owl/trunk/testGraphAverager.java

import java.io.*;
import java.util.*;
import proteinstructure.*;

public class testGraphAverager {

        /**
         * @param args
         */
        public static void main(String[] args) {
                // Plan:
                // use getopt to parse the following parameters
                // - original structure (optional, for benchmarking)
                // - file with list of structures for averaging (required)
                // - contact type & distance cutoff (both required)
                // - consensus edge threshold (defaults to 0.5?)
                // - run tinker or not
                // - parameters for tinker model picking (optional)
                // - number of tinker models to output (only if run tinker is set, defaults to one)
                // - output benchmark result (graph comparison or structure comparison, only if original structure is specified)
                // - output file for resulting graph (optional)
                
                // take pdb files from command line (later: use getopt)
                // create trivial alignment (check alignment)
                // create graphAverager
                // create consensus graph
                // compare with graph for native structure (first argument)
                
                // do tinker reconstruction
                // compare with native structure
                
                // benchmarking:
                // - use score of best model vs. native (either graph based or structure based)
                // - use matlab to optimize input parameters
                // - create matlab executable for optimizing command line parameters:
                //   * specify command line to be run and constant parameters
                //   * specify parameter ranges to be optimized (discrete numerical, continuous numerical with steps, discrete set)
                //   * specify regexp to parse result from output (stdout or file)
                //   * specify optimization procedure (use default matlab ones, e.g. each parameter separate, genetic algorithm, simul. annealing, ...)
                //   * specify convergence threshold or timeout
                //   * run on cluster?
                
                
                // read command line parameters
                if(args.length < 4) {
                        System.out.println("Usage: testGraphAverager <targetPdbFile> <ChainCode> <ListOfPredictionFiles> <edgeThreshold>");
                        System.exit(1);
                }
                
                File targetFile = new File(args[0]);
                File listFile = new File(args[2]);
                String chainCode = args[1];
                String thresholdStr = args[3];
                Pdb target = null;
                Graph targetGraph = null;
                Vector<String> modelFileNames = new Vector<String>();
                Vector<Graph> models = new Vector<Graph>();
                String contactType = "Cb";
                double distCutoff = 7.0;
                double graphAveragingThreshold = Double.parseDouble(thresholdStr); 
                
                if(!targetFile.canRead()) {
                        System.err.println("Can not read from file " + targetFile.getAbsolutePath());
                        System.exit(1);
                }
                if(!listFile.canRead()) {
                        System.err.println("Can not read from file " + listFile.getAbsolutePath());
                        System.exit(1);
                }
                
                // read input files
                try {
                        target = new PdbfilePdb(targetFile.getAbsolutePath(), chainCode);
                        targetGraph = target.get_graph(contactType, distCutoff);
                } catch(PdbChainCodeNotFoundError e) {
                        System.err.println("Chain code " + chainCode + " not found in file " + targetFile.getAbsolutePath());
                        System.exit(1);
                } catch(PdbfileFormatError e) {
                        System.err.println("Formating error in file " + targetFile.getAbsolutePath());  
                        System.exit(1);
                } catch (FileNotFoundException e) {
                        System.err.println("File " + targetFile.getAbsolutePath() + " not found");      
                        System.exit(1);                 
                } catch(IOException e) {
                        System.err.println("Error reading from file " + targetFile.getAbsolutePath());  
                        System.exit(1);                         
                }
                
                try {
                        BufferedReader in = new BufferedReader(new FileReader(listFile));
                        String line;
                        File file;
                        Pdb pdb;
                        Graph graph;
                        while ((line =  in.readLine()  ) != null) {
                                file = new File(line);
                                if(!file.canRead()) {
                                        System.err.println("File " + line + " not found.");
                                        System.exit(1);
                                } else {
                                        modelFileNames.add(line);
                                        try {
                                                pdb = new PdbfilePdb(file.getAbsolutePath(),Pdb.NULL_CHAIN_CODE);
                                                graph = pdb.get_graph(contactType, distCutoff);
                                                models.add(graph);
                                        } catch(PdbChainCodeNotFoundError e) {
                                                System.err.println("Chain code " + chainCode + " not found in file " + file.getAbsolutePath());
                                                System.exit(1);
                                        } catch(PdbfileFormatError e) {
                                                System.err.println("Formating error in file " + file.getAbsolutePath());        
                                                System.exit(1);
                                        } catch (FileNotFoundException e) {
                                                System.err.println("File " + file.getAbsolutePath() + " not found");    
                                                System.exit(1);                 
                                        } catch(IOException e) {
                                                System.err.println("Error reading from file " + file.getAbsolutePath());        
                                                System.exit(1);                         
                                        }
                                }
                        }
                        in.close();

