owl/trunk/genDbGraph.java

import gnu.getopt.Getopt;

import java.io.BufferedReader;
import java.io.FileReader;
import java.io.IOException;
import java.sql.SQLException;


import proteinstructure.Graph;
import proteinstructure.Pdb;
import proteinstructure.PdbChainCodeNotFoundError;
import proteinstructure.PdbCodeNotFoundError;
import proteinstructure.PdbaseInconsistencyError;
import proteinstructure.PdbasePdb;
import proteinstructure.PdbfileFormatError;
import proteinstructure.PdbfilePdb;
import tools.MySQLConnection;


public class genDbGraph {
        /*------------------------------ constants ------------------------------*/
        
        public static final String                      PDB_DB = "pdbase";
        public static final String                      DB_HOST = "white";                                                              
        public static final String                      DB_USER = getUserName();
        public static final String                      DB_PWD = "nieve";
        public static final String                      DSSP_EXE = "/project/StruPPi/bin/dssp";
        public static final String                      DSSP_PARAMS = "--";

        //public static double                  cutoff = 4.2;
        //public static String                  edgeType = "ALL";
        
        /*---------------------------- private methods --------------------------*/
        /** 
         * Get user name from operating system (for use as database username). 
         * */
        private static String getUserName() {
                String user = null;
                user = System.getProperty("user.name");
                if(user == null) {
                        System.err.println("Could not get user name from operating system.");
                }
                return user;
        }
        
        public static void main(String[] args) throws IOException {
                
                
                String help = "Usage, 3 options:\n" +
                                "1)  genDbGraph -i <listfile> -d <distance_cutoff> -t <contact_type> -o <output_db> [-D <pdbase_db>] \n" +
                                "2)  genDbGraph -p <pdb_code> -c <chain_pdb_code> -d <distance_cutoff> -t <contact_type> -o <output_db> [-D <pdbase_db>] \n" +
                                "3)  genDbGraph -f <pdbfile> -c <chain_pdb_code> -d <distance_cutoff> -t <contact_type> -o <output_db> \n" +
                                "In case 2) also a list of comma separated pdb codes and chain codes can be specified, e.g. -p 1bxy,1jos -c A,A\n" +
                                "\nIf pdbase_db not specified, the default pdbase will be used\n" +
                                "\nSecondary structure will be taken from pdbase database. If reading from pdb file and the pdb file is missing the secondary structure, then it will be assigned using dssp\n"; 

                String listfile = "";
                String[] pdbCodes = null;
                String[] pdbChainCodes = null;
                String pdbfile = "";
                String pdbaseDb = PDB_DB;
                String edgeType = "";
                double cutoff = 0.0;
                String outputDb = "";
                
                Getopt g = new Getopt("genDbGraph", args, "i:p:c:f:d:t:o:D:h?");
                int c;
                while ((c = g.getopt()) != -1) {
                        switch(c){
                        case 'i':
                                listfile = g.getOptarg();
                                break;
                        case 'p':
                                pdbCodes = g.getOptarg().split(",");
                                break;
                        case 'c':
                                pdbChainCodes = g.getOptarg().split(",");
                                break;
                        case 'f':
                                pdbfile = g.getOptarg();
                                break;
                        case 'd':
                                cutoff = Double.valueOf(g.getOptarg());
                                break;
                        case 't':
                                edgeType = g.getOptarg();
                                break;
                        case 'o':
                                outputDb = g.getOptarg();
                                break;
                        case 'D':
                                pdbaseDb = g.getOptarg();
                                break;
                        case 'h':
                        case '?':
                                System.out.println(help);
                                System.exit(0);
                                break; // getopt() already printed an error
                        }
                }

                if (outputDb.equals("") || edgeType.equals("") || cutoff==0.0) {
                        System.err.println("Some missing option");
                        System.err.println(help);
                        System.exit(1);
                }
                if (listfile.equals("") && pdbCodes==null && pdbfile.equals("")){
                        System.err.println("Either a listfile, some pdb codes/chain codes or a pdbfile must be given");
                        System.err.println(help);
                        System.exit(1);
                }
                if ((!listfile.equals("") && pdbCodes!=null) || (!listfile.equals("") && !pdbfile.equals("")) || (pdbCodes!=null && !pdbfile.equals(""))) {
                        System.err.println("Options -p/-c, -i and -f/-c are exclusive. Use only one of them");
                        System.err.println(help);
                        System.exit(1);                 
                }

