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" +
                                "If pdbase_db not specified, the default pdbase will be used\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++;
                                }
                        }

                        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);        
                        }
                }
        }

}
Revision:	285
Committed:	Thu Aug 16 15:10:11 2007 UTC (17 years, 2 months ago) by duarte
File size:	6394 byte(s)
Log Message:	Fixed a few bugs in reading of list file Better output and error messages
Line	File contents
1	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	"If pdbase_db not specified, the default pdbase will be used\n";
55
56	String listfile = "";
57	String[] pdbCodes = null;
58	String[] pdbChainCodes = null;
59	String pdbfile = "";
60	String pdbaseDb = PDB_DB;
61	String edgeType = "";
62	double cutoff = 0.0;
63	String outputDb = "";
64
65	Getopt g = new Getopt("genDbGraph", args, "i:p:c:f:d:t:o:D:h?");
66	int c;
67	while ((c = g.getopt()) != -1) {
68	switch(c){
69	case 'i':
70	listfile = g.getOptarg();
71	break;
72	case 'p':
73	pdbCodes = g.getOptarg().split(",");
74	break;
75	case 'c':
76	pdbChainCodes = g.getOptarg().split(",");
77	break;
78	case 'f':
79	pdbfile = g.getOptarg();
80	break;
81	case 'd':
82	cutoff = Double.valueOf(g.getOptarg());
83	break;
84	case 't':
85	edgeType = g.getOptarg();
86	break;
87	case 'o':
88	outputDb = g.getOptarg();
89	break;
90	case 'D':
91	pdbaseDb = g.getOptarg();
92	break;
93	case 'h':
94	case '?':
95	System.out.println(help);
96	System.exit(0);
97	break; // getopt() already printed an error
98	}
99	}
100
101	if (outputDb.equals("") \|\| edgeType.equals("") \|\| cutoff==0.0) {
102	System.err.println("Some missing option");
103	System.err.println(help);
104	System.exit(1);
105	}
106	if (listfile.equals("") && pdbCodes==null && pdbfile.equals("")){
107	System.err.println("Either a listfile, some pdb codes/chain codes or a pdbfile must be given");
108	System.err.println(help);
109	System.exit(1);
110	}
111	if ((!listfile.equals("") && pdbCodes!=null) \|\| (!listfile.equals("") && !pdbfile.equals("")) \|\| (pdbCodes!=null && !pdbfile.equals(""))) {
112	System.err.println("Options -p/-c, -i and -f/-c are exclusive. Use only one of them");
113	System.err.println(help);
114	System.exit(1);
115	}
116
117
118	MySQLConnection conn = null;
119
120	try{
121	conn = new MySQLConnection(DB_HOST, DB_USER, DB_PWD);
122	} catch (Exception e) {
123	System.err.println("Error opening database connection. Exiting");
124	System.exit(1);
125	}
126
127
128	if (pdbfile.equals("")){
129
130	if (!listfile.equals("")) {
131	BufferedReader fpdb = new BufferedReader(new FileReader(listfile));
132	String line = "";
133	int numLines = 0;
134	fpdb.mark(100000);
135	while ((line = fpdb.readLine() ) != null ) {
136	if (line.length()>0) numLines++;
137	}
138	fpdb.reset();
139	pdbCodes = new String[numLines];
140	pdbChainCodes = new String[numLines];
141	numLines = 0;
142	while ((line = fpdb.readLine() ) != null ) {
143	pdbCodes[numLines] = line.split("\\s+")[0].toLowerCase();
144	pdbChainCodes[numLines] = line.split("\\s+")[1];
145	numLines++;
146	}
147	}
148
149	int numPdbs = 0;
150
151	for (int i=0;i<pdbCodes.length;i++) {
152	String pdbCode = pdbCodes[i];
153	String pdbChainCode = pdbChainCodes[i];
154
155	try {
156
157	long start = System.currentTimeMillis();
158
159	Pdb pdb = new PdbasePdb(pdbCode, pdbChainCode, pdbaseDb, conn);
160
161	// get graph
162	Graph graph = pdb.get_graph(edgeType, cutoff);
163
164	graph.write_graph_to_db(conn,outputDb);
165
166	long end = System.currentTimeMillis();
167	double time = (double) (end -start)/1000;
168
169	System.out.printf(pdbCode+"_"+pdbChainCode+" - %5.3f s\n",time);
170	numPdbs++;
171
172	} catch (PdbaseInconsistencyError e) {
173	System.err.println("Inconsistency in " + pdbCode + pdbChainCode);
174	} catch (PdbCodeNotFoundError e) {
175	System.err.println("Couldn't find pdb code "+pdbCode);
176	} catch (SQLException e) {
177	System.err.println("SQL error for structure "+pdbCode+"_"+pdbChainCode+", error: "+e.getMessage());
178	} catch (PdbChainCodeNotFoundError e) {
179	System.err.println("Couldn't find pdb chain code "+pdbChainCode+" for pdb code "+pdbCode);
180	}
181
182	}
183
184	// output results
185	System.out.println("Number of structures loaded successfully: " + numPdbs);
186
187
188	} else {
189	String pdbChainCode = pdbChainCodes[0];
190	try {
191	Pdb pdb = new PdbfilePdb(pdbfile,pdbChainCode);
192	if (!pdb.hasSecondaryStructure()) {
193	pdb.runDssp(DSSP_EXE, DSSP_PARAMS);
194	}
195	Graph graph = pdb.get_graph(edgeType, cutoff);
196	try {
197	graph.write_graph_to_db(conn, outputDb);
198	System.out.println("Loaded to database graph for file "+pdbfile);
199	} catch (SQLException e) {
200	e.printStackTrace();
201	}
202
203	} catch (PdbfileFormatError e) {
204	System.err.println("pdb file "+pdbfile+" doesn't have right format");
205	} catch (PdbChainCodeNotFoundError e) {
206	System.err.println("chain code "+pdbChainCode+" wasn't found in file "+pdbfile);
207	}
208	}
209	}
210
211	}