trunk/tools/Msdsd2Pdb.java

package tools;

import java.io.*;
import java.sql.SQLException;
import java.sql.Statement;
import java.sql.ResultSet;

/**
 * Package:             tools
 * Class:               Msdsd2Pdb
 * Author:              Ioannis Filippis, filippis@molgen.mpg.de
 * Date:                21/03/2006
 *
 * Msdsd2Pdb's static export2File method creates a pdb file by exporting the atom
 * lines directly from the msdsd. This is needed for the visualization of the
 * biological units since pdb files contain the ASUs. Moreover, contact graphs
 * are defined using msdsd-custom fields (like chain code and residue serial)
 * and mapping to pdb fields would be necessary for the graph visualisation 
 * if the original pdb files were preferred.
 * 
 * Notes:
 * - Hetatoms are excluded (pdb_group = "A") and in case of multiple locations of
 *       amino acids, only the default location is considered (graph_alt_code_used = 1) 
 *       (VendruscoloM_00_PSFG.pdf currently found in LitNet/incomingPDF/LAST_ROUND/)
 * - There is also the possibility to send the atom lines directly to PyMol and 
 *       loading the structure without intermediate files. Look at PyMol class and
 *   sendAtomLines method.
 * 
 * Changelog:
 * 21/03/06 first created by IF
 * 02/03/07 JD, Major changes: adapted to msdsd_00_07_a and using my_msdsd_00_07_a. 
 *                      Using MySQLConnection and file PrintStream instead of shell mysql client for output
 */

public class Msdsd2Pdb {

        
        public static String MSDSDDB="msdsd_00_07_a";
        public static String INFODB="my_msdsd_00_07_a";
        public static String HOST="white";
        public static String PWD="nieve";
        
        
    /**
     * Exports to file in pdb format the atom lines of a model (modelId) of a biological unit (assemblyId) 
     * of a protein (accessionCode) directly from msdsd.
     * 
     * Notes:
     * - Hetatoms are excluded (pdb_group = "A") and in case of multiple locations of
     *   amino acids, only the default location is considered (graph_alt_code_used = 1) 
     *   (VendruscoloM_00_PSFG.pdf currently found in LitNet/incomingPDF/LAST_ROUND/)
     * - The chain_pdb_code is used in the chainID field in the atom line, while the chain_code is used 
     *   the segID field (due to its length). Therefore, "segi" and not "chain" must be used in pymol
     *   selections.
     * 
     * @param accessionCode
     * @param assemblyId
     * @param modelId
     * @param pdbFile
     * @param user
     */ 
        public static void export2File(String accessionCode, int assemblyId, int modelId, String pdbFile, String user) throws FileNotFoundException{
                PrintStream Pdb = new PrintStream(new FileOutputStream(pdbFile));
                MySQLConnection conn = new MySQLConnection(HOST,user,PWD,MSDSDDB);

                String query = "SELECT CONCAT("+
                    "RPAD(\"ATOM\", 6, \" \"), "+
                    "LPAD(serial, 5, \" \"), "+
                    "\" \", "+
                    "LPAD(chem_atom_name, 4, \" \"), "+
                    "IF(alt_code IS NULL, \" \", alt_code), "+
                    "code_3_letter, "+
                    "\" \", "+
                    "IF(chain_pdb_code IS NULL, \" \", chain_pdb_code), "+
                    "LPAD(residue_serial, 4, \" \"), "+
                    "IF(residue_pdb_insert_code IS NULL, \" \", residue_pdb_insert_code), "+
                    "REPEAT(\" \", 3), "+
                    "LPAD(x, 8, \" \"), "+
                    "LPAD(y, 8, \" \"), "+
                    "LPAD(z, 8, \" \"), "+
                    "LPAD(occupancy, 6, \" \"), "+
                    "REPEAT(\" \", 6), "+
                    "REPEAT(\" \", 6), "+
                    "RPAD(chain_code, 4, \" \") "+
                    ") AS atom_lines FROM "+MSDSDDB+".atom_data WHERE "+
                    "(assembly_id = "+assemblyId+") AND "+
                    "(model_id = "+modelId+") AND "+
                    "(graph_alt_code_used = 1) AND "+
                    "(pdb_group = \"A\") "+
                    "ORDER BY chain_code, residue_serial, serial;";
                
