#1588 Import of standard IDT monomers (#1904)

Co-authored-by: even1024 <roman.porozhnetov@gmail.com> Co-authored-by: Aliakasndr Dziarkach <Aliakasndr.Dziarkach@gmail.com>
epam · Apr 9, 2024 · cc1b0d1 · cc1b0d1
1 parent 6761533
commit cc1b0d1
Show file tree

Hide file tree

Showing 28 changed files with 4,299 additions and 137 deletions.
diff --git a/api/c/indigo/indigo.h b/api/c/indigo/indigo.h
@@ -203,6 +203,10 @@ CEXPORT int indigoLoadFasta(int source, const char* seq_type);
 CEXPORT int indigoLoadFastaFromString(const char* string, const char* seq_type);
 CEXPORT int indigoLoadFastaFromFile(const char* filename, const char* seq_type);
 
+CEXPORT int indigoLoadIDT(int source);
+CEXPORT int indigoLoadIDTFromString(const char* string);
+CEXPORT int indigoLoadIDTFromFile(const char* filename);
+
 CEXPORT int indigoSaveMolfile(int molecule, int output);
 CEXPORT int indigoSaveMolfileToFile(int molecule, const char* filename);
 CEXPORT const char* indigoMolfile(int molecule);

diff --git a/api/c/indigo/src/indigo_molecule.cpp b/api/c/indigo/src/indigo_molecule.cpp
@@ -626,6 +626,56 @@ CEXPORT int indigoLoadFastaFromFile(const char* filename, const char* seq_type)
     INDIGO_END(-1);
 }
 
+CEXPORT int indigoLoadIDT(int source)
+{
+    INDIGO_BEGIN
+    {
+        IndigoObject& obj = self.getObject(source);
+        SequenceLoader loader(IndigoScanner::get(obj));
+
+        std::unique_ptr<IndigoMolecule> molptr = std::make_unique<IndigoMolecule>();
+
+        Molecule& mol = molptr->mol;
+        loader.loadIDT(mol);
+        return self.addObject(molptr.release());
+    }
+    INDIGO_END(-1);
+}
+
+CEXPORT int indigoLoadIDTFromString(const char* string)
+{
+    INDIGO_BEGIN
+    {
+        int source = indigoReadString(string);
+        int result;
+
+        if (source <= 0)
+            return -1;
+
+        result = indigoLoadIDT(source);
+        indigoFree(source);
+        return result;
+    }
+    INDIGO_END(-1);
+}
+
+CEXPORT int indigoLoadIDTFromFile(const char* filename)
+{
+    INDIGO_BEGIN
+    {
+        int source = indigoReadFile(filename);
+        int result;
+
+        if (source < 0)
+            return -1;
+
+        result = indigoLoadIDT(source);
+        indigoFree(source);
+        return result;
+    }
+    INDIGO_END(-1);
+}
+
 CEXPORT int indigoLoadSmarts(int source)
 {
     INDIGO_BEGIN

diff --git a/api/dotnet/src/Indigo.cs b/api/dotnet/src/Indigo.cs
@@ -454,6 +454,12 @@ public IndigoObject loadFasta(string str, string seq_type)
             return new IndigoObject(this, checkResult(IndigoLib.indigoLoadFastaFromString(str, seq_type)));
         }
 
+        public IndigoObject loadIDT(string str)
+        {
+            setSessionID();
+            return new IndigoObject(this, checkResult(IndigoLib.indigoLoadIDTFromString(str)));
+        }
+
         public IndigoObject loadSmarts(byte[] buf)
         {
             setSessionID();
@@ -478,6 +484,12 @@ public IndigoObject loadFastaFromFile(string path, string seq_type)
             return new IndigoObject(this, checkResult(IndigoLib.indigoLoadFastaFromFile(path, seq_type)));
         }
 
+        public IndigoObject loadIDTFromFile(string path)
+        {
+            setSessionID();
+            return new IndigoObject(this, checkResult(IndigoLib.indigoLoadIDTFromFile(path)));
+        }
+
         public IndigoObject loadReaction(string str)
         {
             setSessionID();

diff --git a/api/dotnet/src/IndigoLib.cs b/api/dotnet/src/IndigoLib.cs
@@ -134,12 +134,18 @@ public unsafe class IndigoLib
         [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
         public static extern int indigoLoadFastaFromString(string str, string seq_type);
 
+        [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
+        public static extern int indigoLoadIDTFromString(string str);
+
         [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
         public static extern int indigoLoadSequenceFromFile(string filename, string seq_type);
 
         [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
         public static extern int indigoLoadFastaFromFile(string filename, string seq_type);
 
+        [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
+        public static extern int indigoLoadIDTFromFile(string filename);
+
         [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
         public static extern int indigoLoadSmartsFromBuffer(byte[] buffer, int size);
 

diff --git a/api/java/indigo/src/main/java/com/epam/indigo/Indigo.java b/api/java/indigo/src/main/java/com/epam/indigo/Indigo.java
@@ -380,6 +380,16 @@ public IndigoObject loadFastaFromFile(String path, String seq_type) {
         return new IndigoObject(this, checkResult(this, lib.indigoLoadFastaFromFile(path, seq_type)));
     }
 
