2 |
26 Feb 07 |
jari |
1 |
/* |
2 |
26 Feb 07 |
jari |
Copyright @ 1999-2003, The Institute for Genomic Research (TIGR). |
2 |
26 Feb 07 |
jari |
All rights reserved. |
2 |
26 Feb 07 |
jari |
4 |
*/ |
2 |
26 Feb 07 |
jari |
5 |
/* |
2 |
26 Feb 07 |
jari |
* $RCSfile: JEASEStatistics.java,v $ |
2 |
26 Feb 07 |
jari |
* $Revision: 1.7 $ |
2 |
26 Feb 07 |
jari |
* $Date: 2006/05/02 20:52:48 $ |
2 |
26 Feb 07 |
jari |
* $Author: eleanorahowe $ |
2 |
26 Feb 07 |
jari |
* $State: Exp $ |
2 |
26 Feb 07 |
jari |
11 |
*/ |
2 |
26 Feb 07 |
jari |
12 |
/* |
2 |
26 Feb 07 |
jari |
* Developed by NIAID LTB Lab. |
2 |
26 Feb 07 |
jari |
14 |
*/ |
2 |
26 Feb 07 |
jari |
15 |
/* |
2 |
26 Feb 07 |
jari |
* Modified at TIGR for use within MeV v. 2.2 and later. |
2 |
26 Feb 07 |
jari |
17 |
*/ |
2 |
26 Feb 07 |
jari |
18 |
package org.tigr.microarray.mev.cluster.algorithm.impl.ease; |
2 |
26 Feb 07 |
jari |
19 |
|
2 |
26 Feb 07 |
jari |
20 |
import java.io.BufferedReader; |
2 |
26 Feb 07 |
jari |
21 |
import java.io.File; |
2 |
26 Feb 07 |
jari |
22 |
import java.io.FileReader; |
2 |
26 Feb 07 |
jari |
23 |
import java.io.FileWriter; |
2 |
26 Feb 07 |
jari |
24 |
import java.math.BigDecimal; |
2 |
26 Feb 07 |
jari |
25 |
import java.util.Date; |
2 |
26 Feb 07 |
jari |
26 |
import java.util.Enumeration; |
2 |
26 Feb 07 |
jari |
27 |
import java.util.Hashtable; |
2 |
26 Feb 07 |
jari |
28 |
import java.util.Vector; |
2 |
26 Feb 07 |
jari |
29 |
|
2 |
26 Feb 07 |
jari |
30 |
public class JEASEStatistics { |
2 |
26 Feb 07 |
jari |
31 |
|
2 |
26 Feb 07 |
jari |
/** the name of the population file */ |
2 |
26 Feb 07 |
jari |
33 |
public String population_file_name = ""; |
2 |
26 Feb 07 |
jari |
34 |
|
2 |
26 Feb 07 |
jari |
/** the name of the input file that contains a list of gene identifiers. */ |
2 |
26 Feb 07 |
jari |
36 |
public String list_file_name = ""; |
2 |
26 Feb 07 |
jari |
37 |
|
2 |
26 Feb 07 |
jari |
/** the name of the file which the result will be written to. */ |
2 |
26 Feb 07 |
jari |
39 |
public String output_file_name; |
2 |
26 Feb 07 |
jari |
40 |
|
2 |
26 Feb 07 |
jari |
/** a vector that contains names of the annotation files. */ |
2 |
26 Feb 07 |
jari |
42 |
public Vector annotation_file_names = new Vector(); |
2 |
26 Feb 07 |
jari |
43 |
|
2 |
26 Feb 07 |
jari |
/** the total number of genes that are imported from a file and belong to categories in the annotation files */ |
2 |
26 Feb 07 |
jari |
45 |
public int population_total; |
2 |
26 Feb 07 |
jari |
46 |
|
2 |
26 Feb 07 |
jari |
/** the total number of genes from the sample list that belong to categories in the annotation files. */ |
2 |
26 Feb 07 |
jari |
48 |
public int list_total; |
2 |
26 Feb 07 |
jari |
49 |
|
2 |
26 Feb 07 |
jari |
/** the categories from the annotation files*/ |
2 |
26 Feb 07 |
jari |
51 |
public Hashtable categories = new Hashtable(); |
2 |
26 Feb 07 |
jari |
52 |
|
2 |
26 Feb 07 |
jari |
/** categories and hits in each of the categories for the population */ |
2 |
26 Feb 07 |
jari |
54 |
public Hashtable categories_population = new Hashtable(); |
2 |
26 Feb 07 |
jari |
55 |
|
2 |
26 Feb 07 |
jari |
/** categories and hits in each of the categories for the sample */ |
2 |
26 Feb 07 |
jari |
57 |
public Hashtable categories_list = new Hashtable(); |
2 |
26 Feb 07 |
jari |
58 |
|
2 |
26 Feb 07 |
jari |
/** stores category names and associated locus id hits as java.lang.String in java.util.Vector */ |
2 |
26 Feb 07 |
jari |
60 |
public Hashtable hitAccumulator = new Hashtable(); //jcb |
2 |
26 Feb 07 |
jari |
61 |
|
2 |
26 Feb 07 |
jari |
/** An instance of HypergeometricProbability class. */ |
2 |
26 Feb 07 |
jari |
63 |
public HypergeometricProbability hgp_computation = new HypergeometricProbability(); |
2 |
26 Feb 07 |
jari |
64 |
|
2 |
26 Feb 07 |
jari |
/** A BufferedReader object used to read an external file. */ |
2 |
26 Feb 07 |
jari |
66 |
public BufferedReader read_in_stream; |
2 |
26 Feb 07 |
jari |
67 |
|
2 |
26 Feb 07 |
jari |
/** a two dimension array to store the result */ |
2 |
26 Feb 07 |
jari |
69 |
public String[][] records; |
2 |
26 Feb 07 |
jari |
70 |
|
2 |
26 Feb 07 |
jari |
/** Category names */ //jcb |
2 |
26 Feb 07 |
jari |
72 |
public String [] categoryNames; |
2 |
26 Feb 07 |
jari |
73 |
|
2 |
26 Feb 07 |
jari |
/** Category list hits */ |
2 |
26 Feb 07 |
jari |
75 |
public String [][] listHitMatrix; //jcb |
2 |
26 Feb 07 |
jari |
76 |
|
2 |
26 Feb 07 |
jari |
/** the order in which the results should be listed. */ |
2 |
26 Feb 07 |
jari |
78 |
public String[] orders; |
2 |
26 Feb 07 |
jari |
79 |
|
2 |
26 Feb 07 |
jari |
/** indicates statistic to report */ |
2 |
26 Feb 07 |
jari |
81 |
public boolean reportEaseScore = true; //jcb |
2 |
26 Feb 07 |
jari |
82 |
|
2 |
26 Feb 07 |
jari |
/** accumulates totals for various systems **/ |
2 |
26 Feb 07 |
jari |
84 |
private Hashtable pop_totals = new Hashtable(); |
2 |
26 Feb 07 |
jari |
85 |
|
2 |
26 Feb 07 |
jari |
/** accumulates totals for various systems **/ |
2 |
26 Feb 07 |
jari |
87 |
private Hashtable sample_totals = new Hashtable(); |
2 |
26 Feb 07 |
jari |
88 |
|
2 |
26 Feb 07 |
jari |
/** OS file separator **/ |
2 |
26 Feb 07 |
jari |
90 |
private String sep; |
2 |
26 Feb 07 |
jari |
91 |
|
2 |
26 Feb 07 |
jari |
/** Constructor */ |
2 |
26 Feb 07 |
jari |
93 |
public JEASEStatistics() { |
2 |
26 Feb 07 |
jari |
94 |
population_total = 0; |
2 |
26 Feb 07 |
jari |
95 |
list_total =0; |
2 |
26 Feb 07 |
jari |
96 |
sep = System.getProperty("file.separator"); |
2 |
26 Feb 07 |
jari |
97 |
} |
2 |
26 Feb 07 |
jari |
98 |
|
2 |
26 Feb 07 |
jari |
/** Constructor */ //jcb |
2 |
26 Feb 07 |
jari |
100 |
public JEASEStatistics(boolean reportEaseScore) { |
2 |
26 Feb 07 |
jari |
101 |
this.reportEaseScore = reportEaseScore; |
2 |
26 Feb 07 |
jari |
102 |
population_total = 0; |
2 |
26 Feb 07 |
jari |
103 |
list_total =0; |
2 |
26 Feb 07 |
jari |
104 |
sep = System.getProperty("file.separator"); |
2 |
26 Feb 07 |
jari |
105 |
} |
2 |
26 Feb 07 |
jari |
106 |
|
2 |
26 Feb 07 |
jari |
/** Choose a new population file |
2 |
26 Feb 07 |
jari |
* @param file_name file name |
2 |
26 Feb 07 |
jari |
109 |
*/ |
2 |
26 Feb 07 |
jari |
110 |
public void UseNewPopulationFileName(String file_name) { |
2 |
26 Feb 07 |
jari |
111 |
population_file_name = file_name; |
2 |
26 Feb 07 |
jari |
112 |
} |
2 |
26 Feb 07 |
jari |
113 |
|
2 |
26 Feb 07 |
jari |
/** Remove all of the annotation files from the list currently being used. */ |
2 |
26 Feb 07 |
jari |
115 |
public void RemoveCurrentAnnotationFileNames() { |
2 |
26 Feb 07 |
jari |
116 |
annotation_file_names.removeAllElements(); |
2 |
26 Feb 07 |
jari |
117 |
} |
2 |
26 Feb 07 |
jari |
118 |
|
2 |
26 Feb 07 |
jari |
/** Add a new annotation file name into the annotation file name list. */ |
2 |
26 Feb 07 |
jari |
120 |