                try {
                        Statement stmt = conn.createStatement();
                        ResultSet rsst = stmt.executeQuery(query);
                        while (rsst.next()) {
                                Pdb.println(rsst.getString(1));
                        }
                        stmt.close();
                        rsst.close();
                } catch (SQLException e) {
                        e.printStackTrace();
                }
                Pdb.close();
                conn.close();
    }

 
    /**
     * Exports to file in pdb format the atom lines of the assymetric unit of a protein (accessionCode) 
     * (all chains) directly from msdsd. 
     * 
     * Notes:
     * - Hetatoms are excluded (pdb_group = "A") and in case of multiple locations of
     *   amino acids, only the default location is considered (graph_alt_code_used = 1) 
     *   (VendruscoloM_00_PSFG.pdf currently found in LitNet/incomingPDF/LAST_ROUND/)
     * - The chain_pdb_code is used in the chainID field in the atom line, while the chain_code is used 
     *   the segID field (due to its length). Therefore, "segi" and not "chain" must be used in pymol
     *   selections.
     * 
     * @param accessionCode
     * @param pdbFile
     * @param user
     */ 
        public static void export2File(String accessionCode, String pdbFile, String user) throws FileNotFoundException {
                PrintStream Pdb = new PrintStream(new FileOutputStream(pdbFile));
                MySQLConnection conn = new MySQLConnection(HOST,user,PWD,MSDSDDB);

                String query = "SELECT CONCAT("+
                    "RPAD(\"ATOM\", 6, \" \"), "+
                    "LPAD(serial, 5, \" \"), "+
                    "\" \", "+
                    "LPAD(chem_atom_name, 4, \" \"), "+
                    "IF(alt_code IS NULL, \" \", alt_code), "+
                    "code_3_letter, "+
                    "\" \", "+
                    "IF(chain_pdb_code IS NULL, \" \", chain_pdb_code), "+
                    "LPAD(residue_serial, 4, \" \"), "+
                    "IF(residue_pdb_insert_code IS NULL, \" \", residue_pdb_insert_code), "+
                    "REPEAT(\" \", 3), "+
                    "LPAD(x, 8, \" \"), "+
                    "LPAD(y, 8, \" \"), "+
                    "LPAD(z, 8, \" \"), "+
                    "LPAD(occupancy, 6, \" \"), "+
                    "REPEAT(\" \", 6), "+
                    "REPEAT(\" \", 6), "+
                    "RPAD(chain_code, 4, \" \") "+
                    ") AS atom_lines FROM "+MSDSDDB+".atom_data WHERE "+
                    "(accession_code = \""+accessionCode+"\") AND "+
                    "(non_assembly_valid = \"Y\") AND "+
                    "(graph_alt_code_used = 1) AND "+
                    "(pdb_group = \"A\") "+
                    "ORDER BY chain_code, residue_serial, serial;";
                try {
                        Statement stmt = conn.createStatement();
                        ResultSet rsst = stmt.executeQuery(query);
                        while (rsst.next()) {
                                Pdb.println(rsst.getString(1));
                        }
                        stmt.close();
                        rsst.close();
                } catch (SQLException e) {
                        e.printStackTrace();
                }
                Pdb.close();
                conn.close();
    }
        
        /**
         * Exports to file in pdb format the atom coordinates for the assymetric unit of a protein given model_id and chain_id
         * @param chainId
         * @param modelId
         * @param pdbFile
         * @param user
         * @return
         */
        public static void export2File(int chainId, int modelId, String pdbFile, String user) throws FileNotFoundException {
                PrintStream Pdb = new PrintStream(new FileOutputStream(pdbFile));
                MySQLConnection conn = new MySQLConnection(HOST,user,PWD,MSDSDDB);

