Backmerge: #2539 - Export of unknown monomer to HELM doesn't work

Backmerge to master

api/tests/integration/ref/formats/helm_to_ket.py.out

@@ -1,6 +1,7 @@
 *** HELM to KET ***
 helm_aminoacids_variants.ket:SUCCEED
 helm_annotations.ket:SUCCEED
+helm_any_chem.ket:SUCCEED
 helm_chem_peptide.ket:SUCCEED
 helm_chem_rna.ket:SUCCEED
 helm_chem_rna_hydro.ket:SUCCEED

api/tests/integration/ref/formats/ket_to_helm.py.out

@@ -2,6 +2,7 @@
 dna_variants.ket:SUCCEED
 helm_aminoacids_variants.ket:SUCCEED
 helm_annotations.ket:SUCCEED
+helm_any_chem.ket:SUCCEED
 helm_chem_peptide.ket:SUCCEED
 helm_chem_rna.ket:SUCCEED
 helm_chem_rna_hydro.ket:SUCCEED

api/tests/integration/tests/formats/helm_to_ket.py

@@ -46,6 +46,7 @@ def find_diff(a, b):
     "helm_chem_rna_hydro": "CHEM1{[MCC]}|RNA1{R(U)P}$CHEM1,RNA1,1:pair-3:pair$$$V2.0",
     "helm_unsplit": "RNA1{[5Br-dU]}$$$$V2.0",
     "helm_smiles_no_ap": "CHEM1{[P(O)(O)(=O)O]}$$$$V2.0",
+    "helm_any_chem": "CHEM1{*}|CHEM2{*}$$$$V2.0",
 }
 
 lib = indigo.loadMonomerLibraryFromFile(

api/tests/integration/tests/formats/ket_to_helm.py

@@ -54,13 +54,18 @@ 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$$$V2.0",
-    "helm_unknown": "CHEM1{[C1([*:1])C([*:4])C([*:3])C1[*:2] |$;_R1;;_R4;;_R3;;_R2$|]}$$$$V2.0",
+    "helm_unknown": "CHEM1{*}$$$$V2.0",
     "helm_different_id": "PEPTIDE1{A}|RNA1{R(A)P}$$$$V2.0",
+    "helm_any_chem": "CHEM1{*}|CHEM2{*}$$$$V2.0",
 }
 
 for filename in sorted(helm_data.keys()):
     mol = indigo.loadKetDocumentFromFile(os.path.join(ref, filename + ".ket"))
-    helm = mol.helm(lib)
+    try:
+        helm = mol.helm(lib)
+    except IndigoException as e:
+        print("Test %s failed: %s" % (filename, getIndigoExceptionText(e)))
+        continue
     helm_ref = helm_data[filename]
     if helm_ref == helm:
         print(filename + ".ket:SUCCEED")

api/tests/integration/tests/formats/ref/helm_any_chem.ket

@@ -0,0 +1,102 @@
+{
+    "root": {
+        "nodes": [
+            {
+                "$ref": "monomer0"
+            },
+            {
+                "$ref": "monomer1"
+            }
+        ],
+        "templates": [
+            {
+                "$ref": "monomerTemplate-unknown_monomer_0"
+            },
+            {
+                "$ref": "monomerTemplate-unknown_monomer_1"
+            }
+        ]
+    },
+    "monomer0": {
+        "type": "monomer",
+        "id": "0",
+        "seqid": 1,
+        "position": {
+            "x": 0.000000,
+            "y": -0.000000
+        },
+        "alias": "unknown_monomer_0",
+        "templateId": "unknown_monomer_0"
+    },
+    "monomer1": {
+        "type": "monomer",
+        "id": "1",
+        "seqid": 2,
+        "position": {
+            "x": 0.000000,
+            "y": -1.500000
+        },
+        "alias": "unknown_monomer_1",
+        "templateId": "unknown_monomer_1"
+    },
+    "monomerTemplate-unknown_monomer_0": {
+        "type": "monomerTemplate",
+        "id": "unknown_monomer_0",
+        "class": "CHEM",
+        "alias": "unknown_monomer_0",
+        "unresolved": true,
+        "idtAliases": {
+            "base": "unknown_monomer_0"
+        },
+        "attachmentPoints": [
+            {
+                "attachmentAtom": -1,
+                "label": "R1"
+            },
+            {
+                "attachmentAtom": -1,
+                "label": "R2"
+            },
+            {
+                "attachmentAtom": -1,
+                "label": "R3"
+            },
+            {
+                "attachmentAtom": -1,
+                "label": "R4"
+            }
+        ],
+        "atoms": [],
+        "bonds": []
+    },
+    "monomerTemplate-unknown_monomer_1": {
+        "type": "monomerTemplate",
+        "id": "unknown_monomer_1",
+        "class": "CHEM",
+        "alias": "unknown_monomer_1",
+        "unresolved": true,
+        "idtAliases": {
+            "base": "unknown_monomer_1"
+        },
+        "attachmentPoints": [
+            {
+                "attachmentAtom": -1,
+                "label": "R1"
+            },
+            {
+                "attachmentAtom": -1,
+                "label": "R2"
+            },
+            {
+                "attachmentAtom": -1,
+                "label": "R3"
+            },
+            {
+                "attachmentAtom": -1,
+                "label": "R4"
+            }
+        ],
+        "atoms": [],
+        "bonds": []
+    }
+}