public void AddAnnotationFileName(String file_name) { |
2 |
26 Feb 07 |
jari |
121 |
annotation_file_names.addElement(file_name); |
2 |
26 Feb 07 |
jari |
122 |
} |
2 |
26 Feb 07 |
jari |
123 |
|
2 |
26 Feb 07 |
jari |
/** Use a new list file name. */ |
2 |
26 Feb 07 |
jari |
125 |
public void UseNewListFileName(String file_name) { |
2 |
26 Feb 07 |
jari |
126 |
list_file_name = file_name; |
2 |
26 Feb 07 |
jari |
127 |
} |
2 |
26 Feb 07 |
jari |
128 |
|
2 |
26 Feb 07 |
jari |
/** set the name of the file that stores the results.*/ |
2 |
26 Feb 07 |
jari |
130 |
public void SetOutputFileName(String file_name){ |
2 |
26 Feb 07 |
jari |
131 |
output_file_name = file_name; |
2 |
26 Feb 07 |
jari |
132 |
} |
2 |
26 Feb 07 |
jari |
133 |
|
2 |
26 Feb 07 |
jari |
/** Obtain the categories from the annotation files and create a hashtable using these categories as keys.*/ |
2 |
26 Feb 07 |
jari |
135 |
public void GetCategories() { |
2 |
26 Feb 07 |
jari |
136 |
BufferedReader in = null; |
2 |
26 Feb 07 |
jari |
137 |
Hashtable hash_table = new Hashtable(); |
2 |
26 Feb 07 |
jari |
138 |
Hashtable implied_associations = new Hashtable(); |
2 |
26 Feb 07 |
jari |
139 |
String term; |
2 |
26 Feb 07 |
jari |
140 |
String line="", category="", file_name=""; |
2 |
26 Feb 07 |
jari |
141 |
int idx, c =0, idx2; |
2 |
26 Feb 07 |
jari |
142 |
|
2 |
26 Feb 07 |
jari |
143 |
try{ |
2 |
26 Feb 07 |
jari |
144 |
for(Enumeration e = annotation_file_names.elements(); e.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
145 |
file_name = e.nextElement().toString(); |
2 |
26 Feb 07 |
jari |
146 |
in = new BufferedReader(new FileReader(file_name)); |
2 |
26 Feb 07 |
jari |
147 |
|
2 |
26 Feb 07 |
jari |
//use last index of . in case a user has a . in the path. |
2 |
26 Feb 07 |
jari |
149 |
term = file_name.substring(file_name.lastIndexOf(sep)+1, file_name.lastIndexOf(".")); |
2 |
26 Feb 07 |
jari |
150 |
|
2 |
26 Feb 07 |
jari |
//store terms in total hits accumulators |
2 |
26 Feb 07 |
jari |
152 |
this.sample_totals.put(term, new Hashtable()); |
2 |
26 Feb 07 |
jari |
153 |
this.pop_totals.put(term, new Hashtable()); |
2 |
26 Feb 07 |
jari |
154 |
|
2 |
26 Feb 07 |
jari |
155 |
while((line = in.readLine()) != null){ |
2 |
26 Feb 07 |
jari |
156 |
|
2 |
26 Feb 07 |
jari |
157 |
idx = line.indexOf("\t"); |
2 |
26 Feb 07 |
jari |
158 |
|
2 |
26 Feb 07 |
jari |
159 |
if( idx >= line.length() || idx < 1){ |
2 |
26 Feb 07 |
jari |
160 |
continue; |
2 |
26 Feb 07 |
jari |
161 |
} |
2 |
26 Feb 07 |
jari |
162 |
|
2 |
26 Feb 07 |
jari |
//put this in to guard against no trailing tab (jcb) |
2 |
26 Feb 07 |
jari |
164 |
idx2 = line.indexOf("\t", idx+1); |
2 |
26 Feb 07 |
jari |
165 |
if(idx2 >= line.length() || idx2 < 1) |
2 |
26 Feb 07 |
jari |
166 |
idx2 = line.length(); |
2 |
26 Feb 07 |
jari |
167 |
|
2 |
26 Feb 07 |
jari |
168 |
category = term + "\t" + line.substring(idx+1, idx2).trim(); |
2 |
26 Feb 07 |
jari |
169 |
|
2 |
26 Feb 07 |
jari |
170 |
if(!categories.containsKey(category)){ |
2 |
26 Feb 07 |
jari |
171 |
categories.put(category, new Hashtable()); |
2 |
26 Feb 07 |
jari |
172 |
((Hashtable) categories.get(category)).put(line.substring(0,idx).trim(), ""); |
2 |
26 Feb 07 |
jari |
173 |
}else{ |
2 |
26 Feb 07 |
jari |
174 |
((Hashtable) categories.get(category)).put(line.substring(0,idx).trim(), ""); |
2 |
26 Feb 07 |
jari |
175 |
} |
2 |
26 Feb 07 |
jari |
176 |
} |
2 |
26 Feb 07 |
jari |
177 |
} |
2 |
26 Feb 07 |
jari |
178 |
|
2 |
26 Feb 07 |
jari |
//System.out.println("categories size = "+categories.size()); |
2 |
26 Feb 07 |
jari |
180 |
|
2 |
26 Feb 07 |
jari |
//(jcb) |
2 |
26 Feb 07 |
jari |
//create hash table for implies using (implies_associator) |
2 |
26 Feb 07 |
jari |
//This will then be used to add implied categories |
2 |
26 Feb 07 |
jari |
184 |
String fileName, impliesFile; |
2 |
26 Feb 07 |
jari |
185 |
int stringIndex; |
2 |
26 Feb 07 |
jari |
186 |
for(int i = 0; i < annotation_file_names.size(); i++){ |
2 |
26 Feb 07 |
jari |
187 |
fileName = (String)annotation_file_names.elementAt(i); |
2 |
26 Feb 07 |
jari |
188 |
stringIndex = fileName.lastIndexOf(sep); |
2 |
26 Feb 07 |
jari |
189 |
impliesFile = fileName.substring(0, stringIndex) + sep+"Implies"+sep; |
2 |
26 Feb 07 |
jari |
190 |
impliesFile += fileName.substring(stringIndex+1, fileName.length()); |
2 |
26 Feb 07 |
jari |
191 |
|
2 |
26 Feb 07 |
jari |
//System.out.println("implies file = "+impliesFile); |
2 |
26 Feb 07 |
jari |
193 |
|
2 |
26 Feb 07 |
jari |
194 |
File file = new File(impliesFile); |
2 |
26 Feb 07 |
jari |
195 |
if(!file.exists() || !file.isFile()) //if implies file is missing move on |
2 |
26 Feb 07 |
jari |
196 |
continue; |
2 |
26 Feb 07 |
jari |
197 |
|
2 |
26 Feb 07 |
jari |
198 |
in = new BufferedReader(new FileReader(impliesFile)); |
2 |
26 Feb 07 |
jari |
//term = fileName.substring(file_name.lastIndexOf("/")+1, file_name.indexOf(".")); |
2 |
26 Feb 07 |
jari |
200 |
term = fileName.substring(fileName.lastIndexOf(sep)+1, fileName.lastIndexOf(".")); |
2 |
26 Feb 07 |
jari |
201 |
|
2 |
26 Feb 07 |
jari |
202 |
|
2 |
26 Feb 07 |
jari |
203 |
while((line = in.readLine()) != null){ |
2 |
26 Feb 07 |
jari |
204 |
idx = line.indexOf('\t'); |
2 |
26 Feb 07 |
jari |
205 |
|
2 |
26 Feb 07 |
jari |
206 |
if(idx >= line.length() || idx < 1) //must include a tab |
2 |
26 Feb 07 |
jari |
207 |
continue; |
2 |
26 Feb 07 |
jari |
208 |
|
2 |
26 Feb 07 |
jari |
209 |
if(!implied_associations.containsKey(term + "\t" +line.substring(0,idx).trim())){ |
2 |
26 Feb 07 |
jari |
210 |
implied_associations.put(term + "\t" +line.substring(0,idx).trim(), new Vector()); |
2 |
26 Feb 07 |
jari |
211 |
((Vector)(implied_associations.get(term + "\t" +line.substring(0,idx).trim()))).addElement(term + "\t" + line.substring(idx, line.length()).trim()); |
2 |
26 Feb 07 |
jari |
212 |
} else { |
2 |
26 Feb 07 |
jari |
213 |
((Vector)(implied_associations.get(term + "\t" +line.substring(0,idx).trim()))).addElement(term + "\t" + line.substring(idx, line.length()).trim()); |
2 |
26 Feb 07 |
jari |
214 |
} |
2 |
26 Feb 07 |
jari |
215 |
} |
2 |
26 Feb 07 |
jari |
216 |
} |
2 |
26 Feb 07 |
jari |
217 |
|
2 |
26 Feb 07 |
jari |
218 |
boolean end = false; |
2 |
26 Feb 07 |
jari |
//(jcb) append associated categories to the list |
2 |
26 Feb 07 |
jari |
220 |
for(int k = 0; k < 10 && !end; k++){ // !end will short circuit if stable |
2 |
26 Feb 07 |
jari |
221 |
String cat=""; |
2 |
26 Feb 07 |
jari |
222 |
Hashtable catHash; |
2 |
26 Feb 07 |
jari |
223 |
Vector impVector; |
2 |
26 Feb 07 |
jari |
224 |
String impCat; |
2 |
26 Feb 07 |
jari |
225 |
|
2 |
26 Feb 07 |
jari |
226 |
end = true; //start at true until no new indices are inserted |
2 |
26 Feb 07 |
jari |
227 |
|
2 |
26 Feb 07 |
jari |
228 |
for( Enumeration _enum = implied_associations.keys(); _enum.hasMoreElements(); ){ |
