owl/trunk/rewriteNBHood.java

import tools.MySQLConnection;

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

public class rewriteNBHood {

        /**
         * rewriting the nbhoodstring such that BB dominated nbs are represented by SSType
         * only single gapsymbol is used for nbhood and fullstring _With _Gaps (nwg and fwg)  
         * output to stdout such that it can be read by
         * mysql> load data infile '/scratch/local/cullpdb/cullpdb_20.nbhoods' into table nbstrings;
         * see bottom for table definition 
         * @author lappe
         */

        static String user = "lappe", fullnbs="", gapnbs="", nums="", gaphood=""; // change user name!!
        static MySQLConnection conn;
        static int k=0, k_sc=0, k_bb=0;
        static int minDmeansSC=10; // above this threshold contacts are assumed to be SC dominated 
        static String backgrndDB = "cullpdb_90"; 

        public static void main(String[] args) throws SQLException {
                String nbhood = "", msql, cid, res, sectype;
                int graph_id=0, node_id=0, resnr=0; 
                // System.out.println("rewriteNBHood v.0.2");
                conn = new MySQLConnection("white",user,"nieve", backgrndDB);
                Statement mstmt; 
                ResultSet mrsst;

                try {
                        msql = "select graph_id, node_id, cid, num, res, sstype from single_model_node where sstype is not null;";
                        // System.out.println("nodes> "+ msql);
                        mstmt = conn.createStatement();
                        mrsst = mstmt.executeQuery(msql);

                        while (mrsst.next()) {
                                // System.out.print(">");
                                graph_id=mrsst.getInt(1);
                                node_id =mrsst.getInt(2);
                                cid     =mrsst.getString(3);
                                resnr   =mrsst.getInt(4);
                                res     =mrsst.getString(5);
                                sectype =mrsst.getString(6);
                                System.out.print(graph_id); // graph
                                System.out.print("\t"+node_id); // node
                                System.out.print("\t"+cid); // cid
                                System.out.print("\t"+resnr); // num
                                System.out.print("\t"+res); // res
                                System.out.print("\t"+sectype); // sstype

                                nbhood= getSSNbHoodString( graph_id, cid, resnr);
                                System.out.print("\t"+k);
                                System.out.print("\t"+k_bb);
                                System.out.print("\t"+k_sc);  
                                System.out.print("\t"+nbhood);
                                System.out.print("\t"+gaphood);
                                System.out.print("\t"+fullnbs);
                                System.out.print("\t"+gapnbs);
                                // System.out.print("\t"+nums);
                                System.out.println(); 

                        }
                        mrsst.close();
                        mstmt.close();

                } catch (SQLException e) {
                        e.printStackTrace();
                        System.err.println("SQLException: " + e.getMessage());
                        System.err.println("SQLState:     " + e.getSQLState());
                }

                // System.out.println("fin.");
        }

        public static String getSSNbHoodString( int gid, String cid, int num ) {
                int j_num, j_bb, j_sc, lastnum=0;
                boolean overx=false; 
                String sql, nbs="", j_res, j_ss, nextnb, gapsym="", xgapsym=""; 
                Statement stmt;  
                ResultSet rsst;

