#2015 Import/Export of variant monomers from Fasta/Sequence

Add variant monomers support for Sequence/Fasta. Add UT
epam · Aug 14, 2024 · a6d4c91 · a6d4c91
1 parent 179c0bc
commit a6d4c91
Show file tree

Hide file tree

Showing 39 changed files with 23,451 additions and 12,739 deletions.
diff --git a/api/c/indigo/src/indigo_molecule.cpp b/api/c/indigo/src/indigo_molecule.cpp
@@ -568,11 +568,10 @@ CEXPORT int indigoLoadSequence(int source, const char* seq_type, int library)
         IndigoObject& lib_obj = self.getObject(library);
         SequenceLoader loader(IndigoScanner::get(obj), IndigoMonomerLibrary::get(lib_obj));
 
-        std::unique_ptr<IndigoMolecule> molptr = std::make_unique<IndigoMolecule>();
+        std::unique_ptr<IndigoKetDocument> docptr = std::make_unique<IndigoKetDocument>();
 
-        Molecule& mol = molptr->mol;
-        loader.loadSequence(mol, seq_type);
-        return self.addObject(molptr.release());
+        loader.loadSequence(docptr->get(), seq_type);
+        return self.addObject(docptr.release());
     }
     INDIGO_END(-1);
 }
@@ -619,11 +618,10 @@ CEXPORT int indigoLoadFasta(int source, const char* seq_type, int library)
         IndigoObject& lib_obj = self.getObject(library);
         SequenceLoader loader(IndigoScanner::get(obj), IndigoMonomerLibrary::get(lib_obj));
 
-        std::unique_ptr<IndigoMolecule> molptr = std::make_unique<IndigoMolecule>();
+        std::unique_ptr<IndigoKetDocument> docptr = std::make_unique<IndigoKetDocument>();
 
-        Molecule& mol = molptr->mol;
-        loader.loadFasta(mol, seq_type);
-        return self.addObject(molptr.release());
+        loader.loadFasta(docptr->get(), seq_type);
+        return self.addObject(docptr.release());
     }
     INDIGO_END(-1);
 }
@@ -671,11 +669,8 @@ CEXPORT int indigoLoadIdt(int source, int library)
         MonomerTemplateLibrary& lib = IndigoMonomerLibrary::get(lib_obj);
         SequenceLoader loader(IndigoScanner::get(obj), lib);
 
-        // std::unique_ptr<IndigoMolecule> molptr = std::make_unique<IndigoMolecule>();
         std::unique_ptr<IndigoKetDocument> docptr = std::make_unique<IndigoKetDocument>();
 
-        // Molecule& mol = molptr->mol;
-        // loader.loadIdt(mol);
         loader.loadIdt(docptr->get());
         return self.addObject(docptr.release());
     }
@@ -3067,7 +3062,6 @@ CEXPORT int indigoAddDataSGroup(int molecule, int natoms, int* atoms, int nbonds
         BaseMolecule& mol = self.getObject(molecule).getBaseMolecule();
         int idx = mol.sgroups.addSGroup(SGroup::SG_TYPE_DAT);
         DataSGroup& dsg = (DataSGroup&)mol.sgroups.getSGroup(idx);
-        int i;
         if (atoms != nullptr)
             dsg.atoms.concat(atoms, natoms);
 
@@ -4780,7 +4774,7 @@ CEXPORT int indigoNameToStructure(const char* name, const char* params)
             Duplicate params string as we call destructive function strtok() on callee side
             We can get rid of it if we have sustainable options in the future
             */
-            char* options = ::strdup(params);
+            char* options = ::_strdup(params);
             if (options)
             {
                 parser.setOptions(options);

diff --git a/api/c/indigo/src/indigo_savers.cpp b/api/c/indigo/src/indigo_savers.cpp
@@ -591,7 +591,7 @@ CEXPORT int indigoSaveSequence(int item, int output, int library)
     {
         IndigoObject& obj = self.getObject(item);
         Output& out = IndigoOutput::get(self.getObject(output));
-        if (IndigoBaseMolecule::is(obj) || IndigoKetDocument::is(obj))
+        if (IndigoBaseMolecule::is(obj))
         {
             IndigoObject& lib_obj = self.getObject(library);
             SequenceSaver saver(out, IndigoMonomerLibrary::get(lib_obj));
@@ -600,6 +600,14 @@ CEXPORT int indigoSaveSequence(int item, int output, int library)
             out.flush();
             return 1;
         }
+        else if (IndigoKetDocument::is(obj))
+        {
+            IndigoObject& lib_obj = self.getObject(library);
+            SequenceSaver saver(out, IndigoMonomerLibrary::get(lib_obj));
+            saver.saveKetDocument(static_cast<IndigoKetDocument&>(obj).get());
+            out.flush();
+            return 1;
+        }
         throw IndigoError("indigoSaveSequence(): expected molecule, got %s", obj.debugInfo());
     }
     INDIGO_END(-1);
@@ -620,6 +628,14 @@ CEXPORT int indigoSaveFasta(int item, int output, int library)
             out.flush();
             return 1;
         }
+        else if (IndigoKetDocument::is(obj))
+        {
+            IndigoObject& lib_obj = self.getObject(library);
+            SequenceSaver saver(out, IndigoMonomerLibrary::get(lib_obj));
+            saver.saveKetDocument(static_cast<IndigoKetDocument&>(obj).get(), SequenceSaver::SeqFormat::FASTA);
+            out.flush();
+            return 1;
+        }
         throw IndigoError("indigoSaveFasta(): expected molecule, got %s", obj.debugInfo());
     }
     INDIGO_END(-1);