2 |
26 Feb 07 |
jari |
229 |
|
2 |
26 Feb 07 |
jari |
230 |
cat = (String) _enum.nextElement(); |
2 |
26 Feb 07 |
jari |
231 |
|
2 |
26 Feb 07 |
jari |
//if the category is represented, add the implied associations if they don't exist |
2 |
26 Feb 07 |
jari |
233 |
if(categories.containsKey(cat)){ |
2 |
26 Feb 07 |
jari |
234 |
catHash = (Hashtable)categories.get(cat); |
2 |
26 Feb 07 |
jari |
235 |
|
2 |
26 Feb 07 |
jari |
//associated categories |
2 |
26 Feb 07 |
jari |
237 |
impVector = ((Vector)implied_associations.get(cat)); |
2 |
26 Feb 07 |
jari |
238 |
|
2 |
26 Feb 07 |
jari |
239 |
for(int i = 0; i < impVector.size(); i++){ |
2 |
26 Feb 07 |
jari |
240 |
|
2 |
26 Feb 07 |
jari |
241 |
impCat = ((String)impVector.elementAt(i)); |
2 |
26 Feb 07 |
jari |
242 |
|
2 |
26 Feb 07 |
jari |
243 |
if(!categories.containsKey(impCat)){ |
2 |
26 Feb 07 |
jari |
244 |
end = false; |
2 |
26 Feb 07 |
jari |
245 |
categories.put(impCat, new Hashtable()); |
2 |
26 Feb 07 |
jari |
246 |
for(Enumeration categoryEnum = catHash.keys(); categoryEnum.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
247 |
((Hashtable)categories.get(impCat)).put((String)categoryEnum.nextElement(), ""); |
2 |
26 Feb 07 |
jari |
248 |
} |
2 |
26 Feb 07 |
jari |
249 |
} else { //category exists, need to append locus link numbers |
2 |
26 Feb 07 |
jari |
250 |
for(Enumeration categoryEnum = catHash.keys(); categoryEnum.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
251 |
String indexString = (String)categoryEnum.nextElement(); |
2 |
26 Feb 07 |
jari |
252 |
if(!((Hashtable)categories.get(impCat)).containsKey(indexString) ){ |
2 |
26 Feb 07 |
jari |
// ((Hashtable)categories.get(impCat)).put((String)categoryEnum.nextElement(), ""); |
2 |
26 Feb 07 |
jari |
254 |
((Hashtable)categories.get(impCat)).put(indexString, ""); |
2 |
26 Feb 07 |
jari |
255 |
|
2 |
26 Feb 07 |
jari |
256 |
end = false; |
2 |
26 Feb 07 |
jari |
257 |
} |
2 |
26 Feb 07 |
jari |
258 |
} |
2 |
26 Feb 07 |
jari |
259 |
} |
2 |
26 Feb 07 |
jari |
260 |
} |
2 |
26 Feb 07 |
jari |
261 |
} |
2 |
26 Feb 07 |
jari |
262 |
} |
2 |
26 Feb 07 |
jari |
263 |
} |
2 |
26 Feb 07 |
jari |
264 |
|
2 |
26 Feb 07 |
jari |
265 |
}catch(Exception error){ |
2 |
26 Feb 07 |
jari |
266 |
System.out.println("Error occured collecting categories"); |
2 |
26 Feb 07 |
jari |
267 |
System.out.println(error.getMessage()); |
2 |
26 Feb 07 |
jari |
268 |
error.printStackTrace(); |
2 |
26 Feb 07 |
jari |
269 |
} |
2 |
26 Feb 07 |
jari |
270 |
} |
2 |
26 Feb 07 |
jari |
271 |
|
2 |
26 Feb 07 |
jari |
272 |
|
2 |
26 Feb 07 |
jari |
/** Obtain the categories from the annotation files and create a hashtable using these categories as keys.*/ |
2 |
26 Feb 07 |
jari |
274 |
public void GetCategories(Vector popVector) { |
2 |
26 Feb 07 |
jari |
275 |
BufferedReader in = null; |
2 |
26 Feb 07 |
jari |
276 |
Hashtable hash_table = new Hashtable(); |
2 |
26 Feb 07 |
jari |
277 |
Hashtable implied_associations = new Hashtable(); |
2 |
26 Feb 07 |
jari |
278 |
String term; |
2 |
26 Feb 07 |
jari |
279 |
String line="", category="", file_name=""; |
2 |
26 Feb 07 |
jari |
280 |
int idx, c =0, idx2; |
2 |
26 Feb 07 |
jari |
281 |
|
2 |
26 Feb 07 |
jari |
282 |
try{ |
2 |
26 Feb 07 |
jari |
283 |
for(Enumeration e = annotation_file_names.elements(); e.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
284 |
file_name = e.nextElement().toString(); |
2 |
26 Feb 07 |
jari |
285 |
in = new BufferedReader(new FileReader(file_name)); |
2 |
26 Feb 07 |
jari |
286 |
|
2 |
26 Feb 07 |
jari |
//use last index of . in case a user has a . in the path. |
2 |
26 Feb 07 |
jari |
288 |
term = file_name.substring(file_name.lastIndexOf(sep)+1, file_name.lastIndexOf(".")); |
2 |
26 Feb 07 |
jari |
289 |
|
2 |
26 Feb 07 |
jari |
//store terms in total hits accumulators |
2 |
26 Feb 07 |
jari |
291 |
this.sample_totals.put(term, new Hashtable()); |
2 |
26 Feb 07 |
jari |
292 |
this.pop_totals.put(term, new Hashtable()); |
2 |
26 Feb 07 |
jari |
293 |
|
2 |
26 Feb 07 |
jari |
294 |
while((line = in.readLine()) != null){ |
2 |
26 Feb 07 |
jari |
295 |
|
2 |
26 Feb 07 |
jari |
296 |
idx = line.indexOf("\t"); |
2 |
26 Feb 07 |
jari |
297 |
|
2 |
26 Feb 07 |
jari |
298 |
if( idx >= line.length() || idx < 1){ |
2 |
26 Feb 07 |
jari |
299 |
continue; |
2 |
26 Feb 07 |
jari |
300 |
} |
2 |
26 Feb 07 |
jari |
301 |
|
2 |
26 Feb 07 |
jari |
//put this in to guard against no trailing tab (jcb) |
2 |
26 Feb 07 |
jari |
303 |
idx2 = line.indexOf("\t", idx+1); |
2 |
26 Feb 07 |
jari |
304 |
if(idx2 >= line.length() || idx2 < 1) |
2 |
26 Feb 07 |
jari |
305 |
idx2 = line.length(); |
2 |
26 Feb 07 |
jari |
306 |
|
2 |
26 Feb 07 |
jari |
307 |
category = term + "\t" + line.substring(idx+1, idx2).trim(); |
2 |
26 Feb 07 |
jari |
308 |
if(popVector.contains(line.substring(0,idx).trim())) { |
2 |
26 Feb 07 |
jari |
309 |
if(!categories.containsKey(category)){ |
2 |
26 Feb 07 |
jari |
310 |
categories.put(category, new Hashtable()); |
2 |
26 Feb 07 |
jari |
311 |
((Hashtable) categories.get(category)).put(line.substring(0,idx).trim(), ""); |
2 |
26 Feb 07 |
jari |
312 |
}else{ |
2 |
26 Feb 07 |
jari |
313 |
((Hashtable) categories.get(category)).put(line.substring(0,idx).trim(), ""); |
2 |
26 Feb 07 |
jari |
314 |
} |
2 |
26 Feb 07 |
jari |
315 |
} |
2 |
26 Feb 07 |
jari |
316 |
} |
2 |
26 Feb 07 |
jari |
317 |
} |
2 |
26 Feb 07 |
jari |
318 |
|
2 |
26 Feb 07 |
jari |
//(jcb) |
2 |
26 Feb 07 |
jari |
//create hash table for implies using (implies_associator) |
2 |
26 Feb 07 |
jari |
//This will then be used to add implied categories |
2 |
26 Feb 07 |
jari |
322 |
String fileName, impliesFile; |
2 |
26 Feb 07 |
jari |
323 |
int stringIndex; |
2 |
26 Feb 07 |
jari |
324 |
for(int i = 0; i < annotation_file_names.size(); i++){ |
2 |
26 Feb 07 |
jari |
325 |
fileName = (String)annotation_file_names.elementAt(i); |
2 |
26 Feb 07 |
jari |
326 |
stringIndex = fileName.lastIndexOf(sep); |
2 |
26 Feb 07 |
jari |
327 |
impliesFile = fileName.substring(0, stringIndex) + sep+"Implies"+sep; |
2 |
26 Feb 07 |
jari |
328 |
impliesFile += fileName.substring(stringIndex+1, fileName.length()); |
2 |
26 Feb 07 |
jari |
329 |
|
2 |
26 Feb 07 |
jari |
330 |
File file = new File(impliesFile); |
2 |
26 Feb 07 |
jari |
331 |
if(!file.exists() || !file.isFile()) //if implies file is missing move on |
2 |
26 Feb 07 |
jari |
332 |
continue; |
2 |
26 Feb 07 |
jari |
333 |
|
2 |
26 Feb 07 |
jari |
334 |
in = new BufferedReader(new FileReader(impliesFile)); |
2 |
26 Feb 07 |
jari |
//term = fileName.substring(file_name.lastIndexOf("/")+1, file_name.indexOf(".")); |
2 |
26 Feb 07 |
jari |
336 |
term = fileName.substring(fileName.lastIndexOf(sep)+1, fileName.lastIndexOf(".")); |
2 |
26 Feb 07 |
jari |
337 |
|
2 |
26 Feb 07 |
jari |
338 |
|
2 |
26 Feb 07 |
jari |
339 |
