owl/trunk/listInfoGain.java

import tools.MySQLConnection;

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

public class listInfoGain {

        /**
         * "Hello World" for entropy calculations given a nbhoodstring 
         * 
         * @author lappe
         */
        
        static String user = "lappe"    ; // change user name!!
        static MySQLConnection conn;
        
        public static void main(String[] args) {
                double entropy = 0.0, newentropy=0.0, gain, mingain=99999999.9; 
                String nbhood = "", front, middle, tail, newhood="";
                if (args.length<1){
                        System.err.println("The starting neighborhood-string needs to be given .... i.e. %K%D%L%I%x%D%C%");
                        System.exit(1);
                }               
                nbhood = args[0];
                
                conn = new MySQLConnection("white",user,"nieve","pdb_reps_graph_4_2"); 
                int l = (int)((nbhood.length()-1)/2); 
                int N = 1; 
                        System.out.println("ListInfoGain");
                        System.out.print("0 - (%x%)("+l+") "); 
                        entropy = getEntropy( "%x%"); 
                        System.out.println( entropy + " bits."); 
                        System.out.println("Symbols in nbhood :"+l); 
                        
                        for (int i = 0; i<l; i++) {
                                front = nbhood.substring(0,i*2); 
                                middle = nbhood.substring(i*2, i*2+2); 
                                tail = nbhood.substring(i*2+2);
                                System.out.print((i+1)+" - "+front+"("+middle+")"+tail);
                                
                                if (middle.equals("%x")) { // switch from N to C of X  
                                        N = -1; 
                                        newhood = "%x%"; 
                                } else  {
                                   if (N < 0) // we are in the C-terminal section 
                                       newhood = "%x"+middle+"%"; 
                                   else // N terminal (before X) 
                                           newhood = middle+"%x%";
                                } // end if    
                                System.out.print( " -> " + newhood); 
                                newentropy = getEntropy( newhood); 
                                gain = newentropy-entropy; 
                                System.out.println( " : "+newentropy +"bits, gain="+gain);
                                
                        } // next symbol in nbhood; 
                        
                
         
                System.out.println("fin."); 
        }       

                
        
        public static double getEntropy( String nbs) {
                int total = 0, num; 
                String sql, res; 
                Statement stmt;  
                ResultSet rsst;
                double p, psum=0.0, logp, plogp, plogpsum=0.0; 
                
                try {
                        sql = "select count(*) from single_model_node where n like '"+nbs+"';";
                        // System.out.println( sql); 
                        stmt = conn.createStatement();
                        rsst = stmt.executeQuery(sql);
                        if (rsst.next()) total = rsst.getInt( 1); 
                        rsst.close(); 
                        stmt.close(); 
                        
                        sql = "select res, count(*) as t, count(*)/"+total+" as p from single_model_node where n like '"+nbs+"' group by res order by res;";
                        stmt = conn.createStatement();
                        rsst = stmt.executeQuery(sql);
                        // System.out.println("res : total t : fraction p : log2(p) : -p*log2(p)");
                        while (rsst.next()) {                           
                                res = rsst.getString(1); // 1st column -- res
                                num = rsst.getInt(2); // 2nd column -- num
                                p = rsst.getDouble(3); // 3rd: fraction p 
                                // System.out.print(res+"   : "+num+ " : " + p);
                                logp = Math.log(p)/Math.log(2.0); // to basis 2 for info in bits 
                                // System.out.print(" : " + logp); 
                                plogp = -1.0 * p * logp;  
                                // System.out.print(" : " + plogp);
                                plogpsum += plogp;  
                                psum += p; 
                                // System.out.println("");
                        }
                        // System.out.println("Sum :"+total+"       : "+psum+"       : "+plogpsum);
                        rsst.close(); 
                        stmt.close(); 
                        
                } catch (SQLException e) {
                        e.printStackTrace();
                        System.err.println("SQLException: " + e.getMessage());
                        System.err.println("SQLState:     " + e.getSQLState());
                }
                return plogpsum; 
        } // end of getEntropy 