                
                MySQLConnection conn = null;            

                try{
                        conn = new MySQLConnection(DB_HOST, DB_USER, DB_PWD);
                } catch (Exception e) {
                        System.err.println("Error opening database connection. Exiting");
                        System.exit(1);
                }
                
                
                if (pdbfile.equals("")){
                        
                        if (!listfile.equals("")) {                     
                                BufferedReader fpdb = new BufferedReader(new FileReader(listfile));
                                String line = "";
                                int numLines = 0;
                                fpdb.mark(100000);
                                while ((line = fpdb.readLine() ) != null ) {
                                        if (line.length()>0) numLines++;
                                }
                                fpdb.reset();
                                pdbCodes = new String[numLines];
                                pdbChainCodes = new String[numLines];
                                numLines = 0;
                                while ((line = fpdb.readLine() ) != null ) {
                                        pdbCodes[numLines] = line.split("\\s+")[0].toLowerCase();
                                        pdbChainCodes[numLines] = line.split("\\s+")[1];
                                        numLines++;
                                }
                                fpdb.close();
                        }

                        int numPdbs = 0;

                        for (int i=0;i<pdbCodes.length;i++) {
                                String pdbCode = pdbCodes[i];
                                String pdbChainCode = pdbChainCodes[i];

                                try {
                                        
                                        long start = System.currentTimeMillis();
                                        
                                        Pdb pdb = new PdbasePdb(pdbCode, pdbChainCode, pdbaseDb, conn);

                                        // get graph
                                        Graph graph = pdb.get_graph(edgeType, cutoff);

                                        graph.write_graph_to_db(conn,outputDb);

                                        long end = System.currentTimeMillis();
                                        double time = (double) (end -start)/1000;
                                        
                                        System.out.printf(pdbCode+"_"+pdbChainCode+" - %5.3f s\n",time);
                                        numPdbs++;
                                        
                                } catch (PdbaseInconsistencyError e) {
                                        System.err.println("Inconsistency in " + pdbCode + pdbChainCode);
                                } catch (PdbCodeNotFoundError e) {
                                        System.err.println("Couldn't find pdb code "+pdbCode);
                                } catch (SQLException e) {
                                        System.err.println("SQL error for structure "+pdbCode+"_"+pdbChainCode+", error: "+e.getMessage());
                                } catch (PdbChainCodeNotFoundError e) {
                                        System.err.println("Couldn't find pdb chain code "+pdbChainCode+" for pdb code "+pdbCode);
                                }

                        }

                        // output results
                        System.out.println("Number of structures loaded successfully: " + numPdbs);


                } else {
                        String pdbChainCode = pdbChainCodes[0];
                        try {
                                Pdb pdb = new PdbfilePdb(pdbfile,pdbChainCode);
                                if (!pdb.hasSecondaryStructure()) {
                                        pdb.runDssp(DSSP_EXE, DSSP_PARAMS);
                                }
                                Graph graph = pdb.get_graph(edgeType, cutoff);
                                try {
                                        graph.write_graph_to_db(conn, outputDb);
                                        System.out.println("Loaded to database graph for file "+pdbfile);
                                } catch (SQLException e) {
                                        e.printStackTrace();
                                }
                                
                        } catch (PdbfileFormatError e) {
                                System.err.println("pdb file "+pdbfile+" doesn't have right format");
                        } catch (PdbChainCodeNotFoundError e) {
                                System.err.println("chain code "+pdbChainCode+" wasn't found in file "+pdbfile);        
                        }
                }
                