                String query = "SELECT CONCAT("+
                    "RPAD(\"ATOM\", 6, \" \"), "+
                    "LPAD(serial, 5, \" \"), "+
                    "\" \", "+
                    "LPAD(chem_atom_name, 4, \" \"), "+
                    "IF(alt_code IS NULL, \" \", alt_code), "+
                    "code_3_letter, "+
                    "\" \", "+
                    "IF(chain_pdb_code IS NULL, \" \", chain_pdb_code), "+
                    "LPAD(residue_serial, 4, \" \"), "+// check if this is msd or pdb residue serials, do we care?
                    "IF(residue_pdb_insert_code IS NULL, \" \", residue_pdb_insert_code), "+
                    "REPEAT(\" \", 3), "+
                    "LPAD(x, 8, \" \"), "+
                    "LPAD(y, 8, \" \"), "+
                    "LPAD(z, 8, \" \"), "+
                    "LPAD(occupancy, 6, \" \"), "+
                    "REPEAT(\" \", 6), "+
                    "REPEAT(\" \", 6), "+
                    "RPAD(chain_code, 4, \" \") "+
                    ") AS atom_lines FROM "+MSDSDDB+".atom_data WHERE "+
                    "(model_id = "+modelId+") AND "+
                    "(chain_id = "+chainId+") AND "+
                    "(graph_alt_code_used = 1) AND "+
                    "(pdb_group = \"A\") "+
                    "ORDER BY chain_code, residue_serial, serial;";
                try {
                        Statement stmt = conn.createStatement();
                        ResultSet rsst = stmt.executeQuery(query);
                        while (rsst.next()) {
                                Pdb.println(rsst.getString(1));
                        }
                        stmt.close();
                        rsst.close();
                } catch (SQLException e) {
                        e.printStackTrace();
                }
                Pdb.close();
                conn.close();           
    }

        /**
         * Exports to file in pdb format the atom coordinates for the assymetric unit of a protein given an accession_code and chain_pdb_code 
         * (if NMR just the model with model_serial=1)
         * @param accessionCode
         * @param chainPdbCode
         * @param pdbFile
         * @param user
         * @return
         */
        public static void export2File(String accessionCode, String chainPdbCode, String pdbFile, String user) throws FileNotFoundException{
                MySQLConnection conn = new MySQLConnection(HOST,user,PWD,MSDSDDB);
                int chainId=0;
                int modelId=0;
                String chainStr="='"+chainPdbCode+"'";
                if (chainPdbCode.equals("NULL")) {
                        chainStr="IS NULL";
                }
                String query = "SELECT chain_id, model_id " +
                                "FROM "+INFODB+".mmol_chain_info " +
                                "WHERE accession_code='"+accessionCode+"' " +
                                "AND chain_pdb_code " + chainStr +
                                "AND chain_type='C' " +
                                "AND asu_chain=1 " +
                                "AND model_serial=1;";
                try {
                        Statement stmt = conn.createStatement();
                        ResultSet rsst = stmt.executeQuery(query);
                        while (rsst.next()) {
                                chainId=rsst.getInt(1);
                                modelId=rsst.getInt(2);
                        }
                        stmt.close();
                        rsst.close();
                } catch (SQLException e) {
                        e.printStackTrace();
                }
                conn.close();
                export2File(chainId,modelId,pdbFile,user);
    }


