ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/owl/trunk/cnEvalTargetNbhood.java
Revision: 243
Committed: Wed Aug 1 14:21:09 2007 UTC (17 years, 2 months ago) by lappe
File size: 23363 byte(s)
Log Message: 
Added correction by matrix size (yields avgdeltaRank)
plus correction jcnsize-1 for direct nbs 
and pessimistic tie-break
Line User Rev File contents
1 lappe 227 import tools.MySQLConnection;
2    
3     import java.sql.SQLException;
4     import java.sql.Statement;
5     import java.sql.ResultSet;
6    
7     public class cnEvalTargetNbhood {
8    
9     /**
10     * CN2 size x sumdelta
11     * iterating over all steps by exchange of one neighbor at a time by a common neighbor
12     * with subsequent scoring
13     * @author lappe
14     */
15     static int maxRank = 21; // value to replace for non-existence of central redue in the resultvector (rank=0)
16     // higher values should penalize non-existence more
17 lappe 231 static int VL=1; // Verbosity Level
18 lappe 227 static String user = "lappe" ; // change user name!!
19     static MySQLConnection conn;
20     static String backgrndDB = "pdb_reps_graph_4_2";
21     static String targetNodes = "target_node";
22     static String targetEdges = "target_edge";
23     static double lastEntropy=0.0, lastFreq, lastAUC, lastavgk, lastdevk;
24     static int lastRank, lastTotal, lastxcn=0;
25     static int graphid=0, resnr=0;
26     static int size1=0, size2=0; // the dimensions of the matrices = |shell1|x|shell2| = nr of direct(1) x indirect(2) nbs
27     static String rsideline[], rheadline[]={" -->\t(-)1st shell\t0"," |\t \t-"," |\t(+) \t0"," V\t2nd shell\t-", " \t \tX"};
28     static String sideline[], headline[]={"\t\t0","\t\t-","\t\t0","\t\t-","\t\tX"};
29     static String restype="?", ressec="?", newnbhood="";
30 lappe 243 static int cn1[], cn2[], matsize[][];
31 lappe 227 static int sumdelta[][], rank[][], total[][], cnsize[][], cnall[][];
32     static double entropy[][], freq[][], AUC[][];
33     static String newnbs[][], nbstring[][], moveset[][];
34    
35     static int printRank=1,
36     printTotal = 2,
37     printEntropy = 3,
38     printFreq = 4,
39     printAUC = 5,
40     printNbstring= 6,
41     printMoveset = 7,
42     printdeltaRank=8,
43     printCNSize =9,
44 lappe 243 printCNSxdelta=10,
45     printMatSize =11,
46     printCNSdMSxdelta=12;
47 lappe 227
48     public static void main(String[] args) {
49    
50     if (args.length<2){
51     System.err.println("The graph_id and residue-nr. needs to be given .... i.e. 7 110");
52     System.exit(1);
53     }
54     graphid = Integer.parseInt( args[0]);
55     resnr = Integer.parseInt( args[1]);
56     int j_num=0, oj_num=0, oj_shell, oj_cnsize, i, j, x, oi, oj, o1, o2, ixnum, jxnum, score=0;
57     boolean overx = false;
58     String sql, oj_res, oj_sec;
59     Statement mstmt, mjst, nstmt;
60     ResultSet mrsst, nrsst;
61    
62     try {
63     conn = new MySQLConnection("white",user,"nieve", backgrndDB); // the UPPERCASE DB!