                // closing db connection
                conn.close();
        }

}
Revision:	358
Committed:	Wed Oct 17 13:22:01 2007 UTC (17 years ago) by duarte
File size:	6642 byte(s)
Log Message:	Now closing db connection. Better help string.
Line	User	Rev	File contents
1	duarte	264	import gnu.getopt.Getopt;
2
3			import java.io.BufferedReader;
4			import java.io.FileReader;
5			import java.io.IOException;
6			import java.sql.SQLException;
7
8
9			import proteinstructure.Graph;
10			import proteinstructure.Pdb;
11			import proteinstructure.PdbChainCodeNotFoundError;
12			import proteinstructure.PdbCodeNotFoundError;
13			import proteinstructure.PdbaseInconsistencyError;
14			import proteinstructure.PdbasePdb;
15			import proteinstructure.PdbfileFormatError;
16			import proteinstructure.PdbfilePdb;
17			import tools.MySQLConnection;
18
19
20			public class genDbGraph {
21			/------------------------------ constants ------------------------------/
22
23			public static final String PDB_DB = "pdbase";
24			public static final String DB_HOST = "white";
25			public static final String DB_USER = getUserName();
26			public static final String DB_PWD = "nieve";
27			public static final String DSSP_EXE = "/project/StruPPi/bin/dssp";
28			public static final String DSSP_PARAMS = "--";
29
30			//public static double cutoff = 4.2;
31			//public static String edgeType = "ALL";
32
33			/---------------------------- private methods --------------------------/
34			/**
35			* Get user name from operating system (for use as database username).
36			* */
37			private static String getUserName() {
38			String user = null;
39			user = System.getProperty("user.name");
40			if(user == null) {
41			System.err.println("Could not get user name from operating system.");
42			}
43			return user;
44			}
45
46			public static void main(String[] args) throws IOException {
47
48
49			String help = "Usage, 3 options:\n" +
50			"1) genDbGraph -i <listfile> -d <distance_cutoff> -t <contact_type> -o <output_db> [-D <pdbase_db>] \n" +
51			"2) genDbGraph -p <pdb_code> -c <chain_pdb_code> -d <distance_cutoff> -t <contact_type> -o <output_db> [-D <pdbase_db>] \n" +
52			"3) genDbGraph -f <pdbfile> -c <chain_pdb_code> -d <distance_cutoff> -t <contact_type> -o <output_db> \n" +
53			"In case 2) also a list of comma separated pdb codes and chain codes can be specified, e.g. -p 1bxy,1jos -c A,A\n" +
54	duarte	358	"\nIf pdbase_db not specified, the default pdbase will be used\n" +
55			"\nSecondary structure will be taken from pdbase database. If reading from pdb file and the pdb file is missing the secondary structure, then it will be assigned using dssp\n";
56	duarte	264
57			String listfile = "";
58			String[] pdbCodes = null;
59			String[] pdbChainCodes = null;
60			String pdbfile = "";
61			String pdbaseDb = PDB_DB;
62			String edgeType = "";
63			double cutoff = 0.0;
64			String outputDb = "";
65
66			Getopt g = new Getopt("genDbGraph", args, "i:p:c:f:d:t:o:D:h?");
67			int c;
68			while ((c = g.getopt()) != -1) {
69			switch(c){
70			case 'i':
71			listfile = g.getOptarg();
72			break;
73			case 'p':
74			pdbCodes = g.getOptarg().split(",");
75			break;
76			case 'c':
77			pdbChainCodes = g.getOptarg().split(",");
78			break;
79			case 'f':
80			pdbfile = g.getOptarg();
81			break;
82			case 'd':
83			cutoff = Double.valueOf(g.getOptarg());
84			break;
85			case 't':
86			edgeType = g.getOptarg();
87			break;
88			case 'o':
89			outputDb = g.getOptarg();
90			break;
91			case 'D':
92			pdbaseDb = g.getOptarg();
93			break;
94			case 'h':
95			case '?':
96			System.out.println(help);
97			System.exit(0);
98			break; // getopt() already printed an error
99			}
100			}
101
102			if (outputDb.equals("") \|\| edgeType.equals("") \|\| cutoff==0.0) {
103			System.err.println("Some missing option");