} // end of class Msdsd2Pdb
Revision:	115
Committed:	Fri Mar 2 13:53:04 2007 UTC (17 years, 7 months ago) by duarte
File size:	9196 byte(s)
Log Message:	Adapted to msdsd_00_07_a Now uses MySQLConnection and PrintWriters and not mysql client through shell (got rid of that method) Thus all methos taking new arguments user and pdbFile New methods: export2file taking accession_code, chain_pdb_code export2file taking chain_id, model_id
Line	File contents
1	package tools;
2
3	import java.io.*;
4	import java.sql.SQLException;
5	import java.sql.Statement;
6	import java.sql.ResultSet;
7
8	/**
9	* Package: tools
10	* Class: Msdsd2Pdb
11	* Author: Ioannis Filippis, filippis@molgen.mpg.de
12	* Date: 21/03/2006
13	*
14	* Msdsd2Pdb's static export2File method creates a pdb file by exporting the atom
15	* lines directly from the msdsd. This is needed for the visualization of the
16	* biological units since pdb files contain the ASUs. Moreover, contact graphs
17	* are defined using msdsd-custom fields (like chain code and residue serial)
18	* and mapping to pdb fields would be necessary for the graph visualisation
19	* if the original pdb files were preferred.
20	*
21	* Notes:
22	* - Hetatoms are excluded (pdb_group = "A") and in case of multiple locations of
23	* amino acids, only the default location is considered (graph_alt_code_used = 1)
24	* (VendruscoloM_00_PSFG.pdf currently found in LitNet/incomingPDF/LAST_ROUND/)
25	* - There is also the possibility to send the atom lines directly to PyMol and
26	* loading the structure without intermediate files. Look at PyMol class and
27	* sendAtomLines method.
28	*
29	* Changelog:
30	* 21/03/06 first created by IF
31	* 02/03/07 JD, Major changes: adapted to msdsd_00_07_a and using my_msdsd_00_07_a.
32	* Using MySQLConnection and file PrintStream instead of shell mysql client for output
33	*/
34
35	public class Msdsd2Pdb {
36
37
38	public static String MSDSDDB="msdsd_00_07_a";
39	public static String INFODB="my_msdsd_00_07_a";
40	public static String HOST="white";
41	public static String PWD="nieve";
42
43
44	/**
45	* Exports to file in pdb format the atom lines of a model (modelId) of a biological unit (assemblyId)
46	* of a protein (accessionCode) directly from msdsd.
47	*
48	* Notes:
49	* - Hetatoms are excluded (pdb_group = "A") and in case of multiple locations of
50	* amino acids, only the default location is considered (graph_alt_code_used = 1)
51	* (VendruscoloM_00_PSFG.pdf currently found in LitNet/incomingPDF/LAST_ROUND/)
52	* - The chain_pdb_code is used in the chainID field in the atom line, while the chain_code is used
53	* the segID field (due to its length). Therefore, "segi" and not "chain" must be used in pymol
54	* selections.
55	*
56	* @param accessionCode
57	* @param assemblyId
58	* @param modelId
59	* @param pdbFile
60	* @param user
61	*/
62	public static void export2File(String accessionCode, int assemblyId, int modelId, String pdbFile, String user) throws FileNotFoundException{
63	PrintStream Pdb = new PrintStream(new FileOutputStream(pdbFile));
64	MySQLConnection conn = new MySQLConnection(HOST,user,PWD,MSDSDDB);
65
66	String query = "SELECT CONCAT("+
67	"RPAD(\"ATOM\", 6, \" \"), "+
