom query
+            }
+
             if (b.HasMember("cip"))
             {
                 std::string cip = b["cip"].GetString();
@@ -844,11 +851,25 @@ void MoleculeJsonLoader::parseSGroups(const rapidjson::Value& sgroups, BaseMolec
     for (SizeType i = 0; i < sgroups.Size(); i++)
     {
         const Value& s = sgroups[i];
+        const Value& atoms = s["atoms"];
         std::string sg_type_str = s["type"].GetString(); // GEN, MUL, SRU, SUP
+        if (sg_type_str == "queryComponent")
+        {
+            if (_pqmol)
+            {
+                _pqmol->components.expandFill(_pqmol->components.size() + atoms.Size(), 0);
+                components_count++;
+                for (int j = 0; j < atoms.Size(); ++j)
+                {
+                    int atom_idx = atoms[j].GetInt();
+                    _pqmol->components[atom_idx] = components_count;
+                }
+            }
+            continue;
+        }
         int sg_type = SGroup::getType(sg_type_str.c_str());
         int grp_idx = mol.sgroups.addSGroup(sg_type);
         SGroup& sgroup = mol.sgroups.getSGroup(grp_idx);
-        const Value& atoms = s["atoms"];
         // add atoms
         std::unordered_set<int> sgroup_atoms;
         for (int j = 0; j < atoms.Size(); ++j)

diff --git a/core/indigo-core/molecule/src/molecule_json_saver.cpp b/core/indigo-core/molecule/src/molecule_json_saver.cpp
@@ -149,7 +149,19 @@ void MoleculeJsonSaver::saveSGroups(BaseMolecule& mol, JsonWriter& writer)
 {
     QS_DEF(Array<int>, sgs_sorted);
     _checkSGroupIndices(mol, sgs_sorted);
-    if (mol.countSGroups() > 0)
+    int sGroupsCount = mol.countSGroups();
+    bool componentDefined = false;
+    if (mol.isQueryMolecule())
+    {
+        QueryMolecule& qmol = static_cast<QueryMolecule&>(mol);
+        if (qmol.components.size() > 0 && qmol.components[0])
+        {
+            componentDefined = true;
+            sGroupsCount++;
+        }
+    }
+
+    if (sGroupsCount > 0)
     {
         writer.Key("sgroups");
         writer.StartArray();
@@ -160,6 +172,25 @@ void MoleculeJsonSaver::saveSGroups(BaseMolecule& mol, JsonWriter& writer)
             auto& sgrp = mol.sgroups.getSGroup(sg_idx);
             saveSGroup(sgrp, writer);
         }
+        // save queryComponent
+        if (mol.isQueryMolecule() && componentDefined)
+        {
+            QueryMolecule& qmol = static_cast<QueryMolecule&>(mol);
+            writer.StartObject();
+            writer.Key("type");
+            writer.String("queryComponent");
+            writer.Key("atoms");
+            writer.StartArray();
+            for (int i = 0; i < qmol.vertexCount(); i++)
+            {
+                if (qmol.components[i])
+                {
+                    writer.Int(i);
+                }
+            }
+            writer.EndArray();
+            writer.EndObject();
+        }
         writer.EndArray();
     }
 }

diff --git a/core/indigo-core/molecule/src/query_molecule.cpp b/core/indigo-core/molecule/src/query_molecule.cpp
@@ -2180,7 +2180,6 @@ void QueryMolecule::getComponentNeighbors(std::list<std::unordered_set<int>>& co
             componentAtoms[componentId].insert(i);
         }
     }
-    componentNeighbors.clear();
     for (auto elem : componentAtoms)
     {
         auto atoms = elem.second;