while((line = in.readLine()) != null){ |
2 |
26 Feb 07 |
jari |
340 |
idx = line.indexOf('\t'); |
2 |
26 Feb 07 |
jari |
341 |
|
2 |
26 Feb 07 |
jari |
342 |
if(idx >= line.length() || idx < 1) //must include a tab |
2 |
26 Feb 07 |
jari |
343 |
continue; |
2 |
26 Feb 07 |
jari |
344 |
|
2 |
26 Feb 07 |
jari |
345 |
if(!implied_associations.containsKey(term + "\t" +line.substring(0,idx).trim())){ |
2 |
26 Feb 07 |
jari |
346 |
implied_associations.put(term + "\t" +line.substring(0,idx).trim(), new Vector()); |
2 |
26 Feb 07 |
jari |
347 |
((Vector)(implied_associations.get(term + "\t" +line.substring(0,idx).trim()))).addElement(term + "\t" + line.substring(idx, line.length()).trim()); |
2 |
26 Feb 07 |
jari |
348 |
} else { |
2 |
26 Feb 07 |
jari |
349 |
((Vector)(implied_associations.get(term + "\t" +line.substring(0,idx).trim()))).addElement(term + "\t" + line.substring(idx, line.length()).trim()); |
2 |
26 Feb 07 |
jari |
350 |
} |
2 |
26 Feb 07 |
jari |
351 |
} |
2 |
26 Feb 07 |
jari |
352 |
} |
2 |
26 Feb 07 |
jari |
353 |
|
2 |
26 Feb 07 |
jari |
354 |
boolean end = false; |
2 |
26 Feb 07 |
jari |
//(jcb) append associated categories to the list |
2 |
26 Feb 07 |
jari |
356 |
for(int k = 0; k < 10 && !end; k++){ // !end will short circuit if stable |
2 |
26 Feb 07 |
jari |
357 |
String cat=""; |
2 |
26 Feb 07 |
jari |
358 |
Hashtable catHash; |
2 |
26 Feb 07 |
jari |
359 |
Vector impVector; |
2 |
26 Feb 07 |
jari |
360 |
String impCat; |
2 |
26 Feb 07 |
jari |
361 |
|
2 |
26 Feb 07 |
jari |
362 |
end = true; //start at true until no new indices are inserted |
2 |
26 Feb 07 |
jari |
363 |
|
2 |
26 Feb 07 |
jari |
364 |
for( Enumeration _enum = implied_associations.keys(); _enum.hasMoreElements(); ){ |
2 |
26 Feb 07 |
jari |
365 |
|
2 |
26 Feb 07 |
jari |
366 |
cat = (String) _enum.nextElement(); |
2 |
26 Feb 07 |
jari |
367 |
|
2 |
26 Feb 07 |
jari |
//if the category is represented, add the implied associations if they don't exist |
2 |
26 Feb 07 |
jari |
369 |
if(categories.containsKey(cat)){ |
2 |
26 Feb 07 |
jari |
370 |
catHash = (Hashtable)categories.get(cat); |
2 |
26 Feb 07 |
jari |
371 |
|
2 |
26 Feb 07 |
jari |
//associated categories |
2 |
26 Feb 07 |
jari |
373 |
impVector = ((Vector)implied_associations.get(cat)); |
2 |
26 Feb 07 |
jari |
374 |
|
2 |
26 Feb 07 |
jari |
375 |
for(int i = 0; i < impVector.size(); i++){ |
2 |
26 Feb 07 |
jari |
376 |
|
2 |
26 Feb 07 |
jari |
377 |
impCat = ((String)impVector.elementAt(i)); |
2 |
26 Feb 07 |
jari |
378 |
|
2 |
26 Feb 07 |
jari |
379 |
if(!categories.containsKey(impCat)){ |
2 |
26 Feb 07 |
jari |
380 |
end = false; |
2 |
26 Feb 07 |
jari |
381 |
categories.put(impCat, new Hashtable()); |
2 |
26 Feb 07 |
jari |
382 |
for(Enumeration categoryEnum = catHash.keys(); categoryEnum.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
383 |
((Hashtable)categories.get(impCat)).put((String)categoryEnum.nextElement(), ""); |
2 |
26 Feb 07 |
jari |
384 |
} |
2 |
26 Feb 07 |
jari |
385 |
} else { //category exists, need to append locus link numbers |
2 |
26 Feb 07 |
jari |
386 |
for(Enumeration categoryEnum = catHash.keys(); categoryEnum.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
387 |
String indexString = (String)categoryEnum.nextElement(); |
2 |
26 Feb 07 |
jari |
388 |
if(!((Hashtable)categories.get(impCat)).containsKey(indexString) ){ |
2 |
26 Feb 07 |
jari |
// ((Hashtable)categories.get(impCat)).put((String)categoryEnum.nextElement(), ""); |
2 |
26 Feb 07 |
jari |
390 |
((Hashtable)categories.get(impCat)).put(indexString, ""); |
2 |
26 Feb 07 |
jari |
391 |
|
2 |
26 Feb 07 |
jari |
392 |
end = false; |
2 |
26 Feb 07 |
jari |
393 |
} |
2 |
26 Feb 07 |
jari |
394 |
} |
2 |
26 Feb 07 |
jari |
395 |
} |
2 |
26 Feb 07 |
jari |
396 |
} |
2 |
26 Feb 07 |
jari |
397 |
} |
2 |
26 Feb 07 |
jari |
398 |
} |
2 |
26 Feb 07 |
jari |
399 |
} |
2 |
26 Feb 07 |
jari |
400 |
|
2 |
26 Feb 07 |
jari |
401 |
}catch(Exception error){ |
2 |
26 Feb 07 |
jari |
402 |
System.out.println("Error occured collecting categories"); |
2 |
26 Feb 07 |
jari |
403 |
System.out.println(error.getMessage()); |
2 |
26 Feb 07 |
jari |
404 |
error.printStackTrace(); |
2 |
26 Feb 07 |
jari |
405 |
} |
2 |
26 Feb 07 |
jari |
406 |
} |
2 |
26 Feb 07 |
jari |
407 |
|
2 |
26 Feb 07 |
jari |
/** Get the number of the genes in the population for each category that exists in the annotation files. */ |
2 |
26 Feb 07 |
jari |
409 |
public void GetPopulationHitsByCategory() { |
2 |
26 Feb 07 |
jari |
410 |
BufferedReader in = null; |
2 |
26 Feb 07 |
jari |
411 |
String line="", category="", hits="", key="", locus_id; |
2 |
26 Feb 07 |
jari |
412 |
Hashtable locus_ids = new Hashtable(); |
2 |
26 Feb 07 |
jari |
413 |
Hashtable hash_table = new Hashtable(); |
2 |
26 Feb 07 |
jari |
414 |
Hashtable count_ids = new Hashtable(); |
2 |
26 Feb 07 |
jari |
415 |
|
2 |
26 Feb 07 |
jari |
416 |
int c = 0; |
2 |
26 Feb 07 |
jari |
417 |
|
2 |
26 Feb 07 |
jari |
418 |
try{ |
2 |
26 Feb 07 |
jari |
419 |
in = new BufferedReader(new FileReader(population_file_name)); |
2 |
26 Feb 07 |
jari |
420 |
|
2 |
26 Feb 07 |
jari |
421 |
while((line = in.readLine()) != null){ |
2 |
26 Feb 07 |
jari |
422 |
locus_ids.put(line.trim(), ""); |
2 |
26 Feb 07 |
jari |
423 |
} |
2 |
26 Feb 07 |
jari |
424 |
|
2 |
26 Feb 07 |
jari |
425 |
for(Enumeration enum1 = locus_ids.keys(); enum1.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
426 |
locus_id = (String)(enum1.nextElement()); |
2 |
26 Feb 07 |
jari |
427 |
|
2 |
26 Feb 07 |
jari |
428 |
for(Enumeration _enum = categories.keys(); _enum.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
429 |
key = (String)(_enum.nextElement()); |
2 |
26 Feb 07 |
jari |
430 |
hash_table = (Hashtable) (categories.get(key)); |
2 |
26 Feb 07 |
jari |
431 |
|
2 |
26 Feb 07 |
jari |
432 |
if(hash_table.containsKey(locus_id)){ |
2 |
26 Feb 07 |
jari |
433 |
if(!categories_population.containsKey(key)){ |
2 |
26 Feb 07 |
jari |
434 |
categories_population.put(key, "1"); |
2 |
26 Feb 07 |
jari |
435 |
}else{ |
2 |
26 Feb 07 |
jari |
436 |
hits = String.valueOf(Integer.parseInt((String) categories_population.get(key)) + 1); |
2 |
26 Feb 07 |
jari |
437 |
categories_population.put(key, hits); |
2 |
26 Feb 07 |
jari |
438 |
} |
2 |
26 Feb 07 |
jari |
439 |
|
2 |
26 Feb 07 |
jari |
440 |
count_ids.put(locus_id, ""); |
2 |
26 Feb 07 |
jari |
441 |
} |
2 |
26 Feb 07 |
jari |
442 |
} |
2 |
26 Feb 07 |
jari |
443 |
} |
2 |
26 Feb 07 |
jari |
444 |
|
2 |
26 Feb 07 |
jari |
445 |
population_total = count_ids.size(); |
2 |
26 Feb 07 |
jari |
446 |
|
2 |
26 Feb 07 |
jari |
447 |
}catch(Exception e){ |
2 |
26 Feb 07 |
jari |
448 |
System.out.println(e.getMessage()); |
2 |
26 Feb 07 |
jari |
449 |
} |
2 |
26 Feb 07 |
jari |
450 |
} |
2 |
26 Feb 07 |
jari |
451 |
|
2 |
26 Feb 07 |
jari |