64     System.out.println("Scoring Target neighborhoods v.0.4. ");
65    
66     sql = "select num, res, sstype from "+targetNodes+" where graph_id="+graphid+" and num="+resnr+";";
67     mstmt = conn.createStatement();
68     mrsst = mstmt.executeQuery(sql);
69     if (mrsst.next()) {
70     // this is the central node -> get type and secondary structure
71     restype = mrsst.getString( 2).toUpperCase();
72     ressec = mrsst.getString( 3).toUpperCase();
73     } // end if central residue
74     mrsst.close();
75     mstmt.close();
76     System.out.println("GraphID "+graphid+" Central residue is "+restype+":"+resnr+":"+ressec);
77    
78     // retrieve the original nbhood into orig_shell
79     System.out.println("retrieving original first shell ... ");
80     mstmt = conn.createStatement();
81     mstmt.executeUpdate("drop table if exists orig_shell;");
82     mstmt.close();
83    
84     mstmt = conn.createStatement();
85     mstmt.executeUpdate("create table orig_shell as select i_num, i_res, j_num, j_res, j_sstype, 1 as shell from "+targetEdges+" where graph_id="+graphid+" and i_num="+resnr+";");
86     mstmt.close();
87    
88     System.out.println("adding the original 2nd shell ...");
89     sql = "select j_num from orig_shell where shell=1;";
90     mstmt = conn.createStatement();
91     mrsst = mstmt.executeQuery(sql);
92     i=0;
93     while (mrsst.next()) {
94     i++;
95     oj_num = mrsst.getInt(1);
96     System.out.println(i+":"+oj_num);
97     mjst = conn.createStatement();
98     sql = "insert into orig_shell select i_num, i_res, j_num, j_res, j_sstype, 2 as shell from "+targetEdges+" where graph_id="+graphid+" and i_num="+oj_num+";";
99     // System.out.println(">"+sql);
100     mjst.executeUpdate( sql);
101     mjst.close();
102     } // end while
103     mrsst.close();
104     mstmt.close();
105    
106     System.out.println("gathering the original 1st and 2nd shell nbs.");
107     sql = "select j_num, j_res, j_sstype, min(shell) as shell, count(*) as cn from orig_shell where j_num!="+resnr+" group by j_num order by j_num;";
108     mstmt = conn.createStatement();
109     mrsst = mstmt.executeQuery(sql);
110     o1=0;
111     o2=0;
112     while (mrsst.next()) {
113     if ( mrsst.getInt( 4)==1) { // count 1st shell entry
114     o1++;
115     System.out.print("1#"+o1);
116     rheadline[0]+="\t"+o1;
117     rheadline[1]+="\t"+mrsst.getString(2); // res
118     rheadline[2]+="\t"+mrsst.getInt(1); // resnum
119     rheadline[3]+="\t"+mrsst.getString(3); // SStype
120     rheadline[4]+="\t("+mrsst.getInt(5)+")"; // CN
121     } // end if 2st shell
122     if ( mrsst.getInt( 4)==2) { // count 2nd shell entry
123     o2++;
124     System.out.print("2#"+o2);
125     } // end if 2nd shell
126     System.out.println(" :\t"+mrsst.getInt( 1)+"\t"+mrsst.getString( 2)+"\t"+mrsst.getString( 3)+"\t"+mrsst.getInt( 4)+"\t"+mrsst.getInt( 5));
127     } // end while
128     System.out.println("Orig.SIZE 1st shell "+o1);
129     System.out.println("Orig.SIZE 2nd shell "+o2);
130     rheadline[4] = rheadline[4].replace("X", ("("+o1)+")");
131     rsideline = new String[o2+1];
132     rsideline[0]="+0\tRnum:S(cn)";
133     sumdelta = new int[(o1+1)][(o2+1)];
134     newnbs = new String[(o1+1)][(o2+1)];
135     cnall = new int[(o1+1)][(o2+1)];
136 lappe 243 matsize = new int[(o1+1)][(o2+1)];
137 lappe 227 // creating the perturbed version of shell 1 into temp_shell
138     for (j=0; j<=o2; j++) { // <=o2 outer loop through all originally indirect contacts
139    
140     for (i=0; i<=o1; i++) { // inner loop through all originally direct contacts
141     if (VL>=1) {
142     System.out.println("---------------------------------------------");
143     System.out.println("Creating perturbed nbhood ("+i+","+j+")\t");
144     }
145     // clear first
146     nstmt = conn.createStatement();
147     nstmt.executeUpdate("drop table if exists temp_shell;");
148     nstmt.close();
149     nstmt = conn.createStatement();
150     nstmt.executeUpdate("create table temp_shell select * from orig_shell limit 0;");
151     nstmt.close();
152     oi = 0;
153     oj = 0;
154     mrsst.beforeFirst();
155     newnbhood="";
156     overx = false;
157     ixnum=0;
158     jxnum=0;
159     while (mrsst.next()) {