104			System.err.println(help);
105			System.exit(1);
106			}
107			if (listfile.equals("") && pdbCodes==null && pdbfile.equals("")){
108			System.err.println("Either a listfile, some pdb codes/chain codes or a pdbfile must be given");
109			System.err.println(help);
110			System.exit(1);
111			}
112			if ((!listfile.equals("") && pdbCodes!=null) \|\| (!listfile.equals("") && !pdbfile.equals("")) \|\| (pdbCodes!=null && !pdbfile.equals(""))) {
113			System.err.println("Options -p/-c, -i and -f/-c are exclusive. Use only one of them");
114			System.err.println(help);
115			System.exit(1);
116			}
117
118
119			MySQLConnection conn = null;
120
121			try{
122			conn = new MySQLConnection(DB_HOST, DB_USER, DB_PWD);
123			} catch (Exception e) {
124			System.err.println("Error opening database connection. Exiting");
125			System.exit(1);
126			}
127
128
129			if (pdbfile.equals("")){
130
131			if (!listfile.equals("")) {
132			BufferedReader fpdb = new BufferedReader(new FileReader(listfile));
133			String line = "";
134			int numLines = 0;
135			fpdb.mark(100000);
136			while ((line = fpdb.readLine() ) != null ) {
137	duarte	285	if (line.length()>0) numLines++;
138	duarte	264	}
139			fpdb.reset();
140			pdbCodes = new String[numLines];
141			pdbChainCodes = new String[numLines];
142			numLines = 0;
143			while ((line = fpdb.readLine() ) != null ) {
144	duarte	285	pdbCodes[numLines] = line.split("\\s+")[0].toLowerCase();
145			pdbChainCodes[numLines] = line.split("\\s+")[1];
146	duarte	264	numLines++;
147			}
148	duarte	358	fpdb.close();
149	duarte	264	}
150
151			int numPdbs = 0;
152
153			for (int i=0;i<pdbCodes.length;i++) {
154			String pdbCode = pdbCodes[i];
155			String pdbChainCode = pdbChainCodes[i];
156
157			try {
158	duarte	285
159			long start = System.currentTimeMillis();
160
161	duarte	277	Pdb pdb = new PdbasePdb(pdbCode, pdbChainCode, pdbaseDb, conn);
162	duarte	264
163			// get graph
164			Graph graph = pdb.get_graph(edgeType, cutoff);
165
166			graph.write_graph_to_db(conn,outputDb);
167
168	duarte	285	long end = System.currentTimeMillis();
169			double time = (double) (end -start)/1000;
170
171			System.out.printf(pdbCode+"_"+pdbChainCode+" - %5.3f s\n",time);
172	duarte	277	numPdbs++;
173
174	duarte	264	} catch (PdbaseInconsistencyError e) {
175	duarte	277	System.err.println("Inconsistency in " + pdbCode + pdbChainCode);
176	duarte	264	} catch (PdbCodeNotFoundError e) {
177	duarte	277	System.err.println("Couldn't find pdb code "+pdbCode);
178	duarte	264	} catch (SQLException e) {
179	duarte	285	System.err.println("SQL error for structure "+pdbCode+"_"+pdbChainCode+", error: "+e.getMessage());
180	duarte	264	} catch (PdbChainCodeNotFoundError e) {
181	duarte	277	System.err.println("Couldn't find pdb chain code "+pdbChainCode+" for pdb code "+pdbCode);
182	duarte	264	}
183
184			}
185
186			// output results
187	duarte	277	System.out.println("Number of structures loaded successfully: " + numPdbs);
188	duarte	264
189
190			} else {
191			String pdbChainCode = pdbChainCodes[0];
192			try {
193			Pdb pdb = new PdbfilePdb(pdbfile,pdbChainCode);
194			if (!pdb.hasSecondaryStructure()) {
195			pdb.runDssp(DSSP_EXE, DSSP_PARAMS);
196			}
197			Graph graph = pdb.get_graph(edgeType, cutoff);
198			try {
199			graph.write_graph_to_db(conn, outputDb);
200			System.out.println("Loaded to database graph for file "+pdbfile);
201			} catch (SQLException e) {
202			e.printStackTrace();
203			}
204
205			} catch (PdbfileFormatError e) {
206			System.err.println("pdb file "+pdbfile+" doesn't have right format");
207			} catch (PdbChainCodeNotFoundError e) {
208			System.err.println("chain code "+pdbChainCode+" wasn't found in file "+pdbfile);
209			}
210			}
211	duarte	358
212			// closing db connection
213			conn.close();
214	duarte	264	}
215
216			}