Merge pull request ModelSEED#92 from ModelSEED/updating_structures

Updating out-of-date InChI structures

Author: samseaver

Biochemistry/Structures/MetaCyc/InChI_ChargedStrings.txt

@@ -73,7 +73,7 @@ def loadStructures(self,sources_array=[],db_array=[]):
 
         structures_dict = dict()
         if(len(db_array)==1 and db_array[0]=="ModelSEED"):
-            struct_file = "ModelSEED/ModelSEED_Strings.txt"
+            struct_file = "ModelSEED_Structures.txt"
             struct_file = self.StructRoot+struct_file
             reader = DictReader(open(struct_file), dialect = "excel-tab", fieldnames = ['ID','Source','Aliases','Structure'])
             for line in reader:

Biochemistry/Structures/MetaCyc/InChI_OriginalStrings.txt

@@ -0,0 +1,87 @@
+import java.io.IOException;
+
+import chemaxon.formats.MolImporter;
+import chemaxon.formats.MolExporter;
+import chemaxon.formats.MolFormatException;
+import chemaxon.struc.Molecule;
+import chemaxon.marvin.calculations.MajorMicrospeciesPlugin;
+import chemaxon.marvin.calculations.TautomerizationPlugin;
+import chemaxon.marvin.plugin.PluginException;
+
+public final class ChargeInChI {
+
+    public static void main(String[] args){
+	Molecule mol = new Molecule();
+	double pH = 7.0;
+
+	//Import Mol File
+	try {
+	
+	    MolImporter mi = new MolImporter(args[0]);
+
+	    try {
+		mol = mi.read();
+	    } finally {
+		mi.close();
+	    }
+
+	} catch (MolFormatException e) {
+	    throw new IllegalArgumentException("Invalid molecule format", e);
+	} catch (IOException e) {
+	    throw new IllegalArgumentException("Error reading input file", e);
+	}
+
+	//Check for pH
+	if(args.length == 2){
+	    pH = Double.parseDouble(args[1]);
+	}
+
+	//Ready Charge Plugin
+	MajorMicrospeciesPlugin mms_plugin = new MajorMicrospeciesPlugin();
+	mms_plugin.setpH(pH);
+
+	//Ready Tautomerizer Plugin
+	TautomerizationPlugin t_plugin = new TautomerizationPlugin();
+	t_plugin.setDominantTautomerDistributionCalculation(true);
+	t_plugin.setpH(pH);
+
+	//Molecule for fusing all fragments
+	Molecule FusedMol = new Molecule();
+
+	//Fragment Molecule and iterate
+	Molecule frags[] = mol.convertToFrags();
+	for(int i = 0; i < frags.length; i++){
+	    
+	    try{
+
+		//Run Charge Plugin
+		mms_plugin.setMolecule(frags[i]);
+		mms_plugin.standardize(frags[i]);
+		mms_plugin.run();
+		Molecule mms_mol = mms_plugin.getMajorMicrospecies();
+		
+		//Run Tautomer Plugin
+		t_plugin.setMolecule(mms_mol);
+		t_plugin.standardize(mms_mol);
+		t_plugin.run();
+		Molecule t_mol = t_plugin.getStructure(0);
+		
+		//Fuse fragment
+		FusedMol.fuse(t_mol,false);
+		
+	    }catch(PluginException PIE){
+		System.out.println("Error: "+PIE.getMessage());
+	    }
+	}
+
+	//Print to string
+	try{
+	
+	    String FusedString = MolExporter.exportToFormat(FusedMol, "inchi");
+	    System.out.print(FusedString);
+
+	}catch(IOException IOE) {
+	    System.out.println("Error: "+IOE.getMessage());
+	}
+    }
+}

Biochemistry/Structures/MetaCyc/InChI_SearchStrings.txt

@@ -1,3 +1,5 @@
 #!/bin/bash
+source /vol/kbase/deployment/user-env.sh
 javac -cp "/homes/seaver/Software/MarvinBeans/lib/*:." ChargeMol.java
+javac -cp "/homes/seaver/Software/MarvinBeans/lib/*:." ChargeInChI.java
 

Biochemistry/Structures/MetaCyc/InChI_StereoFreeSearchStrings.txt

@@ -19,8 +19,8 @@
 
 #Source_Aliases_Dict =  CompoundsHelper.loadSourceAliases()
 
-ms_structs_file = open("../../Biochemistry/Structures/ModelSEED/ModelSEED_Strings.txt",'w')
-ms_structs_file.write("ID\tSource\tAliases\tStructure\n")
+ms_structs_file = open("../../Biochemistry/Structures/ModelSEED_Structures.txt",'w')
+ms_structs_file.write("ID\tType\tAliases\tStructure\n")
 for msid in sorted(MS_Aliases_Dict.keys()):
 
     #Build collection of all structures for the ModelSEED ID