160     oj_num = mrsst.getInt( 1);
161     oj_res = mrsst.getString(2);
162     oj_sec = mrsst.getString(3);
163     oj_shell = mrsst.getInt( 4);
164     oj_cnsize = mrsst.getInt( 5);
165     if (oj_num>resnr) { // we are over x
166     if (!overx) {
167     newnbhood+="x";
168     overx=true;
169     } // end if over x
170     } // END IF J > X
171     if (oj_shell==1) { // a direct 1st shell neighbour
172     oi++;
173     if (oi!=i) {// if this is NOT the one direct nb 2B dropped
174     // include as 1st shell nbor into temp_shell
175     nstmt = conn.createStatement();
176     sql = "insert into temp_shell values("+resnr+",\'"+restype+"\',"+oj_num+",\'"+oj_res+"\',\'"+oj_sec+"\', 1);";
177     // System.out.println("oi>"+ sql);
178     nstmt.executeUpdate(sql);
179     nstmt.close();
180     newnbhood+=oj_res;
181     } else {
182     ixnum=oj_num;
183     } // end if ni!=i
184     } else { // 2nd shell neighbour
185     oj++;
186     if (oj==j) { // this is the 2nd shell nb 2B included
187     // put as new 1st shell nbor
188     nstmt = conn.createStatement();
189     sql = "insert into temp_shell values("+resnr+",\'"+restype+"\',"+oj_num+",\'"+oj_res+"\',\'"+oj_sec+"\', 1);";
190     // System.out.println("oj>"+ sql);
191     nstmt.executeUpdate(sql);
192     nstmt.close();
193     newnbhood+=oj_res;
194     jxnum=oj_num;
195     } // end if
196     if (j==0) { // creating the sideline ruler array for the output
197     rsideline[oj] = "+"+oj+"\t"+oj_res+""+oj_num+":"+oj_sec+"("+oj_cnsize+")";
198     } // end if j==0
199     } // end if 1st/2nd shell
200    
201     } // end while through the entire nbhood
202     if (!overx) { // we haven't seen a nb > x yet
203     newnbhood+="x"; // x sits at the end of the nbhoodstring
204     overx=true;
205     } // end if over x
206     // System.out.println("new direct nbhood "+newnbhood);
207     // Now the "updated" / perturbed version of shell 1 is in temp_shell
208     // we can build 2nd shell accordingly.
209     // System.out.println("building the 2nd shell");
210     sql = "select j_num, j_res, j_sstype from temp_shell where shell=1;";
211     nstmt = conn.createStatement();
212     nrsst = nstmt.executeQuery(sql);
213     x = 0;
214     while (nrsst.next()) {
215     x++;
216     j_num = nrsst.getInt( 1);
217     // System.out.println(x+":"+nrsst.getString( 2)+" "+j_num+" "+nrsst.getString( 3));
218     mjst = conn.createStatement();
219     sql = "insert into temp_shell select i_num, i_res, j_num, j_res, j_sstype, 2 as shell from "+targetEdges+" where graph_id="+graphid+" and i_num="+j_num+";";
220     // System.out.println(">"+sql);
221     mjst.executeUpdate( sql);
222     mjst.close();
223     } // end while
224     nrsst.close();
225     nstmt.close();
226     // and score this move
227     lastxcn=0;
228     newnbs[i][j]= newnbhood;
229 lappe 231 System.out.print("\n["+i+"]["+j+"]\t-"+ixnum+"/+"+jxnum+"\t"+newnbs[i][j]+"\t ");
230 lappe 227 score = scoreCurrentNbhood( ixnum, jxnum);
231     sumdelta[i][j] = score;
232     cnall[i][j] = lastxcn;
233 lappe 243 matsize[i][j] = (size1*size2)-1;
234 lappe 227 if (VL>=1) {
235 lappe 243 reportMatrix( printRank );
236     // reportMatrix( printCNSize );
237 lappe 231 reportMatrix( printdeltaRank );
238 lappe 227 System.out.println("SumDeltaRank Score = \t"+score);
239 lappe 243 System.out.println("Matrix size n1*n2-1= \t("+size1+"*"+size2+")-1 ="+matsize[i][j]);
240 lappe 227 System.out.println("CN1 x CN2 product = \t"+cnall[i][j]);
241     } else {
242 lappe 231 System.out.print("\t"+score+"*"+cnall[i][j]+"\t= "+(score*cnall[i][j]));
243 lappe 227 }
244     // reportMatrix( printNbstring );
245     } // next i
246     System.out.println("\t");
247     } // next j
248     // report total matrix sumdelta
249 lappe 231 // if (VL>=1) {
250 lappe 227 System.out.println("GraphID "+graphid+" Central residue is "+restype+":"+resnr+":"+ressec);
251     System.out.println("backgroundDB"+backgrndDB+" \t maxRank : "+maxRank);
252     reportResults( o1, o2, printRank);
253     reportResults( o1, o2, printCNSize);
254 lappe 243 reportResults( o1, o2, printMatSize);
255 lappe 231 //}
256 lappe 243 reportResults( o1, o2, printCNSxdelta );
257     reportResults( o1, o2, printCNSdMSxdelta );
258 lappe 227 // Cleanup ...
