#2472 - Support for three letter sequences (#2590)

api/c/indigo/indigo.h

@@ -229,6 +229,10 @@ CEXPORT int indigoSaveSequence(int molecule, int output, int library);
 CEXPORT int indigoSaveSequenceToFile(int molecule, const char* filename, int library);
 CEXPORT const char* indigoSequence(int molecule, int library);
 
+CEXPORT int indigoSaveSequence3Letter(int molecule, int output, int library);
+CEXPORT int indigoSaveSequence3LetterToFile(int molecule, const char* filename, int library);
+CEXPORT const char* indigoSequence3Letter(int molecule, int library);
+
 CEXPORT int indigoSaveFasta(int molecule, int output, int library);
 CEXPORT int indigoSaveFastaToFile(int molecule, const char* filename, int library);
 CEXPORT const char* indigoFasta(int molecule, int library);

api/c/indigo/src/indigo_macros.c

@@ -134,6 +134,20 @@ CEXPORT int indigoSaveSequenceToFile(int item, const char* filename, int library
     return res;
 }
 
+CEXPORT int indigoSaveSequence3LetterToFile(int item, const char* filename, int library)
+{
+    int f = indigoWriteFile(filename);
+    int res;
+
+    if (f == -1)
+        return -1;
+
+    res = indigoSaveSequence3Letter(item, f, library);
+
+    indigoFree(f);
+    return res;
+}
+
 CEXPORT int indigoSaveFastaToFile(int item, const char* filename, int library)
 {
     int f = indigoWriteFile(filename);
@@ -222,6 +236,22 @@ CEXPORT const char* indigoSequence(int molecule, int library)
     return res;
 }
 
+CEXPORT const char* indigoSequence3Letter(int molecule, int library)
+{
+    int b = indigoWriteBuffer();
+    const char* res;
+
+    if (b == -1)
+        return 0;
+
+    if (indigoSaveSequence3Letter(molecule, b, library) == -1)
+        return 0;
+
+    res = indigoToString(b);
+    indigoFree(b);
+    return res;
+}
+
 CEXPORT const char* indigoFasta(int molecule, int library)
 {
     int b = indigoWriteBuffer();

api/c/indigo/src/indigo_savers.cpp

@@ -616,6 +616,25 @@ CEXPORT int indigoSaveSequence(int item, int output, int library)
     INDIGO_END(-1);
 }
 
+CEXPORT int indigoSaveSequence3Letter(int item, int output, int library)
+{
+    INDIGO_BEGIN
+    {
+        IndigoObject& obj = self.getObject(item);
+        Output& out = IndigoOutput::get(self.getObject(output));
+        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::Sequence3);
+            out.flush();
+            return 1;
+        }
+        throw IndigoError("indigoSaveSequence(): expected molecule, got %s", obj.debugInfo());
+    }
+    INDIGO_END(-1);
+}
+
 CEXPORT int indigoSaveFasta(int item, int output, int library)
 {
     INDIGO_BEGIN

api/dotnet/src/IndigoLib.cs

@@ -182,6 +182,9 @@ public unsafe class IndigoLib
         [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
         public static extern int indigoSaveSequenceToFile(int molecule, int output, int library);
 
+        [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
+        public static extern int indigoSaveSequence3LetterToFile(int molecule, int output, int library);
+
         [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
         public static extern int indigoSaveMolfileToFile(int molecule, string filename);
 
@@ -191,6 +194,9 @@ public unsafe class IndigoLib
         [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
         public static extern byte* indigoSequence(int molecule, int library);
 
+        [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
+        public static extern byte* indigoSequence3Letter(int molecule, int library);
+
         [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
         public static extern byte* indigoFasta(int molecule, int library);
 

api/dotnet/src/IndigoObject.cs

@@ -92,6 +92,12 @@ public string sequence(IndigoObject library)
             return dispatcher.checkResult(IndigoLib.indigoSequence(self, library.self));
         }
 
+        public string sequence3Letter(IndigoObject library)
+        {
+            dispatcher.setSessionID();
+            return dispatcher.checkResult(IndigoLib.indigoSequence3Letter(self, library.self));
+        }
+
         public string fasta(IndigoObject library)
         {
             dispatcher.setSessionID();
@@ -118,6 +124,14 @@ public void saveSequenceToFile(string filename, IndigoObject library)
             dispatcher.checkResult(IndigoLib.indigoFree(s));
         }
 
+        public void saveSequence3LetterToFile(string filename, IndigoObject library)
+        {
+            dispatcher.setSessionID();
+            int s = dispatcher.checkResult(IndigoLib.indigoWriteFile(filename));
+            dispatcher.checkResult(IndigoLib.indigoSaveSequence3LetterToFile(self, s, library.self));
+            dispatcher.checkResult(IndigoLib.indigoFree(s));
+        }
+
         public string cml()
         {
             dispatcher.setSessionID();

api/java/indigo/src/main/java/com/epam/indigo/IndigoLib.java

@@ -163,6 +163,8 @@ public interface IndigoLib extends Library {
 
     int indigoSaveSequenceToFile(int molecule, String filename, int library);
 
+    int indigoSaveSequence3LetterToFile(int molecule, String filename, int library);
+
     int indigoSaveFastaToFile(int molecule, String filename);
 
     int indigoSaveMolfileToFile(int molecule, String filename);
@@ -171,6 +173,8 @@ public interface IndigoLib extends Library {
 