/** Get the number of the genes in the sample for each category that exists in the annotation files. */ |
2 |
26 Feb 07 |
jari |
453 |
public void GetListHitsByCategory() { |
2 |
26 Feb 07 |
jari |
454 |
BufferedReader in = null; |
2 |
26 Feb 07 |
jari |
455 |
String line="", category="", hits="", key="", locus_id; |
2 |
26 Feb 07 |
jari |
456 |
Hashtable locus_ids = new Hashtable(); |
2 |
26 Feb 07 |
jari |
457 |
Hashtable hash_table = new Hashtable(); |
2 |
26 Feb 07 |
jari |
458 |
Hashtable count_ids = new Hashtable(); |
2 |
26 Feb 07 |
jari |
459 |
|
2 |
26 Feb 07 |
jari |
460 |
int c = 0; |
2 |
26 Feb 07 |
jari |
461 |
|
2 |
26 Feb 07 |
jari |
462 |
try{ |
2 |
26 Feb 07 |
jari |
463 |
in = new BufferedReader(new FileReader(list_file_name)); |
2 |
26 Feb 07 |
jari |
464 |
|
2 |
26 Feb 07 |
jari |
465 |
while((line = in.readLine()) != null){ |
2 |
26 Feb 07 |
jari |
466 |
locus_ids.put(line.trim(), ""); |
2 |
26 Feb 07 |
jari |
467 |
} |
2 |
26 Feb 07 |
jari |
468 |
|
2 |
26 Feb 07 |
jari |
469 |
for(Enumeration enum1 = locus_ids.keys(); enum1.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
470 |
locus_id = (String)(enum1.nextElement()); |
2 |
26 Feb 07 |
jari |
471 |
for(Enumeration _enum = categories.keys(); _enum.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
472 |
key = (String)(_enum.nextElement()); |
2 |
26 Feb 07 |
jari |
473 |
hash_table = (Hashtable) (categories.get(key)); |
2 |
26 Feb 07 |
jari |
474 |
|
2 |
26 Feb 07 |
jari |
475 |
if(hash_table.containsKey(locus_id)){ |
2 |
26 Feb 07 |
jari |
476 |
if(!categories_list.containsKey(key)){ |
2 |
26 Feb 07 |
jari |
477 |
categories_list.put(key, "1"); |
2 |
26 Feb 07 |
jari |
478 |
}else{ |
2 |
26 Feb 07 |
jari |
479 |
hits = String.valueOf(Integer.parseInt((String) (categories_list.get(key))) + 1); |
2 |
26 Feb 07 |
jari |
480 |
categories_list.put(key, hits); |
2 |
26 Feb 07 |
jari |
481 |
} |
2 |
26 Feb 07 |
jari |
482 |
count_ids.put(locus_id, ""); |
2 |
26 Feb 07 |
jari |
483 |
} |
2 |
26 Feb 07 |
jari |
484 |
} |
2 |
26 Feb 07 |
jari |
485 |
} |
2 |
26 Feb 07 |
jari |
486 |
|
2 |
26 Feb 07 |
jari |
487 |
list_total = count_ids.size(); |
2 |
26 Feb 07 |
jari |
488 |
|
2 |
26 Feb 07 |
jari |
489 |
}catch(Exception e){ |
2 |
26 Feb 07 |
jari |
490 |
System.out.println(e.getMessage()); |
2 |
26 Feb 07 |
jari |
491 |
} |
2 |
26 Feb 07 |
jari |
492 |
} |
2 |
26 Feb 07 |
jari |
493 |
|
2 |
26 Feb 07 |
jari |
494 |
|
2 |
26 Feb 07 |
jari |
/** Creates the tab-delimited results containing information on system, category, population hits, population total, sample hits |
2 |
26 Feb 07 |
jari |
* and sample totoal, and the one-tailed hypergeometric probability score of over-representation. |
2 |
26 Feb 07 |
jari |
497 |
*/ |
2 |
26 Feb 07 |
jari |
498 |
public void ConstructResults() { |
2 |
26 Feb 07 |
jari |
499 |
|
2 |
26 Feb 07 |
jari |
500 |
String key="", list_hit="", population_hit="", temp=""; |
2 |
26 Feb 07 |
jari |
501 |
records = new String[categories_list.size()][7]; |
2 |
26 Feb 07 |
jari |
502 |
orders = new String[categories_list.size()]; |
2 |
26 Feb 07 |
jari |
503 |
|
2 |
26 Feb 07 |
jari |
504 |
int c=0; |
2 |
26 Feb 07 |
jari |
505 |
|
2 |
26 Feb 07 |
jari |
506 |
for(Enumeration e = categories_list.keys(); e.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
507 |
key = (String)(e.nextElement()); |
2 |
26 Feb 07 |
jari |
508 |
list_hit=""; |
2 |
26 Feb 07 |
jari |
509 |
population_hit=""; |
2 |
26 Feb 07 |
jari |
510 |
|
2 |
26 Feb 07 |
jari |
511 |
temp = key.substring(0, key.indexOf('\t')).trim(); |
2 |
26 Feb 07 |
jari |
512 |
records[c][0] = temp.substring(temp.lastIndexOf(sep)+1).trim(); |
2 |
26 Feb 07 |
jari |
513 |
records[c][1] = key.substring(key.indexOf('\t')).trim(); |
2 |
26 Feb 07 |
jari |
514 |
|
2 |
26 Feb 07 |
jari |
515 |
list_hit = (String)(categories_list.get(key)); |
2 |
26 Feb 07 |
jari |
516 |
|
2 |
26 Feb 07 |
jari |
517 |
records[c][2] = list_hit; |
2 |
26 Feb 07 |
jari |
518 |
|
2 |
26 Feb 07 |
jari |
519 |
list_total = ((Hashtable)(sample_totals.get(temp))).size(); |
2 |
26 Feb 07 |
jari |
520 |
|
2 |
26 Feb 07 |
jari |
521 |
records[c][3] = String.valueOf(list_total); |
2 |
26 Feb 07 |
jari |
522 |
|
2 |
26 Feb 07 |
jari |
523 |
if(categories_population.containsKey(key)){ |
2 |
26 Feb 07 |
jari |
524 |
population_hit = (String)(categories_population.get(key)); |
2 |
26 Feb 07 |
jari |
525 |
records[c][4] = population_hit; |
2 |
26 Feb 07 |
jari |
526 |
}else{ |
2 |
26 Feb 07 |
jari |
527 |
records[c][4] = ""; |
2 |
26 Feb 07 |
jari |
528 |
} |
2 |
26 Feb 07 |
jari |
529 |
|
2 |
26 Feb 07 |
jari |
530 |
population_total = ((Hashtable)(pop_totals.get(temp))).size(); |
2 |
26 Feb 07 |
jari |
531 |
|
2 |
26 Feb 07 |
jari |
532 |
records[c][5] = String.valueOf(population_total); |
2 |
26 Feb 07 |
jari |
533 |
|
2 |
26 Feb 07 |
jari |
534 |
if(!list_hit.equals("") && !population_hit.equals("")){ |
2 |
26 Feb 07 |
jari |
535 |
double p; |
2 |
26 Feb 07 |
jari |
536 |
if(reportEaseScore){ //report ease score (jcb) added option of EASEScore |
2 |
26 Feb 07 |
jari |
537 |
if(Integer.parseInt(list_hit) > 1){ |
2 |
26 Feb 07 |
jari |
538 |
p = hgp_computation.SumHGP(population_total, Integer.parseInt(population_hit), list_total-1, Integer.parseInt(list_hit)-1); |
2 |
26 Feb 07 |
jari |
539 |
} else{ |
2 |
26 Feb 07 |
jari |
540 |
p = 1.0; |
2 |
26 Feb 07 |
jari |
541 |
} |
2 |
26 Feb 07 |
jari |
542 |
} else { //else report Fisher Exact Prob. |
2 |
26 Feb 07 |
jari |
543 |
p = hgp_computation.SumHGP(population_total, Integer.parseInt(population_hit), list_total, Integer.parseInt(list_hit)); |
2 |
26 Feb 07 |
jari |
544 |
} |
2 |
26 Feb 07 |
jari |
545 |
records[c][6] = String.valueOf(p); |
2 |
26 Feb 07 |
jari |
546 |
}else{ |
2 |
26 Feb 07 |
jari |
547 |
records[c][6] = "1.0"; |
2 |
26 Feb 07 |
jari |
548 |
} |
2 |
26 Feb 07 |
jari |
549 |
c++; |
2 |
26 Feb 07 |
jari |
550 |
} |
2 |
26 Feb 07 |
jari |
551 |
|
2 |
26 Feb 07 |
jari |
//jcb |
2 |
26 Feb 07 |
jari |
//accumulates hits for the result matrix and the category names |
2 |
26 Feb 07 |
jari |
554 |
categoryNames = new String[hitAccumulator.size()]; |
2 |
26 Feb 07 |
jari |
555 |
listHitMatrix = new String[categoryNames.length][]; |
2 |
26 Feb 07 |
jari |
556 |
|
2 |
26 Feb 07 |
jari |
557 |
int cnt = 0; |
2 |
26 Feb 07 |
jari |
558 |
Vector locusVector; |
2 |
26 Feb 07 |
jari |
559 |
|
2 |
26 Feb 07 |
jari |
560 |
for(Enumeration e = hitAccumulator.keys(); e.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
561 |
String cat = (String)(e.nextElement()); |
2 |
26 Feb 07 |
jari |
562 |
categoryNames[cnt] = cat; |
2 |
26 Feb 07 |
jari |
563 |
locusVector = ((Vector)hitAccumulator.get(cat)); |
2 |
26 Feb 07 |
jari |
564 |
listHitMatrix[cnt] = new String[locusVector.size()]; |
2 |
26 Feb 07 |
jari |
565 |
for(int i = 0; i < listHitMatrix[cnt].length; i++){ |
2 |
26 Feb 07 |