259     mrsst.close();
260     mstmt.close();
261    
262     } catch (SQLException e) {
263     e.printStackTrace();
264     System.err.println("SQLException: " + e.getMessage());
265     System.err.println("SQLState: " + e.getSQLState());
266     } // end try/catch
267     System.out.println("fin.");
268     } // end main
269    
270    
271    
272     public static void reportResults( int o1, int o2, int what2print) {
273     System.out.println("Printing "+what2print);
274     if (what2print==printRank) System.out.println("Overall resulting SumDeltaRank Matrix" );
275     if (what2print==printNbstring) System.out.println("Overall tested new nbhoodStrings" );
276     if (what2print==printCNSize) System.out.println("Overall cnsize" );
277     if (what2print==printCNSxdelta) System.out.println("Overall (cnsize*sumDeltaRank)" );
278     /*if (what2print==printTotal) System.out.print("total[i][j]" );
279     if (what2print==printEntropy) System.out.print("entropy[i][j]" );
280     if (what2print==printFreq ) System.out.print("freq[i][j]" );
281     if (what2print==printAUC) System.out.print("AUC[i][j]" );
282    
283     if (what2print==printMoveset) System.out.print("moveset[i][j]" );
284     if (what2print==printdeltaRank) System.out.print("rank[i][j]-rank[0][0]" );) */
285    
286     // print headerline(s)
287     System.out.println(rheadline[0]);
288     System.out.println(rheadline[1]);
289     System.out.println(rheadline[2]);
290     System.out.println(rheadline[3]);
291     System.out.println(rheadline[4]);
292     for (int j=0; j<=o2; j++) {
293     // print rsideline
294     System.out.print( rsideline[j]+"\t");
295     for ( int i=0; i<=o1; i++) {
296     if (what2print==printRank) System.out.print( sumdelta[i][j] );
297     if (what2print==printNbstring) System.out.print( newnbs[i][j] );
298     if (what2print==printCNSize) System.out.print( cnall[i][j] );
299 lappe 243 if (what2print==printCNSxdelta) System.out.print( sumdelta[i][j]*cnall[i][j] );
300     if (what2print==printMatSize) System.out.print( matsize[i][j] );
301     if (what2print==printCNSdMSxdelta) System.out.print( sumdelta[i][j]*cnall[i][j]/matsize[i][j] );
302 lappe 227 System.out.print("\t");
303     } // next i
304     System.out.println("");
305     } // next j
306     } // end of reportResults
307    
308    
309    
310     public static int scoreCurrentNbhood( int ixnum, int jxnum) {
311     int ixcn=0, jxcn=0, n1=0, n2=0, ni, nj, i, j, j_num, j_shell, j_cnsize, sumdeltarank=0;
312     String sql, j_res, j_sec, nbs, mymove, precol;
313     boolean overx = false;
314     Statement stmt;
315     ResultSet rsst;
316    
317     try {
318     headline[0]="\t\t\t0";
319     headline[1]="\t\t\t-";
320     headline[2]="\t\t\t0";
321     headline[3]="\t\t\t-";
322     headline[4]="\t\t\tX";
323    
324     // System.out.println("retrieving the entire nbhood (1st and 2nd shell)");
325     sql = "select j_num, j_res, j_sstype, min(shell) as shell, count(*) as cn from temp_shell where j_num!="+resnr+" group by j_num order by j_num;";
326     stmt = conn.createStatement();
327     rsst = stmt.executeQuery(sql);
328     // counting shell2
329     n2=0;
330     while (rsst.next()) {
331     if ( rsst.getInt( 4)==1) { // count 1st shell entry
332     n1++;
333     // System.out.print("1#"+n1);
334     headline[0]+="\t"+n1;
335     headline[1]+="\t"+rsst.getString(2); // res
336     headline[2]+="\t"+rsst.getInt(1); // resnum
337     headline[3]+="\t"+rsst.getString(3); // SStype
338     headline[4]+="\t("+rsst.getInt(5)+")"; // CNSize
339     } // end if 2st shell
340     if ( rsst.getInt( 4)==2) { // count 2nd shell entry
341     n2++;
342     // System.out.print("2#"+n2);
343     } // end if 2nd shell
344     // System.out.println(" :\t"+rsst.getInt( 1)+"\t"+rsst.getString( 2)+"\t"+rsst.getString( 3)+"\t"+rsst.getInt( 4)+"\t"+rsst.getInt( 5));
345     } // end while
346     size1 = n1;
347     size2 = n2;
348     ixcn = n1;
349     jxcn = n1;
350     if (VL>=1) {
351     System.out.println("|1st shell|="+size1+" \tx\t |2nd shell|="+size2);
352     }
353     headline[4]=headline[4].replace("X",("("+size1+")"));
354    
355     // n1 and n2 are known, initialise matrices accordingly.