                try {
                        k=0; k_sc=0; k_bb=0; // initialising the degreecounts with 0  
                        sql = "select j_num, j_res, j_sstype, weight_BB, weight_SC from single_model_edge where graph_id="+gid+" and i_cid='"+cid+"' and i_num="+num+" order by j_num;";
                        stmt = conn.createStatement();
                        rsst = stmt.executeQuery(sql);
                        fullnbs=""; 
                        gapnbs="";
                        gaphood="";
                        nums=""; 
                        while (rsst.next()) {   
                                gapsym=""; 
                                xgapsym=""; 
                                j_num = rsst.getInt(1); 
                                j_res = rsst.getString(2);                              
                                j_ss = rsst.getString(3); 
                                j_bb = rsst.getInt(4); 
                                j_sc = rsst.getInt(5);
                                // System.out.print("\n-> "+j_res+" "+j_num+" "+j_ss+" "+"("+j_sc+"/"+j_bb+")");
                                nextnb="?"; 
                                if (lastnum>0 && ((j_num-lastnum)>=2)) gapsym="%";   
                                if (!overx && j_num>num) { 
                                        overx=true; 
                                        nbs+="x";
                                        fullnbs+="x";
                                        nums+=(" ("+num+") "); 
                                        if (lastnum>0 && ((num-lastnum)>2)) xgapsym="%";   
                                        gapnbs+=xgapsym+"x"; 
                                        gaphood+=xgapsym+"x"; 
                                        gapsym=""; 
                                        if ((j_num-num)>=2) gapsym="%";  
                                }
                                // conversion of BB dominated contacts
                                if (j_sc==0 && j_bb==0) { // both values are 0 -> decide by sequence sparation
                                        // System.out.print("SC&BB=0"); 
                                        if ( Math.abs( num-j_num)>=minDmeansSC) {
                                                nextnb=j_res;
                                                k_sc++; 
                                                // System.out.print(" -> SC ");
                                        } else {
                                                nextnb=xlateSS( j_ss);
                                                k_bb++; 
                                                // System.out.print(" -> BB ");
                                        } // end if Math.abs( num-j_num)>=minDmeansSC
                                } else { // at least 1 value is non zero 
                                        // System.out.print("SC>0 or BB>0");
                                        if ( (j_bb>j_sc) ) { // bb dominated -> only symbol for sstype  
                                                k_bb++; 
                                                nextnb=xlateSS( j_ss);
                                                // System.out.print(" -> BB ");
                                        } else { // sc dominated contact
                                                k_sc++; 
                                                nextnb=j_res;
                                                // System.out.print(" -> SC ");
                                        } // end if bb dominated
                                } // end if we have non-zero entries
                                // System.out.print(" : "+nextnb);
                                nbs+= nextnb;
                                nums+=" "+j_num; 
                                fullnbs+= j_res; 
                                k++;

                                gapnbs+=gapsym+j_res;
                                gaphood+=gapsym+nextnb; 
                                lastnum=j_num; 
                                // System.out.print( nextnb);
                        }
                        rsst.close(); 
                        stmt.close();
                        if (!overx) {
                                nbs+="x"; // add if last in string == not seen yet
                                fullnbs+="x"; 
                                if (lastnum>0 && ((num-lastnum)==2)) xgapsym="_";
                                if (lastnum>0 && ((num-lastnum)>2)) xgapsym="%";
                                gapnbs+=xgapsym+"x"; 
                                gaphood+=xgapsym+"x"; 
                        }

                } catch (SQLException e) {
                        e.printStackTrace();
                        System.err.println("SQLException: " + e.getMessage());
                        System.err.println("SQLState:     " + e.getSQLState());
                }
                return nbs; 
        } // end of getnbhoodstring

        public static String xlateSS( String sstype) { // xlate SecSTructure-Symbol 
                String SSymbol = "o";
                sstype=sstype.toUpperCase(); 
                if (sstype.equals("H")) SSymbol="z"; // Helix 
                if (sstype.equals("S")) SSymbol="b"; // beta strand 
                return SSymbol; 
        } // end xlateSS

} // end 

/* table definition for the results

CREATE TABLE `nbstrings` (
`graph_id` int(10) unsigned DEFAULT NULL,
`node_id` int(10) unsigned DEFAULT NULL,
`cid` varchar(6) COLLATE latin1_general_cs DEFAULT NULL,
`num` int(10) unsigned DEFAULT NULL,
`res` char(1) COLLATE latin1_general_cs DEFAULT NULL,
`sstype` char(1) COLLATE latin1_general_cs DEFAULT NULL,
`k` int(10) unsigned DEFAULT NULL,
`k_BB` int(10) unsigned DEFAULT NULL,
`k_SC` int(10) unsigned DEFAULT NULL,
`n` varchar(30) COLLATE latin1_general_cs DEFAULT NULL,
`nwg` varchar(40) COLLATE latin1_general_cs DEFAULT NULL,
`f` varchar(30) COLLATE latin1_general_cs DEFAULT NULL,
`fwg` varchar(40) COLLATE latin1_general_cs DEFAULT NULL
) ENGINE=MyISAM DEFAULT CHARSET=latin1 COLLATE=latin1_general_cs; 