diff --git a/api/tests/integration/ref/formats/fasta_to_ket.py.out b/api/tests/integration/ref/formats/fasta_to_ket.py.out
@@ -7,3 +7,6 @@ break_peptide.ket:SUCCEED
 break_rna.fasta:SEQUENCE loader: Invalid symbols in the sequence: *
 comment.ket:SUCCEED
 test_1881.ket:SUCCEED
+aminoacids_variants.ket:SUCCEED
+rna_variants.ket:SUCCEED
+dna_variants.ket:SUCCEED
diff --git a/api/tests/integration/ref/formats/idt_unresolved.py.out b/api/tests/integration/ref/formats/idt_unresolved.py.out
@@ -1,5 +1,5 @@
 *** IDT unresolved to misc. unsupported formats ***
-Sequence saver: i2AmPr cannot be written in sequence/FASTA format.
+Sequence saver: Can't save chem 'i2AmPr' to sequence format
 molecule CDXML saver: i2AmPr cannot be written in CDXML/CDX format.
 molfile saver: i2AmPr cannot be written in MDL Molfile format.
 SMILES saver: i2AmPr cannot be written in SMILES/SMARTS format.

diff --git a/api/tests/integration/ref/formats/seq_to_ket.py.out b/api/tests/integration/ref/formats/seq_to_ket.py.out
@@ -3,3 +3,6 @@ all_aminoacids.ket:SUCCEED
 rna_acgtu.ket:SUCCEED
 dna_acgtu.ket:SUCCEED
 spaces.ket:SUCCEED
+aminoacids_variants.ket:SUCCEED
+rna_variants.ket:SUCCEED
+dna_variants.ket:SUCCEED
diff --git a/api/tests/integration/tests/formats/fasta_to_fasta.py b/api/tests/integration/tests/formats/fasta_to_fasta.py
@@ -30,8 +30,9 @@ def find_diff(a, b):
     {"file": "multiseq", "seq_type": "DNA"},
 ]
 
-# empty library - internal used for now
-lib = indigo.loadMonomerLibrary('{"root":{}}')
+lib = indigo.loadMonomerLibraryFromFile(
+    os.path.join(ref_path, "monomer_library.ket")
+)
 
 for desc in fasta_files:
     filename = desc["file"]

diff --git a/api/tests/integration/tests/formats/fasta_to_ket.py b/api/tests/integration/tests/formats/fasta_to_ket.py
@@ -32,6 +32,9 @@ def find_diff(a, b):
     {"file": "break_rna", "seq_type": "RNA"},
     {"file": "comment", "seq_type": "PEPTIDE"},
     {"file": "test_1881", "seq_type": "PEPTIDE"},
+    {"file": "aminoacids_variants", "seq_type": "PEPTIDE"},
+    {"file": "rna_variants", "seq_type": "RNA"},
+    {"file": "dna_variants", "seq_type": "DNA"},
 ]
 
 
@@ -41,8 +44,9 @@ def remove_prefix(s, prefix="com.epam.indigo.IndigoException: "):
     return s
 
 
-# empty library - internal used for now
-lib = indigo.loadMonomerLibrary('{"root":{}}')
+lib = indigo.loadMonomerLibraryFromFile(
+    os.path.join(ref_path, "monomer_library.ket")
+)
 
 for desc in fasta_files:
     filename = desc["file"]

diff --git a/api/tests/integration/tests/formats/genbank_to_seq.py b/api/tests/integration/tests/formats/genbank_to_seq.py
@@ -27,8 +27,9 @@ def find_diff(a, b):
     {"file": "1844-gen_pept", "seq_type": "PEPTIDE"},
 ]
 
-# empty library - internal used for now
-lib = indigo.loadMonomerLibrary('{"root":{}}')
+lib = indigo.loadMonomerLibraryFromFile(
+    os.path.join(ref_path, "monomer_library.ket")
+)
 
 for infile in files:
     filename = infile["file"] + ".seq"

diff --git a/api/tests/integration/tests/formats/ket_to_fasta.py b/api/tests/integration/tests/formats/ket_to_fasta.py
@@ -28,8 +28,9 @@ def find_diff(a, b):
     "1950-mixed-seq",
 ]
 
-# empty library - internal used for now
-lib = indigo.loadMonomerLibrary('{"root":{}}')
+lib = indigo.loadMonomerLibraryFromFile(
+    os.path.join(ref_path, "monomer_library.ket")
+)
 
 files.sort()
 for filename in files:

diff --git a/api/tests/integration/tests/formats/molecules/aminoacids_variants.fasta b/api/tests/integration/tests/formats/molecules/aminoacids_variants.fasta
@@ -0,0 +1,2 @@
+>Sequence1
+BJZX
diff --git a/api/tests/integration/tests/formats/molecules/dna_variants.fasta b/api/tests/integration/tests/formats/molecules/dna_variants.fasta
@@ -0,0 +1,2 @@
+>Sequence1
+BN
diff --git a/api/tests/integration/tests/formats/molecules/rna_variants.fasta b/api/tests/integration/tests/formats/molecules/rna_variants.fasta
@@ -0,0 +1,2 @@
+>Sequence1
+KN