356     // nbhood, move, rank, entropy, freq, AUC etc. (evtl.+ degree(?))
357     rank = new int[(n1+1)][(n2+1)];
358     rank[0][0]=maxRank;
359     total = new int[(n1+1)][(n2+1)];
360     entropy = new double[(n1+1)][(n2+1)];
361     freq = new double[(n1+1)][(n2+1)];
362     AUC = new double[(n1+1)][(n2+1)];
363     nbstring = new String[(n1+1)][(n2+1)];
364     moveset = new String[(n1+1)][(n2+1)];
365     sideline = new String[n2+1];
366     cn1 = new int[n1+1];
367     cn2 = new int[n2+1];
368     cnsize = new int[(n1+1)][(n2+1)];
369    
370     for (j=0; j<=n2; j++) { // outer loop through all indirect contacts
371     for (i=0; i<=n1; i++) { // inner loop through all direct contacts
372     mymove = "";
373     overx = false;
374     if (VL>=1) {
375     System.out.print("("+i+","+j+")\t");
376     }
377     ni = 0;
378     nj = 0;
379     sideline[0]="+0\tRnum:S("+n1+")";
380     cn1[0]=n1;
381     cn2[0]=n1;
382     nbs="%";
383     rsst.beforeFirst();
384     while (rsst.next()) {
385     j_num = rsst.getInt(1);
386     j_res = rsst.getString(2);
387     j_sec = rsst.getString(3);
388     j_shell = rsst.getInt(4);
389     j_cnsize = rsst.getInt(5);
390     if (j_num>resnr) { // we are over x
391     if (!overx) {
392     nbs+="x%";
393     overx=true;
394     } // end if over x
395     } // END IF J > X
396     if (j_shell==1) { // a direct 1st shell neighbour
397     ni++;
398     if (ni!=i) {// if this is NOT the one direct nb 2B dropped
399     nbs+=j_res.toUpperCase()+"%"; // it is included
400     if ( j_num==jxnum && j==0) { // This is the direct nb dropped
401 lappe 243 jxcn=j_cnsize-1;
402 lappe 231 if (VL>=2) System.out.print("(j"+jxnum+":"+jxcn+")");
403 lappe 227 }
404     } else { // this one IS dropped
405     mymove += "(-"+j_res+":"+j_num+":"+j_sec+"/"+j_cnsize+")";
406     cn1[ni]=j_cnsize;
407     } // end if ni!=i
408    
409     } else { // 2nd shell neighbour
410     nj++;
411     if (nj==j) { // this is the 2nd shell nb 2B included
412     nbs+=j_res.toUpperCase()+"%";
413     mymove += "(+"+j_res+":"+j_num+":"+j_sec+"/"+j_cnsize+")";
414     if ( j_num==ixnum && i==0) { // This is the dropped direct nb, no 2b found in 2ns shell
415     ixcn=j_cnsize;
416 lappe 231 if (VL>=2) System.out.print("(i"+ixnum+":"+ixcn+")");
417 lappe 227 }
418     } // end if
419    
420     // // only once for building the sidelines
421     if (j==0) {
422     sideline[nj] = "+"+nj+"\t"+j_res+""+j_num+":"+j_sec+"("+j_cnsize+")";
423     cn2[nj] = j_cnsize;
424     } // end if sideline
425     } // end if 1st/2nd shell
426    
427     } // end while through the entire nbhood
428     if (!overx) { // in case x is the very last we haven't seen it yet
429     nbs+="x%"; // add it in the end
430     overx=true;
431     } // end if over x
432     if (VL>=1) {
433     System.out.print("("+nbs+")\t");
434     }
435     nbstring[i][j] = nbs;
436     moveset[i][j] = mymove;
437     precol = nbstring[i][0];
438     getEntropy( nbs, restype, precol);
439     if (lastRank==0) lastRank = maxRank;
440     rank[i][j] = lastRank;
441     entropy[i][j] = lastEntropy;
442     freq[i][j] = lastFreq;
443     AUC[i][j] = lastAUC;
444     total[i][j]= lastTotal;
445     cnsize[i][j]=cn1[i]*cn2[j];
446 lappe 243 // if (VL>=1) System.out.print(""+cnsize[i][j]+"\t");
447 lappe 227 if (lastRank > 0) {
448     sumdeltarank += ( (lastRank-rank[0][0]) );