*/ 
Revision:	492
Committed:	Wed Jan 2 13:18:57 2008 UTC (16 years, 8 months ago) by duarte
File size:	6363 byte(s)
Log Message:	Copied the aglappe-jung branch into trunk.
Line	File contents
1	import tools.MySQLConnection;
2
3	import java.sql.SQLException;
4	import java.sql.Statement;
5	import java.sql.ResultSet;
6
7	public class rewriteNBHood {
8
9	/**
10	* rewriting the nbhoodstring such that BB dominated nbs are represented by SSType
11	* only single gapsymbol is used for nbhood and fullstring _With _Gaps (nwg and fwg)
12	* output to stdout such that it can be read by
13	* mysql> load data infile '/scratch/local/cullpdb/cullpdb_20.nbhoods' into table nbstrings;
14	* see bottom for table definition
15	* @author lappe
16	*/
17
18	static String user = "lappe", fullnbs="", gapnbs="", nums="", gaphood=""; // change user name!!
19	static MySQLConnection conn;
20	static int k=0, k_sc=0, k_bb=0;
21	static int minDmeansSC=10; // above this threshold contacts are assumed to be SC dominated
22	static String backgrndDB = "cullpdb_90";
23
24	public static void main(String[] args) throws SQLException {
25	String nbhood = "", msql, cid, res, sectype;
26	int graph_id=0, node_id=0, resnr=0;
27	// System.out.println("rewriteNBHood v.0.2");
28	conn = new MySQLConnection("white",user,"nieve", backgrndDB);
29	Statement mstmt;
30	ResultSet mrsst;
31
32	try {
33	msql = "select graph_id, node_id, cid, num, res, sstype from single_model_node where sstype is not null;";
34	// System.out.println("nodes> "+ msql);
35	mstmt = conn.createStatement();
36	mrsst = mstmt.executeQuery(msql);
37
38	while (mrsst.next()) {
39	// System.out.print(">");
40	graph_id=mrsst.getInt(1);
41	node_id =mrsst.getInt(2);
42	cid =mrsst.getString(3);
43	resnr =mrsst.getInt(4);
44	res =mrsst.getString(5);
45	sectype =mrsst.getString(6);
46	System.out.print(graph_id); // graph
47	System.out.print("\t"+node_id); // node
48	System.out.print("\t"+cid); // cid
49	System.out.print("\t"+resnr); // num
50	System.out.print("\t"+res); // res
51	System.out.print("\t"+sectype); // sstype
52
53	nbhood= getSSNbHoodString( graph_id, cid, resnr);
54	System.out.print("\t"+k);
55	System.out.print("\t"+k_bb);
56	System.out.print("\t"+k_sc);
57	System.out.print("\t"+nbhood);
58	System.out.print("\t"+gaphood);
59	System.out.print("\t"+fullnbs);
60	System.out.print("\t"+gapnbs);
61	// System.out.print("\t"+nums);
62	System.out.println();
63
64	}
65	mrsst.close();
66	mstmt.close();
67
68	} catch (SQLException e) {
69	e.printStackTrace();
70	System.err.println("SQLException: " + e.getMessage());
71	System.err.println("SQLState: " + e.getSQLState());
72	}
73
74	// System.out.println("fin.");
75	}
76
77	public static String getSSNbHoodString( int gid, String cid, int num ) {
78	int j_num, j_bb, j_sc, lastnum=0;
79	boolean overx=false;
80	String sql, nbs="", j_res, j_ss, nextnb, gapsym="", xgapsym="";
81	Statement stmt;
82	ResultSet rsst;
83
84	try {
85	k=0; k_sc=0; k_bb=0; // initialising the degreecounts with 0
86	sql = "select j_num, j_res, j_sstype, weight_BB, weight_SC from single_model_edge where graph_id="+gid+" and i_cid='"+cid+"' and i_num="+num+" order by j_num;";
87	stmt = conn.createStatement();
88	rsst = stmt.executeQuery(sql);
89	fullnbs="";
90	gapnbs="";
91	gaphood="";
92	nums="";
93	while (rsst.next()) {