     Pointer indigoSequence(int molecule, int library);
 
+    Pointer indigoSequence3Letter(int molecule, int library);
+
     Pointer indigoFasta(int molecule, int library);
 
     Pointer indigoIdt(int molecule, int library);

api/java/indigo/src/main/java/com/epam/indigo/IndigoObject.java

@@ -82,6 +82,11 @@ public String sequence(IndigoObject library) {
         return Indigo.checkResultString(this, lib.indigoSequence(self, library.self));
     }
 
+    public String sequence3Letter(IndigoObject library) {
+        dispatcher.setSessionID();
+        return Indigo.checkResultString(this, lib.indigoSequence3Letter(self, library.self));
+    }
+
     public String fasta(IndigoObject library) {
         dispatcher.setSessionID();
         return Indigo.checkResultString(this, lib.indigoFasta(self, library.self));
@@ -112,6 +117,11 @@ public void saveSequenceToFile(String filename, IndigoObject library) {
         Indigo.checkResult(this, lib.indigoSaveSequenceToFile(self, filename, library.self));
     }
 
+    public void saveSequence3LetterToFile(String filename, IndigoObject library) {
+        dispatcher.setSessionID();
+        Indigo.checkResult(this, lib.indigoSaveSequence3LetterToFile(self, filename, library.self));
+    }
+
     public void saveFastaToFile(String filename) {
         dispatcher.setSessionID();
         Indigo.checkResult(this, lib.indigoSaveFastaToFile(self, filename));

api/python/indigo/indigo/indigo_lib.py

@@ -905,6 +905,14 @@ def __init__(self) -> None:
             c_char_p,
             c_int,
         ]
+        IndigoLib.lib.indigoSequence3Letter.restype = c_char_p
+        IndigoLib.lib.indigoSequence3Letter.argtypes = [c_int, c_int]
+        IndigoLib.lib.indigoSaveSequence3LetterToFile.restype = c_int
+        IndigoLib.lib.indigoSaveSequence3LetterToFile.argtypes = [
+            c_int,
+            c_char_p,
+            c_int,
+        ]
         IndigoLib.lib.indigoFasta.restype = c_char_p
         IndigoLib.lib.indigoFasta.argtypes = [c_int, c_int]
         IndigoLib.lib.indigoSaveFastaToFile.restype = c_int

api/python/indigo/indigo/indigo_object.py

@@ -3434,6 +3434,33 @@ def sequence(self, library):
             self._lib().indigoSequence(self.id, library.id)
         )
 
+    def saveSequence3Letter(self, filename, library):
+        """Saves macromolecule to monomers 3 letter sequence file
+
+        Args:
+            filename (str): full file path to the output file
+
+        Returns:
+            int: 1 if file is saved successfully
+        """
+
+        return IndigoLib.checkResult(
+            self._lib().indigoSaveSequence3LetterToFile(
+                self.id, filename.encode(), library.id
+            )
+        )
+
+    def sequence3Letter(self, library):
+        """Molecule or reaction method returns monomer 3 letter sequence for the structure
+
+        Returns:
+            str: sequence string
+        """
+
+        return IndigoLib.checkResultString(
+            self._lib().indigoSequence3Letter(self.id, library.id)
+        )
+
     def saveFasta(self, filename, library):
         """Saves macromolecule to FASTA file
 

api/tests/integration/ref/formats/ket_to_seq3letter.py.out

@@ -0,0 +1,2 @@
+*** KET to 3 LETTER SEQUENCE ***
+peptides_3letter : SUCCEED

api/tests/integration/ref/formats/seq3letter_to_ket.py.out

@@ -0,0 +1,5 @@
+*** 3 LETTER SEQUENCE to KET ***
+peptides_3letter : SUCCEED
+Test 'ALA': got expected error 'Given string cannot be interpreted as a valid three letter sequence because of incorrect formatting.'
+Test 'Al a': got expected error 'Given string cannot be interpreted as a valid three letter sequence because of incorrect formatting.'
+Test 'ala': got expected error 'Given string cannot be interpreted as a valid three letter sequence because of incorrect formatting.'

api/tests/integration/tests/formats/ket_to_seq3letter.py

@@ -0,0 +1,45 @@
+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 Indigo, joinPathPy  # noqa
+
+indigo = Indigo()
+indigo.setOption("ignore-stereochemistry-errors", True)
+
+print("*** KET to 3 LETTER SEQUENCE ***")
+
+refp = joinPathPy("ref/", __file__)
+
+files = [
+    "peptides_3letter",
+]
+
+lib = indigo.loadMonomerLibraryFromFile(
+    os.path.join(refp, "monomer_library.ket")
+)
+
+files.sort()
+for filename in files:
+    doc = indigo.loadKetDocumentFromFile(os.path.join(refp, filename + ".ket"))
+    with open(os.path.join(refp, filename) + ".seq3", "w") as file:
+        file.write(doc.sequence3Letter(lib))
+    with open(os.path.join(refp, filename) + ".seq3", "r") as file:
+        seq_ref = file.read()
+    seq = doc.sequence3Letter(lib)
+    diff = find_diff(seq_ref, seq)
+    if not diff:
+        print(filename + " : SUCCEED")
+    else:
+        print(filename + " : FAILED")
+        print(diff)