                } catch(FileNotFoundException e) {
                        System.err.println("File " + listFile.getAbsolutePath() + " not found.");
                        System.exit(1);
                } catch(IOException e) {
                        System.err.println("Error reading from file " + listFile.getAbsolutePath());
                        System.exit(1);
                }
                
                // create alignment
                TreeMap<String,String> sequences = new TreeMap<String, String>();
                TreeMap<String, Graph> templateGraphs = new TreeMap<String, Graph>();
                sequences.put(targetFile.getAbsolutePath(), targetGraph.getSequence());
                for(int i=0; i < models.size(); i++) {
                        sequences.put(modelFileNames.get(i), models.get(i).getSequence());
                        templateGraphs.put(modelFileNames.get(i), models.get(i));
                }
                Alignment al = new Alignment(sequences);
                
                // create GraphAverager
                GraphAverager grav = new GraphAverager(targetGraph.getSequence(), al, templateGraphs, targetFile.getAbsolutePath());
                Graph resultGraph = grav.doAveraging(graphAveragingThreshold);
                
                // compare prediction with target
                PredEval eval = resultGraph.evaluatePrediction(targetGraph);
                eval.print();
                
                // write result graph to file
                try {
                        resultGraph.write_graph_to_file("predictedGraph.cm");
                } catch (IOException e) {
                        // TODO Auto-generated catch block
                        e.printStackTrace();
                }
                