94	gapsym="";
95	xgapsym="";
96	j_num = rsst.getInt(1);
97	j_res = rsst.getString(2);
98	j_ss = rsst.getString(3);
99	j_bb = rsst.getInt(4);
100	j_sc = rsst.getInt(5);
101	// System.out.print("\n-> "+j_res+" "+j_num+" "+j_ss+" "+"("+j_sc+"/"+j_bb+")");
102	nextnb="?";
103	if (lastnum>0 && ((j_num-lastnum)>=2)) gapsym="%";
104	if (!overx && j_num>num) {
105	overx=true;
106	nbs+="x";
107	fullnbs+="x";
108	nums+=(" ("+num+") ");
109	if (lastnum>0 && ((num-lastnum)>2)) xgapsym="%";
110	gapnbs+=xgapsym+"x";
111	gaphood+=xgapsym+"x";
112	gapsym="";
113	if ((j_num-num)>=2) gapsym="%";
114	}
115	// conversion of BB dominated contacts
116	if (j_sc==0 && j_bb==0) { // both values are 0 -> decide by sequence sparation
117	// System.out.print("SC&BB=0");
118	if ( Math.abs( num-j_num)>=minDmeansSC) {
119	nextnb=j_res;
120	k_sc++;
121	// System.out.print(" -> SC ");
122	} else {
123	nextnb=xlateSS( j_ss);
124	k_bb++;
125	// System.out.print(" -> BB ");
126	} // end if Math.abs( num-j_num)>=minDmeansSC
127	} else { // at least 1 value is non zero
128	// System.out.print("SC>0 or BB>0");
129	if ( (j_bb>j_sc) ) { // bb dominated -> only symbol for sstype
130	k_bb++;
131	nextnb=xlateSS( j_ss);
132	// System.out.print(" -> BB ");
133	} else { // sc dominated contact
134	k_sc++;
135	nextnb=j_res;
136	// System.out.print(" -> SC ");
137	} // end if bb dominated
138	} // end if we have non-zero entries
139	// System.out.print(" : "+nextnb);
140	nbs+= nextnb;
141	nums+=" "+j_num;
142	fullnbs+= j_res;
143	k++;
144
145	gapnbs+=gapsym+j_res;
146	gaphood+=gapsym+nextnb;
147	lastnum=j_num;
148	// System.out.print( nextnb);
149	}
150	rsst.close();
151	stmt.close();
152	if (!overx) {
153	nbs+="x"; // add if last in string == not seen yet
154	fullnbs+="x";
155	if (lastnum>0 && ((num-lastnum)==2)) xgapsym="_";
156	if (lastnum>0 && ((num-lastnum)>2)) xgapsym="%";
157	gapnbs+=xgapsym+"x";
158	gaphood+=xgapsym+"x";
159	}
160
161	} catch (SQLException e) {
162	e.printStackTrace();
163	System.err.println("SQLException: " + e.getMessage());
164	System.err.println("SQLState: " + e.getSQLState());
165	}
166	return nbs;
167	} // end of getnbhoodstring
168
169	public static String xlateSS( String sstype) { // xlate SecSTructure-Symbol
170	String SSymbol = "o";
171	sstype=sstype.toUpperCase();
172	if (sstype.equals("H")) SSymbol="z"; // Helix
173	if (sstype.equals("S")) SSymbol="b"; // beta strand
174	return SSymbol;
175	} // end xlateSS
176
177	} // end
178
179	/* table definition for the results
180
181	CREATE TABLE `nbstrings` (
182	`graph_id` int(10) unsigned DEFAULT NULL,
183	`node_id` int(10) unsigned DEFAULT NULL,
184	`cid` varchar(6) COLLATE latin1_general_cs DEFAULT NULL,
185	`num` int(10) unsigned DEFAULT NULL,
186	`res` char(1) COLLATE latin1_general_cs DEFAULT NULL,
187	`sstype` char(1) COLLATE latin1_general_cs DEFAULT NULL,
188	`k` int(10) unsigned DEFAULT NULL,
189	`k_BB` int(10) unsigned DEFAULT NULL,
190	`k_SC` int(10) unsigned DEFAULT NULL,
191	`n` varchar(30) COLLATE latin1_general_cs DEFAULT NULL,
192	`nwg` varchar(40) COLLATE latin1_general_cs DEFAULT NULL,
193	`f` varchar(30) COLLATE latin1_general_cs DEFAULT NULL,
194	`fwg` varchar(40) COLLATE latin1_general_cs DEFAULT NULL
195	) ENGINE=MyISAM DEFAULT CHARSET=latin1 COLLATE=latin1_general_cs;
196
197	*/