449     // sumdeltarank += ( (lastRank-rank[0][0]) * (cnsize[i][j]) );
450     } else {
451     sumdeltarank += ( (maxRank-rank[0][0]) );
452     // sumdeltarank += ( (maxRank-rank[0][0]) * (cnsize[i][j]) );
453     } // end if lastRank was defined
454 lappe 243 if (VL>=1) System.out.print(""+lastRank+"\t");
455 lappe 227 } // close inner loop (i)
456     if (VL>=1) {
457     System.out.println(".");
458     } else {
459     System.out.print(".");
460     }
461     } // next outerloop (j)
462     lastxcn=(ixcn*jxcn);
463     if (VL>=1) {
464     System.out.println("lastxcn=(ixcn*jxcn)=("+ixcn+"*"+jxcn+")="+lastxcn);
465     }
466     rsst.close();
467     stmt.close();
468    
469     } catch (SQLException e) {
470     e.printStackTrace();
471     System.err.println("SQLException: " + e.getMessage());
472     System.err.println("SQLState: " + e.getSQLState());
473     } // end try/catch
474     // System.out.println("fin.");
475     return sumdeltarank;
476     } // end scoreCurrentNbhood
477    
478    
479     public static void reportMatrix( int what2print) {
480     System.out.println("\nPrinting "+what2print);
481     if (what2print==printRank) System.out.print("rank[i][j]" );
482     if (what2print==printTotal) System.out.print("total[i][j]" );
483     if (what2print==printEntropy) System.out.print("entropy[i][j]" );
484     if (what2print==printFreq ) System.out.print("freq[i][j]" );
485     if (what2print==printAUC) System.out.print("AUC[i][j]" );
486     if (what2print==printNbstring) System.out.print("nbstring[i][j]" );
487     if (what2print==printMoveset) System.out.print("moveset[i][j]" );
488     if (what2print==printdeltaRank) System.out.print("rank[i][j]-rank[0][0]" );
489     if (what2print==printCNSize) System.out.print("cnsize[i][j]" );
490     if (what2print==printCNSxdelta) System.out.print("cnsize[i][j]*(rank[i][j]-rank[0][0])" );
491    
492     System.out.println("...");
493     // print headerline(s)
494     System.out.println(headline[0]);
495     System.out.println(headline[1]);
496     System.out.println(headline[2]);
497     System.out.println(headline[3]);
498     System.out.println(headline[4]);
499     for (int j=0; j<=size2; j++) {
500     // print sideline
501     System.out.print( sideline[j]+"\t");
502     for ( int i=0; i<=size1; i++) {
503     if (what2print==printRank) System.out.print( rank[i][j] );
504     if (what2print==printTotal) System.out.print( total[i][j] );
505     if (what2print==printEntropy) System.out.print( entropy[i][j] );
506     if (what2print==printFreq ) System.out.print( freq[i][j] );
507     if (what2print==printAUC) System.out.print( AUC[i][j] );
508     if (what2print==printNbstring) System.out.print( nbstring[i][j] );
509     if (what2print==printMoveset) System.out.print( moveset[i][j] );
510     if (what2print==printdeltaRank) System.out.print( rank[i][j]-rank[0][0] );
511     if (what2print==printCNSize) System.out.print( cnsize[i][j] );
512     if (what2print==printCNSxdelta) System.out.print( cnsize[i][j]*(rank[i][j]-rank[0][0]) );
513     System.out.print("\t");
514     } // next i
515     System.out.println("");
516     } // next j
517     } // end of report
518    
519     public static void getEntropy( String nbs, String centRes, String predec) {
520     String sql, res, this_n, prec_n;
521     Statement stmt;
522     ResultSet rsst;
523 lappe 243 double p, psum=0.0, logp, plogp, plogpsum=0.0;
