| 1 |
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 |
static double orgFreq, lastFreq, orgAUC, lastAUC; |
| 18 |
static int orgRank, lastRank; |
| 19 |
|
| 20 |
public static void main(String[] args) throws SQLException { |
| 21 |
double entropy = 0.0, newentropy=0.0, gain; |
| 22 |
String nbhood = "", front, middle, tail, newhood="", central ="I"; |
| 23 |
if (args.length<2){ |
| 24 |
System.err.println("The starting neighborhood-string and central residue type needs to be given .... i.e. %K%D%L%I%x%D%C% I"); |
| 25 |
System.exit(1); |
| 26 |
} |
| 27 |
nbhood = args[0]; |
| 28 |
central = args[1]; |
| 29 |
|
| 30 |
conn = new MySQLConnection("white",user,"nieve","pdb_reps_graph_4_2"); |
| 31 |
int l = (int)((nbhood.length()-1)/2); |
| 32 |
int N = 1; |
| 33 |
System.out.println("ListInfoGain"); |
| 34 |
System.out.print("0 - (%x%)("+l+") "); |
| 35 |
entropy = getEntropy( "%x%", central); |
| 36 |
System.out.print( entropy + " bits."); |
| 37 |
// System.out.println("Symbols in nbhood :"+l); |
| 38 |
orgFreq = lastFreq; |
| 39 |
orgAUC = lastAUC; |
| 40 |
orgRank = lastRank; |
| 41 |
System.out.println("central Residue "+central+ " rank#"+lastRank+" p="+lastFreq+" AUC="+String.format("%.5f",lastAUC)); |
| 42 |
// System.out.println("Symbols in nbhood :"+l); |
| 43 |
|
| 44 |
for (int i = 0; i<l; i++) { |
| 45 |
front = nbhood.substring(0,i*2); |
| 46 |
middle = nbhood.substring(i*2, i*2+2); |
| 47 |
tail = nbhood.substring(i*2+2); |
| 48 |
System.out.print((i+1)+" - "+front+"("+middle+")"+tail); |
| 49 |
|
| 50 |
if (middle.equals("%x")) { // switch from N to C of X |
| 51 |
N = -1; |
| 52 |
newhood = "%x%"; |
| 53 |
} else { |
| 54 |
if (N < 0) // we are in the C-terminal section |
| 55 |
newhood = "%x"+middle+"%"; |
| 56 |
else // N terminal (before X) |
| 57 |
newhood = middle+"%x%"; |
| 58 |
} // end if |
| 59 |
System.out.print( " -> " + newhood); |
| 60 |
newentropy = getEntropy( newhood, central); |
| 61 |
gain = newentropy-entropy; |
| 62 |
System.out.print( " : "+String.format("%.5f", newentropy) +"bits ("+String.format("%.5f", gain)+")"); |
| 63 |
System.out.print( " : #"+lastRank +" ("+(lastRank-orgRank)+")"); |
| 64 |
System.out.print( " : "+String.format("%.5f", lastFreq) +"("+String.format("%.5f", (lastFreq-orgFreq))+")"); |
| 65 |
System.out.print( " : "+String.format("%.5f", lastAUC) +"("+String.format("%.5f", (lastAUC-orgAUC))+")"); |
| 66 |
|
| 67 |
System.out.println( "."); |
| 68 |
} // next symbol in nbhood; |
| 69 |
|
| 70 |
System.out.println("fin."); |
| 71 |
} |
| 72 |
|
| 73 |
|
| 74 |
|
| 75 |
public static double getEntropy( String nbs, String centRes) { |
| 76 |
int total = 0; |
| 77 |
String sql, res; |
| 78 |
Statement stmt; |
| 79 |
ResultSet rsst; |
| 80 |
double p, psum=0.0, logp, plogp, plogpsum=0.0; |
| 81 |
|
| 82 |
try { |
| 83 |
sql = "select count(*) from single_model_node where n like '"+nbs+"';"; |
| 84 |
// System.out.println( sql); |
| 85 |
stmt = conn.createStatement(); |
| 86 |
rsst = stmt.executeQuery(sql); |
| 87 |
if (rsst.next()) total = rsst.getInt( 1); |
| 88 |
rsst.close(); |
| 89 |
stmt.close(); |
| 90 |
|
| 91 |
sql = "select res, count(*) as t, count(*)/"+total+" as p from single_model_node where n like '"+nbs+"' group by res order by p DESC;"; |
| 92 |
stmt = conn.createStatement(); |
| 93 |
rsst = stmt.executeQuery(sql); |
| 94 |
// System.out.println("rank : res : total t : fraction p : log2(p) : -p*log2(p)"); |
| 95 |
int rank = 0; |
| 96 |
boolean seenCentRes = false; |
| 97 |
lastAUC= 0; |
| 98 |
while (rsst.next()) { |
| 99 |
rank ++; |
| 100 |
res = rsst.getString(1); // 1st column -- res |
| 101 |
//num = rsst.getInt(2); // 2nd column -- num |
| 102 |
p = rsst.getDouble(3); // 3rd: fraction p |
| 103 |
// System.out.print(rank+ " : " + res+" : "+num+ " : " + p); |
| 104 |
logp = Math.log(p)/Math.log(2.0); // to basis 2 for info in bits |
| 105 |
// System.out.print(" : " + logp); |
| 106 |
plogp = -1.0 * p * logp; |
| 107 |
// System.out.print(" : " + plogp); |
| 108 |
plogpsum += plogp; |
| 109 |
psum += p; |
| 110 |
|
| 111 |
if (res.equals(centRes)) { |
| 112 |
// System.out.print(" <==" + centRes); |
| 113 |
seenCentRes = true; |
| 114 |
lastFreq = p; |
| 115 |
lastRank = rank; |
| 116 |
} |
| 117 |
if (seenCentRes) lastAUC += p; |
| 118 |
// System.out.println(""); |
| 119 |
} |
| 120 |
// System.out.println("Sum :"+total+" : "+psum+" : "+plogpsum); |
| 121 |
rsst.close(); |
| 122 |
stmt.close(); |
| 123 |
|
| 124 |
} catch (SQLException e) { |
| 125 |
e.printStackTrace(); |
| 126 |
System.err.println("SQLException: " + e.getMessage()); |
| 127 |
System.err.println("SQLState: " + e.getSQLState()); |
| 128 |
} |
| 129 |
return plogpsum; |
| 130 |
} // end of getEntropy |
| 131 |
|
| 132 |
} |