+    public IndigoObject loadIDT(String str) {
+        setSessionID();
+        return new IndigoObject(this, checkResult(this, lib.indigoLoadIDTFromString(str)));
+    }
+
+    public IndigoObject loadIDTFromFile(String path) {
+        setSessionID();
+        return new IndigoObject(this, checkResult(this, lib.indigoLoadIDTFromFile(path)));
+    }
+
     public IndigoObject loadReaction(String str) {
         setSessionID();
         return new IndigoObject(this, checkResult(this, lib.indigoLoadReactionFromString(str)));

diff --git a/api/java/indigo/src/main/java/com/epam/indigo/IndigoLib.java b/api/java/indigo/src/main/java/com/epam/indigo/IndigoLib.java
@@ -129,6 +129,12 @@ public interface IndigoLib extends Library {
 
     int indigoLoadFastaFromFile(String filename, String seq_type);
 
+    int indigoLoadIDT(int source);
+
+    int indigoLoadIDTFromString(String str);
+
+    int indigoLoadIDTFromFile(String filename);
+
     int indigoLoadStructureFromString(String str, String params);
 
     int indigoLoadStructureFromFile(String filename, String params);

diff --git a/api/python/indigo/indigo/indigo.py b/api/python/indigo/indigo/indigo.py
@@ -600,6 +600,46 @@ def loadFastaFromFile(self, filename, seq_type):
             ),
         )
 
+    def loadIDT(self, string):
+        """Loads molecule from IDT string
+
+        Args:
+            string (str): sequence string
+
+        Returns:
+            IndigoObject: loaded query molecular structure
+
+        Raises:
+            IndigoException: Exception if structure format is incorrect
+        """
+
+        return IndigoObject(
+            self,
+            IndigoLib.checkResult(
+                self._lib().indigoLoadIDTFromString(string.encode())
+            ),
+        )
+
+    def loadIDTFromFile(self, filename):
+        """Loads query molecule from file in IDT sequence format
+
+        Args:
+            filename (str): full path to the file with sequence string
+
+        Returns:
+            IndigoObject: loaded query molecular structure
+
+        Raises:
+            IndigoException: Exception if structure format is incorrect
+        """
+
+        return IndigoObject(
+            self,
+            IndigoLib.checkResult(
+                self._lib().indigoLoadIDTFromFile(filename.encode())
+            ),
+        )
+
     def loadReaction(self, string):
         """Loads reaction from string. Format will be automatically recognized.
 

diff --git a/api/python/indigo/indigo/indigo_lib.py b/api/python/indigo/indigo/indigo_lib.py
@@ -152,6 +152,10 @@ def __init__(self) -> None:
             c_char_p,
             c_char_p,
         ]
+        IndigoLib.lib.indigoLoadIDTFromString.restype = c_int
+        IndigoLib.lib.indigoLoadIDTFromString.argtypes = [c_char_p]
+        IndigoLib.lib.indigoLoadIDTFromFile.restype = c_int
+        IndigoLib.lib.indigoLoadIDTFromFile.argtypes = [c_char_p]
         IndigoLib.lib.indigoLoadReactionFromString.restype = c_int
         IndigoLib.lib.indigoLoadReactionFromString.argtypes = [c_char_p]
         IndigoLib.lib.indigoLoadReactionFromFile.restype = c_int

diff --git a/api/tests/integration/ref/formats/idt_to_ket.py.out b/api/tests/integration/ref/formats/idt_to_ket.py.out
@@ -0,0 +1,3 @@
+*** IDT to KET ***
+idt_acg.ket:SUCCEED
+idt_maxmgc.ket:SUCCEED
diff --git a/api/tests/integration/tests/formats/idt_to_ket.py b/api/tests/integration/tests/formats/idt_to_ket.py
@@ -0,0 +1,43 @@
+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("json-saving-pretty", True)
+indigo.setOption("ignore-stereochemistry-errors", True)
+
+print("*** IDT to KET ***")
+
+root = joinPathPy("molecules/", __file__)
+ref_path = joinPathPy("ref/", __file__)
+
+fasta_files = [
+    "idt_acg",
+    "idt_maxmgc",
+]
+
+for filename in fasta_files:
+    mol = indigo.loadIDTFromFile(os.path.join(root, filename + ".idt"))
+    # with open(os.path.join(ref_path, filename) + ".ket", "w") as file:
+    #     file.write(mol.json())
+    with open(os.path.join(ref_path, filename) + ".ket", "r") as file:
+        ket_ref = file.read()
+    ket = mol.json()
+    diff = find_diff(ket_ref, ket)
+    if not diff:
+        print(filename + ".ket:SUCCEED")
+    else:
+        print(filename + ".ket:FAILED")
+        print(diff)
diff --git a/api/tests/integration/tests/formats/molecules/idt_acg.idt b/api/tests/integration/tests/formats/molecules/idt_acg.idt
@@ -0,0 +1 @@
+ACG
diff --git a/api/tests/integration/tests/formats/molecules/idt_maxmgc.idt b/api/tests/integration/tests/formats/molecules/idt_maxmgc.idt
@@ -0,0 +1 @@
+mA*mGC