| 1 |
import tools.MySQLConnection; |
| 2 |
|
| 3 |
import java.sql.SQLException; |
| 4 |
import java.sql.Statement; |
| 5 |
import java.sql.ResultSet; |
| 6 |
|
| 7 |
public class maxInfoGain { |
| 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) throws SQLException { |
| 19 |
double entropy = 0.0, newentropy=0.0, gain, maxgain=0.0; |
| 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 |
|
| 30 |
while (l>=2) { |
| 31 |
System.out.println("MaxInfoGain"); |
| 32 |
System.out.print("0 - ("+nbhood+")("+l+") "); |
| 33 |
entropy = getEntropy( nbhood); |
| 34 |
System.out.println( entropy + " bits."); |
| 35 |
System.out.println("Symbols in nbhood :"+l); |
| 36 |
maxgain = 0.0; |
| 37 |
for (int i = 0; i<l; i++) { |
| 38 |
front = nbhood.substring(0,i*2); |
| 39 |
middle = nbhood.substring(i*2, i*2+2); |
| 40 |
tail = nbhood.substring(i*2+2); |
| 41 |
System.out.print((i+1)+" - "+front+"("+middle+")"+tail); |
| 42 |
newentropy = getEntropy( (front+tail)); |
| 43 |
gain = newentropy-entropy; |
| 44 |
System.out.print( " : "+newentropy +"bits, gain="+gain); |
| 45 |
if (gain>maxgain && !(middle.equals("%x"))) { |
| 46 |
maxgain = gain; |
| 47 |
newhood = front+tail; |
| 48 |
System.out.println(" <<-"); |
| 49 |
} else { |
| 50 |
System.out.println(" ."); |
| 51 |
} |
| 52 |
} // next symbol in nbhood; |
| 53 |
if (newhood.equals(nbhood)) { // in case we got stcuk with the same string |
| 54 |
l=0; // - exit loop |
| 55 |
} else { // continue with new nbhoodstring |
| 56 |
nbhood = newhood; |
| 57 |
l = (int)((nbhood.length()-1)/2); |
| 58 |
} |
| 59 |
System.out.println("-->> "+nbhood+"("+l+")"); |
| 60 |
} // |
| 61 |
System.out.println("fin."); |
| 62 |
} |
| 63 |
|
| 64 |
|
| 65 |
|
| 66 |
public static double getEntropy( String nbs) { |
| 67 |
int total = 0; |
| 68 |
String sql; |
| 69 |
Statement stmt; |
| 70 |
ResultSet rsst; |
| 71 |
double p, psum=0.0, logp, plogp, plogpsum=0.0; |
| 72 |
|
| 73 |
try { |
| 74 |
sql = "select count(*) from single_model_node where n like '"+nbs+"';"; |
| 75 |
// System.out.println( sql); |
| 76 |
stmt = conn.createStatement(); |
| 77 |
rsst = stmt.executeQuery(sql); |
| 78 |
if (rsst.next()) total = rsst.getInt( 1); |
| 79 |
rsst.close(); |
| 80 |
stmt.close(); |
| 81 |
|
| 82 |
sql = "select res, count(*) as t, count(*)/"+total+" as p from single_model_node where n like '"+nbs+"' group by res order by res;"; |
| 83 |
stmt = conn.createStatement(); |
| 84 |
rsst = stmt.executeQuery(sql); |
| 85 |
// System.out.println("res : total t : fraction p : log2(p) : -p*log2(p)"); |
| 86 |
while (rsst.next()) { |
| 87 |
//res = rsst.getString(1); // 1st column -- res |
| 88 |
//num = rsst.getInt(2); // 2nd column -- num |
| 89 |
p = rsst.getDouble(3); // 3rd: fraction p |
| 90 |
// System.out.print(res+" : "+num+ " : " + p); |
| 91 |
logp = Math.log(p)/Math.log(2.0); // to basis 2 for info in bits |
| 92 |
// System.out.print(" : " + logp); |
| 93 |
plogp = -1.0 * p * logp; |
| 94 |
// System.out.print(" : " + plogp); |
| 95 |
plogpsum += plogp; |
| 96 |
psum += p; |
| 97 |
// System.out.println(""); |
| 98 |
} |
| 99 |
// System.out.println("Sum :"+total+" : "+psum+" : "+plogpsum); |
| 100 |
rsst.close(); |
| 101 |
stmt.close(); |
| 102 |
|
| 103 |
} catch (SQLException e) { |
| 104 |
e.printStackTrace(); |
| 105 |
System.err.println("SQLException: " + e.getMessage()); |
| 106 |
System.err.println("SQLState: " + e.getSQLState()); |
| 107 |
} |
| 108 |
return plogpsum; |
| 109 |
} // end of getEntropy |
| 110 |
|
| 111 |
} |