jari |
566 |
listHitMatrix[cnt][i] = ((String)locusVector.elementAt(i)); |
2 |
26 Feb 07 |
jari |
567 |
} |
2 |
26 Feb 07 |
jari |
568 |
cnt++; |
2 |
26 Feb 07 |
jari |
569 |
} |
2 |
26 Feb 07 |
jari |
570 |
} |
2 |
26 Feb 07 |
jari |
571 |
|
2 |
26 Feb 07 |
jari |
572 |
|
2 |
26 Feb 07 |
jari |
573 |
|
2 |
26 Feb 07 |
jari |
/** Sort the results in the ascending order of p-score and write the tab-delimited result to a file and standard IO output*/ |
2 |
26 Feb 07 |
jari |
575 |
public void SortRecords() { |
2 |
26 Feb 07 |
jari |
576 |
|
2 |
26 Feb 07 |
jari |
577 |
BigDecimal big1, big2; |
2 |
26 Feb 07 |
jari |
578 |
String holder; |
2 |
26 Feb 07 |
jari |
579 |
|
2 |
26 Feb 07 |
jari |
580 |
int len = records.length; |
2 |
26 Feb 07 |
jari |
581 |
|
2 |
26 Feb 07 |
jari |
582 |
for(int i0=0; i0 < len; i0++){ |
2 |
26 Feb 07 |
jari |
583 |
for(int i=0; i < len-1 - i0; i++){ |
2 |
26 Feb 07 |
jari |
584 |
big1 = new BigDecimal(records[i][6]); |
2 |
26 Feb 07 |
jari |
585 |
big2 = new BigDecimal(records[i+1][6]); |
2 |
26 Feb 07 |
jari |
586 |
|
2 |
26 Feb 07 |
jari |
587 |
if(big1.compareTo(big2) > 0){ |
2 |
26 Feb 07 |
jari |
588 |
holder = records[i][0]; |
2 |
26 Feb 07 |
jari |
589 |
records[i][0] = records[i+1][0]; |
2 |
26 Feb 07 |
jari |
590 |
records[i+1][0] = holder; |
2 |
26 Feb 07 |
jari |
591 |
|
2 |
26 Feb 07 |
jari |
592 |
holder = records[i][1]; |
2 |
26 Feb 07 |
jari |
593 |
records[i][1] = records[i+1][1]; |
2 |
26 Feb 07 |
jari |
594 |
records[i+1][1] = holder; |
2 |
26 Feb 07 |
jari |
595 |
|
2 |
26 Feb 07 |
jari |
596 |
holder = records[i][2]; |
2 |
26 Feb 07 |
jari |
597 |
records[i][2] = records[i+1][2]; |
2 |
26 Feb 07 |
jari |
598 |
records[i+1][2] = holder; |
2 |
26 Feb 07 |
jari |
599 |
|
2 |
26 Feb 07 |
jari |
600 |
holder = records[i][3]; |
2 |
26 Feb 07 |
jari |
601 |
records[i][3] = records[i+1][3]; |
2 |
26 Feb 07 |
jari |
602 |
records[i+1][3] = holder; |
2 |
26 Feb 07 |
jari |
603 |
|
2 |
26 Feb 07 |
jari |
604 |
holder = records[i][4]; |
2 |
26 Feb 07 |
jari |
605 |
records[i][4] = records[i+1][4]; |
2 |
26 Feb 07 |
jari |
606 |
records[i+1][4] = holder; |
2 |
26 Feb 07 |
jari |
607 |
|
2 |
26 Feb 07 |
jari |
608 |
holder = records[i][5]; |
2 |
26 Feb 07 |
jari |
609 |
records[i][5] = records[i+1][5]; |
2 |
26 Feb 07 |
jari |
610 |
records[i+1][5] = holder; |
2 |
26 Feb 07 |
jari |
611 |
|
2 |
26 Feb 07 |
jari |
612 |
holder = records[i][6]; |
2 |
26 Feb 07 |
jari |
613 |
records[i][6] = records[i+1][6]; |
2 |
26 Feb 07 |
jari |
614 |
records[i+1][6] = holder; |
2 |
26 Feb 07 |
jari |
615 |
} |
2 |
26 Feb 07 |
jari |
616 |
} |
2 |
26 Feb 07 |
jari |
617 |
} |
2 |
26 Feb 07 |
jari |
//jcb, elected not to output file by default here. |
2 |
26 Feb 07 |
jari |
//DisplayResults(); |
2 |
26 Feb 07 |
jari |
620 |
} |
2 |
26 Feb 07 |
jari |
621 |
|
2 |
26 Feb 07 |
jari |
/** Writes the tab-delimited results to the specified output file and to the standout IO output as well */ |
2 |
26 Feb 07 |
jari |
623 |
public void DisplayResults() { |
2 |
26 Feb 07 |
jari |
624 |
try{ |
2 |
26 Feb 07 |
jari |
625 |
File output_file = new File(output_file_name); |
2 |
26 Feb 07 |
jari |
626 |
FileWriter output_writer = new FileWriter(output_file); //, false); |
2 |
26 Feb 07 |
jari |
627 |
|
2 |
26 Feb 07 |
jari |
628 |
for(int i=0; i < records.length; i++){ |
2 |
26 Feb 07 |
jari |
629 |
System.out.print(records[i][0] + "\t"); |
2 |
26 Feb 07 |
jari |
630 |
output_writer.write(records[i][0] + "\t"); |
2 |
26 Feb 07 |
jari |
631 |
|
2 |
26 Feb 07 |
jari |
632 |
System.out.print(records[i][1] + "\t"); |
2 |
26 Feb 07 |
jari |
633 |
output_writer.write(records[i][1] + "\t"); |
2 |
26 Feb 07 |
jari |
634 |
|
2 |
26 Feb 07 |
jari |
635 |
System.out.print(records[i][2] + "\t"); |
2 |
26 Feb 07 |
jari |
636 |
output_writer.write(records[i][2] + "\t"); |
2 |
26 Feb 07 |
jari |
637 |
|
2 |
26 Feb 07 |
jari |
638 |
System.out.print(records[i][3] + "\t"); |
2 |
26 Feb 07 |
jari |
639 |
output_writer.write(records[i][3] + "\t"); |
2 |
26 Feb 07 |
jari |
640 |
|
2 |
26 Feb 07 |
jari |
641 |
System.out.print(records[i][4] + "\t"); |
2 |
26 Feb 07 |
jari |
642 |
output_writer.write(records[i][4] + "\t"); |
2 |
26 Feb 07 |
jari |
643 |
|
2 |
26 Feb 07 |
jari |
644 |
System.out.println(records[i][5] + "\t"); |
2 |
26 Feb 07 |
jari |
645 |
output_writer.write(records[i][5] + "\t"); |
2 |
26 Feb 07 |
jari |
646 |
|
2 |
26 Feb 07 |
jari |
647 |
output_writer.write("\n"); |
2 |
26 Feb 07 |
jari |
648 |
} |
2 |
26 Feb 07 |
jari |
649 |
output_writer.close(); |
2 |
26 Feb 07 |
jari |
650 |
}catch(Exception e){ |
2 |
26 Feb 07 |
jari |
651 |
System.out.println(e.getMessage()); |
2 |
26 Feb 07 |
jari |
652 |
} |
2 |
26 Feb 07 |
jari |
653 |
} |
2 |
26 Feb 07 |
jari |
654 |
|
2 |
26 Feb 07 |
jari |
655 |
public static void main(String[] args) { |
2 |
26 Feb 07 |
jari |
656 |
|
2 |
26 Feb 07 |
jari |
657 |
JEASEStatistics jstatistics = new JEASEStatistics(); |
2 |
26 Feb 07 |
jari |
658 |
System.out.println(new Date(System.currentTimeMillis())); |
2 |
26 Feb 07 |
jari |
659 |
|
2 |
26 Feb 07 |
jari |
660 |
jstatistics.UseNewPopulationFileName("C:/01_Work_Test_4/Files/JStatistics/All Homo sapiens.txt"); |
2 |
26 Feb 07 |
jari |
661 |
jstatistics.RemoveCurrentAnnotationFileNames(); |
2 |
26 Feb 07 |
jari |
662 |
jstatistics.AddAnnotationFileName("C:/01_Work_Test_4/Files/JStatistics/GO Biological Process.txt"); |
2 |
26 Feb 07 |
jari |
663 |
jstatistics.UseNewListFileName("C:/01_Work_Test_4/Files/JStatistics/Bigger_HepC_chimp.txt"); |
2 |
26 Feb 07 |
jari |
664 |
jstatistics.SetOutputFileName("C:/del.txt"); |
2 |
26 Feb 07 |
jari |
665 |
|
2 |
26 Feb 07 |
jari |
666 |
jstatistics.GetCategories(); |
2 |
26 Feb 07 |
jari |
667 |
jstatistics.GetListHitsByCategory(); |
2 |
26 Feb 07 |
jari |
668 |
jstatistics.GetPopulationHitsByCategory(); |
2 |
26 Feb 07 |
jari |
669 |
jstatistics.ConstructResults(); |
2 |
26 Feb 07 |
jari |
670 |
jstatistics.SortRecords(); |
2 |
26 Feb 07 |
jari |
671 |
|
2 |
26 Feb 07 |
jari |
672 |
System.out.println(new Date(System.currentTimeMillis())); |
2 |
26 Feb 07 |
jari |
673 |
System.out.println("\nCompleted."); |
2 |
26 Feb 07 |
jari |
674 |
} |
2 |
26 Feb 07 |
jari |
675 |
|
2 |
26 Feb 07 |
jari |
676 |
|
2 |
26 Feb 07 |
jari |
677 |
/*** |
2 |
26 Feb 07 |
jari |
* (jcb)Modified Hit collection methods to accept a vector argument of annotation keys |
2 |
26 Feb 07 |
jari |
* Modification of code to enable integration into MeV Ease module. |
2 |
26 Feb 07 |
jari |
680 |
*/ |
2 |
26 Feb 07 |
jari |
681 |
|
2 |
26 Feb 07 |
jari |
/** Get the number of the genes in the population for each category that exists in the annotation files. |
2 |
26 Feb 07 |
jari |
* @param list population list. |
2 |
26 Feb 07 |
jari |
684 |
*/ |
2 |
26 Feb 07 |
jari |
685 |
public void GetPopulationHitsByCategory(Vector list) { |
