#2539 - Export of unknown monomer to HELM doesn't work (#2690)

epam · Dec 17, 2024 · cd267b4 · cd267b4
1 parent 4e626c4
commit cd267b4
Show file tree

Hide file tree

Showing 4 changed files with 139 additions and 1 deletion.
diff --git a/api/tests/integration/ref/formats/ket_to_helm.py.out b/api/tests/integration/ref/formats/ket_to_helm.py.out
@@ -19,4 +19,5 @@ helm_rna_without_base.ket:SUCCEED
 helm_simple_rna.ket:SUCCEED
 helm_smiles.ket:SUCCEED
 helm_smiles_sugar.ket:SUCCEED
+helm_unknown.ket:SUCCEED
 rna_variants.ket:SUCCEED
diff --git a/api/tests/integration/tests/formats/ket_to_helm.py b/api/tests/integration/tests/formats/ket_to_helm.py
@@ -54,6 +54,7 @@ def find_diff(a, b):
     "helm_molecule_2418": "PEPTIDE1{A}|CHEM1{[C1C=CC=CC=1[*:1] |$;;;;;;_R1$|]}$PEPTIDE1,CHEM1,1:R1-1:R1$$$V2.0",
     "helm_chem_rna_hydro": "CHEM1{[MCC]}|RNA1{R(U)P}$CHEM1,RNA1,1:pair-3:pair$$$V2.0",
     "helm_monomer_molecule_direct": "PEPTIDE1{A}|CHEM1{[C(=C)N[*:1] |$;;;_R1$|]}$PEPTIDE1,CHEM1,1:R2-1:R1$$$V2.0",
+    "helm_unknown": "CHEM1{[C1([*:1])C([*:4])C([*:3])C1[*:2] |$;_R1;;_R4;;_R3;;_R2$|]}$$$$V2.0",
 }
 
 for filename in sorted(helm_data.keys()):

diff --git a/api/tests/integration/tests/formats/ref/helm_unknown.ket b/api/tests/integration/tests/formats/ref/helm_unknown.ket
@@ -0,0 +1,128 @@
+{
+    "root": {
+        "nodes": [
+            {
+                "$ref": "monomer489"
+            }
+        ],
+        "connections": [],
+        "templates": [
+            {
+                "$ref": "monomerTemplate-unknown_monomer_with_idt_alias_Unknown"
+            }
+        ]
+    },
+    "monomer489": {
+        "type": "monomer",
+        "id": "489",
+        "position": {
+            "x": 13.213368034362787,
+            "y": -10.06310758590698
+        },
+        "alias": "Unknown",
+        "templateId": "unknown_monomer_with_idt_alias_Unknown",
+        "seqid": 0
+    },
+    "monomerTemplate-unknown_monomer_with_idt_alias_Unknown": {
+        "type": "monomerTemplate",
+        "atoms": [
+            {
+                "label": "C",
+                "location": [
+                    0,
+                    0,
+                    0
+                ]
+            },
+            {
+                "label": "C",
+                "location": [
+                    1,
+                    1,
+                    1
+                ]
+            },
+            {
+                "label": "C",
+                "location": [
+                    2,
+                    2,
+                    2
+                ]
+            },
+            {
+                "label": "C",
+                "location": [
+                    3,
+                    3,
+                    3
+                ]
+            }
+        ],
+        "bonds": [
+            {
+                "type": 1,
+                "atoms": [
+                    0,
+                    1
+                ]
+            },
+            {
+                "type": 1,
+                "atoms": [
+                    1,
+                    2
+                ]
+            },
+            {
+                "type": 1,
+                "atoms": [
+                    2,
+                    3
+                ]
+            },
+            {
+                "type": 1,
+                "atoms": [
+                    0,
+                    3
+                ]
+            }
+        ],
+        "class": "CHEM",
+        "id": "unknown_monomer_with_idt_alias_Unknown",
+        "fullName": "Unknown",
+        "alias": "Unknown",
+        "attachmentPoints": [
+            {
+                "attachmentAtom": 0,
+                "leavingGroup": {
+                    "atoms": []
+                }
+            },
+            {
+                "attachmentAtom": 1,
+                "leavingGroup": {
+                    "atoms": []
+                }
+            },
+            {
+                "attachmentAtom": 2,
+                "leavingGroup": {
+                    "atoms": []
+                }
+            },
+            {
+                "attachmentAtom": 3,
+                "leavingGroup": {
+                    "atoms": []
+                }
+            }
+        ],
+        "idtAliases": {
+            "base": "Unknown"
+        },
+        "unresolved": true,
+        "naturalAnalogShort": ""
+    }
+}
diff --git a/core/indigo-core/molecule/src/sequence_saver.cpp b/core/indigo-core/molecule/src/sequence_saver.cpp
@@ -1206,7 +1206,15 @@ void SequenceSaver::add_monomer(KetDocument& document, const std::unique_ptr<Ket
                 auto& ap = sa.attachment_points.at(ap_id);
                 int leaving_atom = ap.lvidx;
                 int ap_idx = getAttachmentOrder(ap.apid.ptr()) + 1;
-                pmol->resetAtom(leaving_atom, ELEM_RSITE);
+                if (leaving_atom >= 0)
+                {
+                    pmol->resetAtom(leaving_atom, ELEM_RSITE);
+                }
+                else
+                {
+                    leaving_atom = pmol->addAtom(ELEM_RSITE);
+                    pmol->addBond(ap.aidx, leaving_atom, BOND_SINGLE);
+                }
                 pmol->allowRGroupOnRSite(leaving_atom, ap_idx);
             }
             sgroups.remove(i);