68	"LPAD(serial, 5, \" \"), "+
69	"\" \", "+
70	"LPAD(chem_atom_name, 4, \" \"), "+
71	"IF(alt_code IS NULL, \" \", alt_code), "+
72	"code_3_letter, "+
73	"\" \", "+
74	"IF(chain_pdb_code IS NULL, \" \", chain_pdb_code), "+
75	"LPAD(residue_serial, 4, \" \"), "+
76	"IF(residue_pdb_insert_code IS NULL, \" \", residue_pdb_insert_code), "+
77	"REPEAT(\" \", 3), "+
78	"LPAD(x, 8, \" \"), "+
79	"LPAD(y, 8, \" \"), "+
80	"LPAD(z, 8, \" \"), "+
81	"LPAD(occupancy, 6, \" \"), "+
82	"REPEAT(\" \", 6), "+
83	"REPEAT(\" \", 6), "+
84	"RPAD(chain_code, 4, \" \") "+
85	") AS atom_lines FROM "+MSDSDDB+".atom_data WHERE "+
86	"(assembly_id = "+assemblyId+") AND "+
87	"(model_id = "+modelId+") AND "+
88	"(graph_alt_code_used = 1) AND "+
89	"(pdb_group = \"A\") "+
90	"ORDER BY chain_code, residue_serial, serial;";
91
92	try {
93	Statement stmt = conn.createStatement();
94	ResultSet rsst = stmt.executeQuery(query);
95	while (rsst.next()) {
96	Pdb.println(rsst.getString(1));
97	}
98	stmt.close();
99	rsst.close();
100	} catch (SQLException e) {
101	e.printStackTrace();
102	}
103	Pdb.close();
104	conn.close();
105	}
106
107
108	/**
109	* Exports to file in pdb format the atom lines of the assymetric unit of a protein (accessionCode)
110	* (all chains) directly from msdsd.
111	*
112	* Notes:
113	* - Hetatoms are excluded (pdb_group = "A") and in case of multiple locations of
114	* amino acids, only the default location is considered (graph_alt_code_used = 1)
115	* (VendruscoloM_00_PSFG.pdf currently found in LitNet/incomingPDF/LAST_ROUND/)
116	* - The chain_pdb_code is used in the chainID field in the atom line, while the chain_code is used
117	* the segID field (due to its length). Therefore, "segi" and not "chain" must be used in pymol
118	* selections.
119	*
120	* @param accessionCode
121	* @param pdbFile
122	* @param user
123	*/
124	public static void export2File(String accessionCode, String pdbFile, String user) throws FileNotFoundException {
125	PrintStream Pdb = new PrintStream(new FileOutputStream(pdbFile));
126	MySQLConnection conn = new MySQLConnection(HOST,user,PWD,MSDSDDB);
127
128	String query = "SELECT CONCAT("+
129	"RPAD(\"ATOM\", 6, \" \"), "+
130	"LPAD(serial, 5, \" \"), "+
131	"\" \", "+
132	"LPAD(chem_atom_name, 4, \" \"), "+
133	"IF(alt_code IS NULL, \" \", alt_code), "+
134	"code_3_letter, "+
135	"\" \", "+
136	"IF(chain_pdb_code IS NULL, \" \", chain_pdb_code), "+
137	"LPAD(residue_serial, 4, \" \"), "+
138	"IF(residue_pdb_insert_code IS NULL, \" \", residue_pdb_insert_code), "+
139	"REPEAT(\" \", 3), "+
140	"LPAD(x, 8, \" \"), "+
141	"LPAD(y, 8, \" \"), "+
142	"LPAD(z, 8, \" \"), "+
143	"LPAD(occupancy, 6, \" \"), "+
144	"REPEAT(\" \", 6), "+
145	"REPEAT(\" \", 6), "+
146	"RPAD(chain_code, 4, \" \") "+
147	") AS atom_lines FROM "+MSDSDDB+".atom_data WHERE "+
148	"(accession_code = \""+accessionCode+"\") AND "+
149	"(non_assembly_valid = \"Y\") AND "+
150	"(graph_alt_code_used = 1) AND "+
151	"(pdb_group = \"A\") "+
152	"ORDER BY chain_code, residue_serial, serial;";
153	try {
154	Statement stmt = conn.createStatement();
155	ResultSet rsst = stmt.executeQuery(query);
156	while (rsst.next()) {
157	Pdb.println(rsst.getString(1));
158	}
159	stmt.close();
160	rsst.close();