        }
}
Revision:	351
Committed:	Thu Oct 11 12:43:54 2007 UTC (17 years ago) by stehr
File size:	5786 byte(s)
Log Message:	added some test programs to default package
Line	User	Rev	File contents
1	stehr	351	import java.io.*;
2			import java.util.*;
3			import proteinstructure.*;
4
5			public class testGraphAverager {
6
7			/**
8			* @param args
9			*/
10			public static void main(String[] args) {
11			// Plan:
12			// use getopt to parse the following parameters
13			// - original structure (optional, for benchmarking)
14			// - file with list of structures for averaging (required)
15			// - contact type & distance cutoff (both required)
16			// - consensus edge threshold (defaults to 0.5?)
17			// - run tinker or not
18			// - parameters for tinker model picking (optional)
19			// - number of tinker models to output (only if run tinker is set, defaults to one)
20			// - output benchmark result (graph comparison or structure comparison, only if original structure is specified)
21			// - output file for resulting graph (optional)
22
23			// take pdb files from command line (later: use getopt)
24			// create trivial alignment (check alignment)
25			// create graphAverager
26			// create consensus graph
27			// compare with graph for native structure (first argument)
28
29			// do tinker reconstruction
30			// compare with native structure
31
32			// benchmarking:
33			// - use score of best model vs. native (either graph based or structure based)
34			// - use matlab to optimize input parameters
35			// - create matlab executable for optimizing command line parameters:
36			// * specify command line to be run and constant parameters
37			// * specify parameter ranges to be optimized (discrete numerical, continuous numerical with steps, discrete set)
38			// * specify regexp to parse result from output (stdout or file)
39			// * specify optimization procedure (use default matlab ones, e.g. each parameter separate, genetic algorithm, simul. annealing, ...)
40			// * specify convergence threshold or timeout
41			// * run on cluster?
42
43
44			// read command line parameters
45			if(args.length < 4) {
46			System.out.println("Usage: testGraphAverager <targetPdbFile> <ChainCode> <ListOfPredictionFiles> <edgeThreshold>");
47			System.exit(1);
48			}
49
50			File targetFile = new File(args[0]);
51			File listFile = new File(args[2]);
52			String chainCode = args[1];
53			String thresholdStr = args[3];
54			Pdb target = null;
55			Graph targetGraph = null;
56			Vector<String> modelFileNames = new Vector<String>();
57			Vector<Graph> models = new Vector<Graph>();
58			String contactType = "Cb";
59			double distCutoff = 7.0;
60			double graphAveragingThreshold = Double.parseDouble(thresholdStr);
61
62			if(!targetFile.canRead()) {
63			System.err.println("Can not read from file " + targetFile.getAbsolutePath());
64			System.exit(1);
65			}
66			if(!listFile.canRead()) {
67			System.err.println("Can not read from file " + listFile.getAbsolutePath());
68			System.exit(1);
69			}
70
71			// read input files
72			try {
73			target = new PdbfilePdb(targetFile.getAbsolutePath(), chainCode);
74			targetGraph = target.get_graph(contactType, distCutoff);
75			} catch(PdbChainCodeNotFoundError e) {
76			System.err.println("Chain code " + chainCode + " not found in file " + targetFile.getAbsolutePath());
77			System.exit(1);
78			} catch(PdbfileFormatError e) {
79			System.err.println("Formating error in file " + targetFile.getAbsolutePath());
80			System.exit(1);
81			} catch (FileNotFoundException e) {
82			System.err.println("File " + targetFile.getAbsolutePath() + " not found");
83			System.exit(1);
84			} catch(IOException e) {
85			System.err.println("Error reading from file " + targetFile.getAbsolutePath());
86			System.exit(1);
87			}
88
89			try {
90			BufferedReader in = new BufferedReader(new FileReader(listFile));
91			String line;
92			File file;
93			Pdb pdb;
94			Graph graph;
95			while ((line = in.readLine() ) != null) {
96			file = new File(line);
97			if(!file.canRead()) {
98			System.err.println("File " + line + " not found.");
99			System.exit(1);
100			} else {
101			modelFileNames.add(line);
102			try {
103			pdb = new PdbfilePdb(file.getAbsolutePath(),Pdb.NULL_CHAIN_CODE);
104			graph = pdb.get_graph(contactType, distCutoff);
105			models.add(graph);
106			} catch(PdbChainCodeNotFoundError e) {
107			System.err.println("Chain code " + chainCode + " not found in file " + file.getAbsolutePath());
108			System.exit(1);
109			} catch(PdbfileFormatError e) {
110			System.err.println("Formating error in file " + file.getAbsolutePath());
111			System.exit(1);
112			} catch (FileNotFoundException e) {
113			System.err.println("File " + file.getAbsolutePath() + " not found");
114			System.exit(1);
115			} catch(IOException e) {
116			System.err.println("Error reading from file " + file.getAbsolutePath());
117			System.exit(1);
118			}
119			}
120			}
121			in.close();
122
123			} catch(FileNotFoundException e) {
124			System.err.println("File " + listFile.getAbsolutePath() + " not found.");
125			System.exit(1);
126			} catch(IOException e) {
127			System.err.println("Error reading from file " + listFile.getAbsolutePath());
128			System.exit(1);
129			}
130
131			// create alignment
132			TreeMap<String,String> sequences = new TreeMap<String, String>();
133			TreeMap<String, Graph> templateGraphs = new TreeMap<String, Graph>();
134			sequences.put(targetFile.getAbsolutePath(), targetGraph.getSequence());
135			for(int i=0; i < models.size(); i++) {
136			sequences.put(modelFileNames.get(i), models.get(i).getSequence());
137			templateGraphs.put(modelFileNames.get(i), models.get(i));
138			}
139			Alignment al = new Alignment(sequences);
140
141			// create GraphAverager
142			GraphAverager grav = new GraphAverager(targetGraph.getSequence(), al, templateGraphs, targetFile.getAbsolutePath());
143			Graph resultGraph = grav.doAveraging(graphAveragingThreshold);
144
145			// compare prediction with target
146			PredEval eval = resultGraph.evaluatePrediction(targetGraph);
147			eval.print();
148
149			// write result graph to file
150			try {
151			resultGraph.write_graph_to_file("predictedGraph.cm");
152			} catch (IOException e) {
153			// TODO Auto-generated catch block
154			e.printStackTrace();
155			}
156
157			}
158			}