core/indigo-core/molecule/src/sequence_loader.cpp

@@ -1248,15 +1248,12 @@ void SequenceLoader::loadIdt(KetDocument& document)
                         checkAddTemplate(document, monomer_template);
                         single_monomer = monomer_template_id;
                         single_monomer_alias = monomer_template.getStringProp("alias");
-                        single_monomer_class = MonomerTemplates::classToStr(monomer_template.monomerClass());
                     }
                     else // IDT alias not found
                     {
-                        unresolved = true;
                         single_monomer = "unknown_monomer_with_idt_alias_" + idt_alias;
                         single_monomer_alias = idt_alias;
                         auto monomer_class = MonomerClass::CHEM;
-                        single_monomer_class = MonomerTemplates::classToStr(monomer_class);
                         // Unresoved monomer could be in any position
                         MonomerTemplate monomer_template(single_monomer, monomer_class, IdtAlias(idt_alias, idt_alias, idt_alias, idt_alias), true);
                         monomer_template.setStringProp("alias", idt_alias);
@@ -1302,6 +1299,14 @@ std::string SequenceLoader::readHelmMonomerAlias(KetDocument& document, MonomerC
     std::string monomer_alias;
     auto ch = _scanner.lookNext();
 
+    if (ch == '*')
+    {
+        if (monomer_class != MonomerClass::CHEM)
+            throw Error("'*' could be used only for CHEM monomers for now.");
+        _scanner.skip(1);
+        return "*";
+    }
+
     if (ch == '[')
     {
         _scanner.skip(1);
@@ -1692,6 +1697,7 @@ void SequenceLoader::loadHELM(KetDocument& document)
     std::string simple_polymer_type = "";
     int monomer_idx = 0;
     int prev_monomer_template_atom_idx = -1;
+    int unknown_count = 0;
     _unknown_ambiguous_count = 0;
     using polymer_map = std::map<std::string, std::map<int, size_t>>;
     polymer_map used_polymer_nums;
@@ -1753,6 +1759,18 @@ void SequenceLoader::loadHELM(KetDocument& document)
                     ch = _scanner.lookNext();
                     if (ch != '}')
                         throw Error("Unexpected symbol. Expected '}' but found '%c'.", ch); // only one monomer in chem
+
+                    auto& alias = std::get<0>(monomer_info);
+                    if (alias == "*") // if monomer_alias == "*"
+                    {
+                        alias = "unknown_monomer_" + std::to_string(unknown_count++);
+                        MonomerTemplate monomer_template(alias, MonomerClass::CHEM, IdtAlias(alias, alias, alias, alias), true);
+                        monomer_template.setStringProp("alias", alias);
+                        for (auto ap : {"R1", "R2", "R3", "R4"})
+                            monomer_template.AddAttachmentPoint(ap, -1);
+                        checkAddTemplate(document, monomer_template);
+                        _added_templates.emplace(monomer_class, alias);
+                    }
                     cur_polymer_map->second[monomer_idx] = addKetMonomer(document, monomer_info, monomer_class, pos);
                 }
                 else if (monomer_class == MonomerClass::AminoAcid)

core/indigo-core/molecule/src/sequence_saver.cpp

@@ -1285,7 +1285,12 @@ std::string SequenceSaver::saveHELM(KetDocument& document, std::vector<std::dequ
             if (monomer_class == MonomerClass::Base)
                 helm_string += '(';
             if (monomer->monomerType() == KetBaseMonomer::MonomerType::Monomer)
-                add_monomer(document, monomer, helm_string);
+            {
+                if (templates.at(monomer->templateId()).unresolved())
+                    helm_string += '*';
+                else
+                    add_monomer(document, monomer, helm_string);
+            }
             else if (monomer->monomerType() == KetBaseMonomer::MonomerType::AmbiguousMonomer)
             {
                 const auto& templ = variant_templates.at(monomer->templateId());