}
Revision:	151
Committed:	Wed May 16 14:05:48 2007 UTC (17 years, 4 months ago) by lappe
File size:	3346 byte(s)
Log Message:	working version for each single nbring residue
Line	User	Rev	File contents
1	lappe	148	import tools.MySQLConnection;
2
3			import java.sql.SQLException;
4			import java.sql.Statement;
5			import java.sql.ResultSet;
6
7			public class listInfoGain {
8
9			/**
10			* "Hello World" for entropy calculations given a nbhoodstring
11			*
12			* @author lappe
13			*/
14
15			static String user = "lappe" ; // change user name!!
16			static MySQLConnection conn;
17
18			public static void main(String[] args) {
19			double entropy = 0.0, newentropy=0.0, gain, mingain=99999999.9;
20			String nbhood = "", front, middle, tail, newhood="";
21			if (args.length<1){
22			System.err.println("The starting neighborhood-string needs to be given .... i.e. %K%D%L%I%x%D%C%");
23			System.exit(1);
24			}
25			nbhood = args[0];
26
27			conn = new MySQLConnection("white",user,"nieve","pdb_reps_graph_4_2");
28			int l = (int)((nbhood.length()-1)/2);
29			int N = 1;
30			System.out.println("ListInfoGain");
31	lappe	151	System.out.print("0 - (%x%)("+l+") ");
32			entropy = getEntropy( "%x%");
33	lappe	148	System.out.println( entropy + " bits.");
34			System.out.println("Symbols in nbhood :"+l);
35
36			for (int i = 0; i<l; i++) {
37			front = nbhood.substring(0,i*2);
38			middle = nbhood.substring(i2, i2+2);
39			tail = nbhood.substring(i*2+2);
40			System.out.print((i+1)+" - "+front+"("+middle+")"+tail);
41
42			if (middle.equals("%x")) { // switch from N to C of X
43			N = -1;
44	lappe	151	newhood = "%x%";
45	lappe	148	} else {
46			if (N < 0) // we are in the C-terminal section
47			newhood = "%x"+middle+"%";
48			else // N terminal (before X)
49			newhood = middle+"%x%";
50			} // end if
51	lappe	151	System.out.print( " -> " + newhood);
52	lappe	148	newentropy = getEntropy( newhood);
53			gain = newentropy-entropy;
54			System.out.println( " : "+newentropy +"bits, gain="+gain);
55
56			} // next symbol in nbhood;
57
58
59
60			System.out.println("fin.");
61			}
62
63
64
65			public static double getEntropy( String nbs) {
66			int total = 0, num;
67			String sql, res;
68			Statement stmt;
69			ResultSet rsst;
70			double p, psum=0.0, logp, plogp, plogpsum=0.0;
71
72			try {
73			sql = "select count(*) from single_model_node where n like '"+nbs+"';";
74			// System.out.println( sql);
75			stmt = conn.createStatement();
76			rsst = stmt.executeQuery(sql);
77			if (rsst.next()) total = rsst.getInt( 1);
78			rsst.close();
79			stmt.close();
80
81			sql = "select res, count() as t, count()/"+total+" as p from single_model_node where n like '"+nbs+"' group by res order by res;";
82			stmt = conn.createStatement();
83			rsst = stmt.executeQuery(sql);
84			// System.out.println("res : total t : fraction p : log2(p) : -p*log2(p)");
85			while (rsst.next()) {
86			res = rsst.getString(1); // 1st column -- res
87			num = rsst.getInt(2); // 2nd column -- num
88			p = rsst.getDouble(3); // 3rd: fraction p
89			// System.out.print(res+" : "+num+ " : " + p);
90			logp = Math.log(p)/Math.log(2.0); // to basis 2 for info in bits
91			// System.out.print(" : " + logp);
92			plogp = -1.0 * p * logp;
93			// System.out.print(" : " + plogp);
94			plogpsum += plogp;
95			psum += p;
96			// System.out.println("");
97			}
98			// System.out.println("Sum :"+total+" : "+psum+" : "+plogpsum);
99			rsst.close();
100			stmt.close();
101
102			} catch (SQLException e) {
103			e.printStackTrace();
104			System.err.println("SQLException: " + e.getMessage());
105			System.err.println("SQLState: " + e.getSQLState());
106			}
107			return plogpsum;
108			} // end of getEntropy
109
110			}