2 |
26 Feb 07 |
jari |
686 |
BufferedReader in = null; |
2 |
26 Feb 07 |
jari |
687 |
String category="", hits="", key="", locus_id; |
2 |
26 Feb 07 |
jari |
688 |
Hashtable locus_ids = new Hashtable(); |
2 |
26 Feb 07 |
jari |
689 |
Hashtable hash_table = new Hashtable(); |
2 |
26 Feb 07 |
jari |
690 |
Hashtable count_ids = new Hashtable(); |
2 |
26 Feb 07 |
jari |
691 |
|
2 |
26 Feb 07 |
jari |
692 |
try{ |
2 |
26 Feb 07 |
jari |
693 |
int size = list.size(); |
2 |
26 Feb 07 |
jari |
694 |
for(int i = 0; i < size; i++) |
2 |
26 Feb 07 |
jari |
695 |
locus_ids.put((String)list.elementAt(i), ""); |
2 |
26 Feb 07 |
jari |
696 |
|
2 |
26 Feb 07 |
jari |
697 |
for(Enumeration enum1 = locus_ids.keys(); enum1.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
698 |
locus_id = (String)(enum1.nextElement()); |
2 |
26 Feb 07 |
jari |
699 |
|
2 |
26 Feb 07 |
jari |
700 |
for(Enumeration _enum = categories.keys(); _enum.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
701 |
key = (String)(_enum.nextElement()); |
2 |
26 Feb 07 |
jari |
702 |
hash_table = (Hashtable) (categories.get(key)); |
2 |
26 Feb 07 |
jari |
703 |
|
2 |
26 Feb 07 |
jari |
704 |
if(hash_table.containsKey(locus_id)){ |
2 |
26 Feb 07 |
jari |
705 |
if(!categories_population.containsKey(key)){ |
2 |
26 Feb 07 |
jari |
706 |
categories_population.put(key, "1"); |
2 |
26 Feb 07 |
jari |
707 |
}else{ |
2 |
26 Feb 07 |
jari |
708 |
hits = String.valueOf(Integer.parseInt((String) categories_population.get(key)) + 1); |
2 |
26 Feb 07 |
jari |
709 |
categories_population.put(key, hits); |
2 |
26 Feb 07 |
jari |
710 |
} |
2 |
26 Feb 07 |
jari |
711 |
count_ids.put(locus_id, ""); |
2 |
26 Feb 07 |
jari |
712 |
|
2 |
26 Feb 07 |
jari |
//accumulate the pop total hits |
2 |
26 Feb 07 |
jari |
714 |
key = key.substring(0, key.indexOf("\t")).trim(); |
2 |
26 Feb 07 |
jari |
715 |
((Hashtable)pop_totals.get(key)).put(locus_id,""); |
2 |
26 Feb 07 |
jari |
716 |
} |
2 |
26 Feb 07 |
jari |
717 |
} |
2 |
26 Feb 07 |
jari |
718 |
} |
2 |
26 Feb 07 |
jari |
719 |
population_total = count_ids.size(); |
2 |
26 Feb 07 |
jari |
720 |
}catch(Exception e){ |
2 |
26 Feb 07 |
jari |
721 |
System.out.println(e.getMessage()); |
2 |
26 Feb 07 |
jari |
722 |
} |
2 |
26 Feb 07 |
jari |
723 |
} |
2 |
26 Feb 07 |
jari |
724 |
|
2 |
26 Feb 07 |
jari |
/** Get the number of the genes in the sample for each category that exists in the annotation files. |
2 |
26 Feb 07 |
jari |
* @param list List indices. |
2 |
26 Feb 07 |
jari |
727 |
*/ |
2 |
26 Feb 07 |
jari |
728 |
public void GetListHitsByCategory(Vector list) { |
2 |
26 Feb 07 |
jari |
729 |
BufferedReader in = null; |
2 |
26 Feb 07 |
jari |
730 |
String category="", hits="", key="", locus_id; |
2 |
26 Feb 07 |
jari |
731 |
Hashtable locus_ids = new Hashtable(); |
2 |
26 Feb 07 |
jari |
732 |
Hashtable hash_table = new Hashtable(); |
2 |
26 Feb 07 |
jari |
733 |
Hashtable count_ids = new Hashtable(); |
2 |
26 Feb 07 |
jari |
734 |
|
2 |
26 Feb 07 |
jari |
735 |
int c = 0; |
2 |
26 Feb 07 |
jari |
736 |
try{ |
2 |
26 Feb 07 |
jari |
737 |
int size = list.size(); |
2 |
26 Feb 07 |
jari |
738 |
for(int i = 0; i < size; i++) |
2 |
26 Feb 07 |
jari |
739 |
locus_ids.put((String)list.elementAt(i), ""); |
2 |
26 Feb 07 |
jari |
740 |
|
2 |
26 Feb 07 |
jari |
741 |
for(Enumeration enum1 = locus_ids.keys(); enum1.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
742 |
locus_id = (String)(enum1.nextElement()); |
2 |
26 Feb 07 |
jari |
743 |
|
2 |
26 Feb 07 |
jari |
744 |
for(Enumeration _enum = categories.keys(); _enum.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
745 |
key = (String)(_enum.nextElement()); |
2 |
26 Feb 07 |
jari |
746 |
hash_table = (Hashtable) (categories.get(key)); |
2 |
26 Feb 07 |
jari |
747 |
|
2 |
26 Feb 07 |
jari |
748 |
if(hash_table.containsKey(locus_id)){ |
2 |
26 Feb 07 |
jari |
749 |
if(!categories_list.containsKey(key)){ |
2 |
26 Feb 07 |
jari |
750 |
categories_list.put(key, "1"); |
2 |
26 Feb 07 |
jari |
//jcb |
2 |
26 Feb 07 |
jari |
752 |
hitAccumulator.put(key, new Vector()); |
2 |
26 Feb 07 |
jari |
753 |
((Vector)hitAccumulator.get(key)).add(locus_id); |
2 |
26 Feb 07 |
jari |
//end mod |
2 |
26 Feb 07 |
jari |
755 |
}else { |
2 |
26 Feb 07 |
jari |
756 |
hits = String.valueOf(Integer.parseInt((String) (categories_list.get(key))) + 1); |
2 |
26 Feb 07 |
jari |
757 |
categories_list.put(key, hits); |
2 |
26 Feb 07 |
jari |
//jcb |
2 |
26 Feb 07 |
jari |
759 |
((Vector)hitAccumulator.get(key)).add(locus_id); |
2 |
26 Feb 07 |
jari |
//end mod |
2 |
26 Feb 07 |
jari |
761 |
} |
2 |
26 Feb 07 |
jari |
762 |
count_ids.put(locus_id, ""); |
2 |
26 Feb 07 |
jari |
763 |
|
2 |
26 Feb 07 |
jari |
//accumulate the sample total hits |
2 |
26 Feb 07 |
jari |
765 |
key = key.substring(0, key.indexOf("\t")).trim(); |
2 |
26 Feb 07 |
jari |
766 |
((Hashtable)sample_totals.get(key)).put(locus_id,""); |
2 |
26 Feb 07 |
jari |
767 |
} |
2 |
26 Feb 07 |
jari |
768 |
} |
2 |
26 Feb 07 |
jari |
769 |
} |
2 |
26 Feb 07 |
jari |
770 |
list_total = count_ids.size(); |
2 |
26 Feb 07 |
jari |
771 |
}catch(Exception e){ |
2 |
26 Feb 07 |
jari |
772 |
System.out.println(e.getMessage()); |
2 |
26 Feb 07 |
jari |
773 |
} |
2 |
26 Feb 07 |
jari |
774 |
} |
2 |
26 Feb 07 |
jari |
775 |
|
2 |
26 Feb 07 |
jari |
776 |
String [][] getResults(){ |
2 |
26 Feb 07 |
jari |
777 |
return this.records; |
2 |
26 Feb 07 |
jari |
778 |
} |
2 |
26 Feb 07 |
jari |
779 |
|
2 |
26 Feb 07 |
jari |
780 |
String [][] getSurveyResults(){ |
2 |
26 Feb 07 |
jari |
781 |
return this.records; |
2 |
26 Feb 07 |
jari |
782 |
} |
2 |
26 Feb 07 |
jari |
783 |
|
2 |
26 Feb 07 |
jari |
784 |
String [][] getListHitMatrix(){ |
2 |
26 Feb 07 |
jari |
785 |
return this.listHitMatrix; |
2 |
26 Feb 07 |
jari |
786 |
} |
2 |
26 Feb 07 |
jari |
787 |
|
2 |
26 Feb 07 |
jari |
788 |
String [] getCategoryNames(){ |
2 |
26 Feb 07 |
jari |
789 |
return this.categoryNames; |
2 |
26 Feb 07 |
jari |
790 |
} |
2 |
26 Feb 07 |
jari |
791 |
|
2 |
26 Feb 07 |
jari |
/** Resets the list total and list hit accumulators. |
2 |
26 Feb 07 |
jari |
793 |
*/ |
2 |
26 Feb 07 |
jari |
794 |
public void resetForNewList(){ |
2 |
26 Feb 07 |
jari |
795 |
this.categories_list = new Hashtable(); |
2 |
26 Feb 07 |
jari |
796 |
this.sample_totals = new Hashtable(); |
2 |
26 Feb 07 |
jari |
797 |
String file_name, term; |
2 |
26 Feb 07 |
jari |
798 |
|
2 |
26 Feb 07 |
jari |
799 |
for(Enumeration e = annotation_file_names.elements(); e.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
800 |
file_name = e.nextElement().toString(); |
2 |
26 Feb 07 |
jari |
801 |
term = file_name.substring(file_name.lastIndexOf(sep)+1, file_name.lastIndexOf(".")); |
2 |
26 Feb 07 |
jari |
802 |
|
2 |
26 Feb 07 |
jari |
//store terms in total hits accumulators |
2 |
26 Feb 07 |
jari |
804 |