161	} catch (SQLException e) {
162	e.printStackTrace();
163	}
164	Pdb.close();
165	conn.close();
166	}
167
168	/**
169	* Exports to file in pdb format the atom coordinates for the assymetric unit of a protein given model_id and chain_id
170	* @param chainId
171	* @param modelId
172	* @param pdbFile
173	* @param user
174	* @return
175	*/
176	public static void export2File(int chainId, int modelId, String pdbFile, String user) throws FileNotFoundException {
177	PrintStream Pdb = new PrintStream(new FileOutputStream(pdbFile));
178	MySQLConnection conn = new MySQLConnection(HOST,user,PWD,MSDSDDB);
179
180	String query = "SELECT CONCAT("+
181	"RPAD(\"ATOM\", 6, \" \"), "+
182	"LPAD(serial, 5, \" \"), "+
183	"\" \", "+
184	"LPAD(chem_atom_name, 4, \" \"), "+
185	"IF(alt_code IS NULL, \" \", alt_code), "+
186	"code_3_letter, "+
187	"\" \", "+
188	"IF(chain_pdb_code IS NULL, \" \", chain_pdb_code), "+
189	"LPAD(residue_serial, 4, \" \"), "+// check if this is msd or pdb residue serials, do we care?
190	"IF(residue_pdb_insert_code IS NULL, \" \", residue_pdb_insert_code), "+
191	"REPEAT(\" \", 3), "+
192	"LPAD(x, 8, \" \"), "+
193	"LPAD(y, 8, \" \"), "+
194	"LPAD(z, 8, \" \"), "+
195	"LPAD(occupancy, 6, \" \"), "+
196	"REPEAT(\" \", 6), "+
197	"REPEAT(\" \", 6), "+
198	"RPAD(chain_code, 4, \" \") "+
199	") AS atom_lines FROM "+MSDSDDB+".atom_data WHERE "+
200	"(model_id = "+modelId+") AND "+
201	"(chain_id = "+chainId+") AND "+
202	"(graph_alt_code_used = 1) AND "+
203	"(pdb_group = \"A\") "+
204	"ORDER BY chain_code, residue_serial, serial;";
205	try {
206	Statement stmt = conn.createStatement();
207	ResultSet rsst = stmt.executeQuery(query);
208	while (rsst.next()) {
209	Pdb.println(rsst.getString(1));
210	}
211	stmt.close();
212	rsst.close();
213	} catch (SQLException e) {
214	e.printStackTrace();
215	}
216	Pdb.close();
217	conn.close();
218	}
219
220	/**
221	* Exports to file in pdb format the atom coordinates for the assymetric unit of a protein given an accession_code and chain_pdb_code
222	* (if NMR just the model with model_serial=1)
223	* @param accessionCode
224	* @param chainPdbCode
225	* @param pdbFile
226	* @param user
227	* @return
228	*/
229	public static void export2File(String accessionCode, String chainPdbCode, String pdbFile, String user) throws FileNotFoundException{
230	MySQLConnection conn = new MySQLConnection(HOST,user,PWD,MSDSDDB);
231	int chainId=0;
232	int modelId=0;
233	String chainStr="='"+chainPdbCode+"'";
234	if (chainPdbCode.equals("NULL")) {
235	chainStr="IS NULL";
236	}
237	String query = "SELECT chain_id, model_id " +
238	"FROM "+INFODB+".mmol_chain_info " +
239	"WHERE accession_code='"+accessionCode+"' " +
240	"AND chain_pdb_code " + chainStr +
241	"AND chain_type='C' " +
242	"AND asu_chain=1 " +
243	"AND model_serial=1;";
244	try {
245	Statement stmt = conn.createStatement();
246	ResultSet rsst = stmt.executeQuery(query);
247	while (rsst.next()) {
248	chainId=rsst.getInt(1);
249	modelId=rsst.getInt(2);
250	}
251	stmt.close();
252	rsst.close();
253	} catch (SQLException e) {
254	e.printStackTrace();
255	}
256	conn.close();
257	export2File(chainId,modelId,pdbFile,user);
258	}
259
260
261
262
263	} // end of class Msdsd2Pdb