524     int counter=0, c=0, lastc=0;
525 lappe 227 try {
526     // Hashing first row tables comes first
527     // System.out.println("nbs : "+nbs);
528     // System.out.println("predec: "+predec);
529     this_n = nbs.replace("%","");
530     prec_n = predec.replace("%","");
531     // System.out.println("this_n: ["+this_n+"]");
532     // System.out.println("prec_n: ["+prec_n+"]");
533     if (prec_n.equals(this_n)) {
534     // System.out.println("have to create db for this "+prec_n);
535     sql = "create table IF NOT EXISTS nbhashtables."+prec_n+" as select res, n, k from single_model_node where n like '"+nbs+"';";
536     // System.out.println(" >> "+sql);
537     stmt = conn.createStatement();
538     stmt.executeUpdate( sql);
539     stmt.close();
540     } // else System.out.println("using preceding db of "+prec_n);
541    
542     // now we can safely derive the estimates from the hashtable
543     sql = "select count(*) from nbhashtables."+prec_n+" where n like '"+nbs+"';";
544     // System.out.println( sql);
545     stmt = conn.createStatement();
546     rsst = stmt.executeQuery(sql);
547     if (rsst.next()) lastTotal = rsst.getInt( 1);
548     rsst.close();
549     stmt.close();
550    
551     sql = "select res, count(*) as t, count(*)/"+lastTotal+" as p, avg( k), stddev( k) from nbhashtables."+prec_n+" where n like '"+nbs+"' group by res order by p DESC;";
552     stmt = conn.createStatement();
553     rsst = stmt.executeQuery(sql);
554     // System.out.println("rank : res : total t : fraction p : log2(p) : -p*log2(p)");
555     int rank = 0;
556     boolean seenCentRes = false;
557     lastAUC = 0.0;
558     lastRank = 0;
559     lastFreq = 0.0;
560     lastavgk = 0.0;
561     lastdevk = 0.0;
562     while (rsst.next()) {
563 lappe 243 counter++;
564 lappe 227 res = rsst.getString(1); // 1st column -- res
565 lappe 243 c = rsst.getInt( 2); // 2nds column : count/residue
566 lappe 227 p = rsst.getDouble(3); // 3rd: fraction p
567 lappe 243 if (VL>=2) System.out.print(counter+"/"+rank+ " : " + res+" : "+c+ " : " + p);
568 lappe 227 logp = Math.log(p)/Math.log(2.0); // to basis 2 for info in bits
569     // System.out.print(" : " + logp);
570     plogp = -1.0 * p * logp;
571     // System.out.print(" : " + plogp);
572     plogpsum += plogp;
573     psum += p;
574    
575 lappe 243 if ((c == lastc) && (lastc>0)) { // tie
576     if (VL>=2) System.out.print(" <-- TIE!");
577     lastRank = counter;
578     } // end if
579    
580 lappe 227 if (res.equals(centRes)) {
581 lappe 243 if (VL>=2) System.out.print(" <== " + centRes);
582 lappe 227 seenCentRes = true;
583     lastFreq = p;
584     lastRank = rank;
585     lastavgk = rsst.getDouble(4);
586     lastdevk = rsst.getDouble(5);
587 lappe 243 lastc = c;
588 lappe 227 }
589 lappe 243 if (seenCentRes) lastAUC += p;
590    
591     if (VL>=2) System.out.println("");
592 lappe 227 }
593     // System.out.println("Sum :"+lastTotal+" : "+psum+" : "+plogpsum);
594 lappe 243 if (VL>=2) System.out.println("=> rank "+lastRank);
595 lappe 227 rsst.close();
596     stmt.close();
597     lastEntropy = plogpsum;
598     if (lastRank==0) lastRank = maxRank;
599     } catch (SQLException e) {
600     e.printStackTrace();
601     System.err.println("SQLException: " + e.getMessage());
602     System.err.println("SQLState: " + e.getSQLState());
603     }
604    
605     } // end of getEntropy
606    
607     } // end class