this.sample_totals.put(term, new Hashtable()); |
2 |
26 Feb 07 |
jari |
805 |
} |
2 |
26 Feb 07 |
jari |
806 |
} |
2 |
26 Feb 07 |
jari |
807 |
|
2 |
26 Feb 07 |
jari |
808 |
|
2 |
26 Feb 07 |
jari |
809 |
/*** |
2 |
26 Feb 07 |
jari |
* Modified Hit collection methods to accept a vector argument of annotation keys |
2 |
26 Feb 07 |
jari |
* Modification of code to enable integration into TIGR MeV EASE module. |
2 |
26 Feb 07 |
jari |
812 |
*/ |
2 |
26 Feb 07 |
jari |
813 |
|
2 |
26 Feb 07 |
jari |
/** Get the number of the genes in the population for each category that exists in the annotation files. |
2 |
26 Feb 07 |
jari |
* @param list Input population list. |
2 |
26 Feb 07 |
jari |
816 |
*/ |
2 |
26 Feb 07 |
jari |
817 |
public void GetPopulationHitsByCategoryForSurvey(Vector list) { |
2 |
26 Feb 07 |
jari |
818 |
BufferedReader in = null; |
2 |
26 Feb 07 |
jari |
819 |
String category="", hits="", key="", locus_id; |
2 |
26 Feb 07 |
jari |
820 |
Hashtable locus_ids = new Hashtable(); |
2 |
26 Feb 07 |
jari |
821 |
Hashtable hash_table = new Hashtable(); |
2 |
26 Feb 07 |
jari |
822 |
Hashtable count_ids = new Hashtable(); |
2 |
26 Feb 07 |
jari |
//jcb |
2 |
26 Feb 07 |
jari |
824 |
String impliedCategory=""; |
2 |
26 Feb 07 |
jari |
825 |
Vector impliedVector; |
2 |
26 Feb 07 |
jari |
826 |
|
2 |
26 Feb 07 |
jari |
827 |
int c = 0; |
2 |
26 Feb 07 |
jari |
828 |
|
2 |
26 Feb 07 |
jari |
829 |
try{ |
2 |
26 Feb 07 |
jari |
830 |
int size = list.size(); |
2 |
26 Feb 07 |
jari |
831 |
for(int i = 0; i < size; i++) |
2 |
26 Feb 07 |
jari |
832 |
locus_ids.put((String)list.elementAt(i), ""); |
2 |
26 Feb 07 |
jari |
833 |
|
2 |
26 Feb 07 |
jari |
834 |
for(Enumeration enum1 = locus_ids.keys(); enum1.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
835 |
locus_id = (String)(enum1.nextElement()); |
2 |
26 Feb 07 |
jari |
836 |
|
2 |
26 Feb 07 |
jari |
837 |
for(Enumeration _enum = categories.keys(); _enum.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
838 |
key = (String)(_enum.nextElement()); |
2 |
26 Feb 07 |
jari |
839 |
hash_table = (Hashtable) (categories.get(key)); |
2 |
26 Feb 07 |
jari |
840 |
|
2 |
26 Feb 07 |
jari |
841 |
if(hash_table.containsKey(locus_id)){ |
2 |
26 Feb 07 |
jari |
842 |
if(!categories_population.containsKey(key)){ |
2 |
26 Feb 07 |
jari |
843 |
categories_population.put(key, "1"); |
2 |
26 Feb 07 |
jari |
844 |
|
2 |
26 Feb 07 |
jari |
845 |
hitAccumulator.put(key, new Vector()); |
2 |
26 Feb 07 |
jari |
846 |
((Vector)hitAccumulator.get(key)).add(locus_id); |
2 |
26 Feb 07 |
jari |
847 |
}else{ |
2 |
26 Feb 07 |
jari |
848 |
hits = String.valueOf(Integer.parseInt((String) categories_population.get(key)) + 1); |
2 |
26 Feb 07 |
jari |
849 |
categories_population.put(key, hits); |
2 |
26 Feb 07 |
jari |
850 |
|
2 |
26 Feb 07 |
jari |
851 |
((Vector)hitAccumulator.get(key)).add(locus_id); |
2 |
26 Feb 07 |
jari |
852 |
} |
2 |
26 Feb 07 |
jari |
853 |
count_ids.put(locus_id, ""); |
2 |
26 Feb 07 |
jari |
854 |
|
2 |
26 Feb 07 |
jari |
//accumulate the pop total hits |
2 |
26 Feb 07 |
jari |
856 |
key = key.substring(0, key.indexOf("\t")).trim(); |
2 |
26 Feb 07 |
jari |
857 |
((Hashtable)pop_totals.get(key)).put(locus_id,""); |
2 |
26 Feb 07 |
jari |
858 |
} |
2 |
26 Feb 07 |
jari |
859 |
} |
2 |
26 Feb 07 |
jari |
860 |
} |
2 |
26 Feb 07 |
jari |
861 |
|
2 |
26 Feb 07 |
jari |
862 |
population_total = count_ids.size(); |
2 |
26 Feb 07 |
jari |
863 |
|
2 |
26 Feb 07 |
jari |
864 |
}catch(Exception e){ |
2 |
26 Feb 07 |
jari |
865 |
System.out.println(e.getMessage()); |
2 |
26 Feb 07 |
jari |
866 |
} |
2 |
26 Feb 07 |
jari |
867 |
} |
2 |
26 Feb 07 |
jari |
868 |
|
2 |
26 Feb 07 |
jari |
869 |
|
2 |
26 Feb 07 |
jari |
/** Creates the tab-delimited results containing information on system, category, population hits, population total, sample hits |
2 |
26 Feb 07 |
jari |
* and sample totoal, and the one-tailed hypergeometric probability score of over-representation. |
2 |
26 Feb 07 |
jari |
872 |
*/ |
2 |
26 Feb 07 |
jari |
873 |
public void ConstructSurveyResults() { |
2 |
26 Feb 07 |
jari |
874 |
|
2 |
26 Feb 07 |
jari |
875 |
String key="", list_hit="", population_hit="", temp=""; |
2 |
26 Feb 07 |
jari |
876 |
|
2 |
26 Feb 07 |
jari |
877 |
records = new String[categories_population.size()][4]; |
2 |
26 Feb 07 |
jari |
878 |
orders = new String[categories_population.size()]; |
2 |
26 Feb 07 |
jari |
879 |
|
2 |
26 Feb 07 |
jari |
880 |
int c=0; |
2 |
26 Feb 07 |
jari |
881 |
|
2 |
26 Feb 07 |
jari |
882 |
for(Enumeration e = categories_population.keys(); e.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
883 |
key = (String)(e.nextElement()); |
2 |
26 Feb 07 |
jari |
884 |
list_hit=""; |
2 |
26 Feb 07 |
jari |
885 |
population_hit=""; |
2 |
26 Feb 07 |
jari |
886 |
|
2 |
26 Feb 07 |
jari |
887 |
temp = key.substring(0, key.indexOf('\t')).trim(); |
2 |
26 Feb 07 |
jari |
888 |
records[c][0] = temp.substring(temp.lastIndexOf(sep)+1).trim(); |
2 |
26 Feb 07 |
jari |
889 |
records[c][1] = key.substring(key.indexOf('\t')).trim(); |
2 |
26 Feb 07 |
jari |
890 |
|
2 |
26 Feb 07 |
jari |
891 |
list_hit = (String)(categories_population.get(key)); |
2 |
26 Feb 07 |
jari |
892 |
|
2 |
26 Feb 07 |
jari |
893 |
records[c][2] = list_hit; |
2 |
26 Feb 07 |
jari |
894 |
|
2 |
26 Feb 07 |
jari |
895 |
population_total = ((Hashtable)(pop_totals.get(temp))).size(); |
2 |
26 Feb 07 |
jari |
896 |
|
2 |
26 Feb 07 |
jari |
897 |
records[c][3] = String.valueOf(population_total); |
2 |
26 Feb 07 |
jari |
898 |
|
2 |
26 Feb 07 |
jari |
899 |
c++; |
2 |
26 Feb 07 |
jari |
900 |
} |
2 |
26 Feb 07 |
jari |
901 |
|
2 |
26 Feb 07 |
jari |
//jcb |
2 |
26 Feb 07 |
jari |
//accumulates hits for the result matrix. |
2 |
26 Feb 07 |
jari |
904 |
categoryNames = new String[hitAccumulator.size()]; |
2 |
26 Feb 07 |
jari |
905 |
listHitMatrix = new String[categoryNames.length][]; |
2 |
26 Feb 07 |
jari |
906 |
|
2 |
26 Feb 07 |
jari |
907 |
int cnt = 0; |
2 |
26 Feb 07 |
jari |
908 |
Vector locusVector; |
2 |
26 Feb 07 |
jari |
909 |
|
2 |
26 Feb 07 |
jari |
910 |
for(Enumeration e = hitAccumulator.keys(); e.hasMoreElements();){ |
2 |
26 Feb 07 |
jari |
911 |
String cat = (String)(e.nextElement()); |
2 |
26 Feb 07 |
jari |
912 |
categoryNames[cnt] = cat; |
2 |
26 Feb 07 |
jari |
913 |
locusVector = ((Vector)hitAccumulator.get(cat)); |
2 |
26 Feb 07 |
jari |
914 |
listHitMatrix[cnt] = new String[locusVector.size()]; |
2 |
26 Feb 07 |
jari |
915 |
for(int i = 0; i < listHitMatrix[cnt].length; i++){ |
2 |
26 Feb 07 |
jari |
916 |
listHitMatrix[cnt][i] = ((String)locusVector.elementAt(i)); |
2 |
26 Feb 07 |
jari |
917 |
} |
2 |
26 Feb 07 |
jari |
918 |
cnt++; |
2 |
26 Feb 07 |
jari |
919 |
} |
2 |
26 Feb 07 |
jari |
920 |
} |
2 |
26 Feb 07 |
jari |
921 |
|
2 |
26 Feb 07 |
jari |
922 |
} |
2 |
26 Feb 07 |
jari |
923 |
|