43 |
21 Nov 05 |
enell |
1 |
/* |
784 |
18 Sep 08 |
jari |
$Id$ |
784 |
18 Sep 08 |
jari |
3 |
|
784 |
18 Sep 08 |
jari |
Copyright (C) 2005 Johan Enell |
793 |
25 Sep 08 |
jari |
Copyright (C) 2008 Jari Häkkinen |
784 |
18 Sep 08 |
jari |
6 |
|
784 |
18 Sep 08 |
jari |
This file is part of the se.lu.onk.ReplicateError plug-in for |
784 |
18 Sep 08 |
jari |
BASE. Available at http://baseplugins.thep.lu.se/ and BASE web |
784 |
18 Sep 08 |
jari |
site is http://base.thep.lu.se |
784 |
18 Sep 08 |
jari |
10 |
|
784 |
18 Sep 08 |
jari |
This is free software; you can redistribute it and/or modify it |
784 |
18 Sep 08 |
jari |
under the terms of the GNU General Public License as published by |
784 |
18 Sep 08 |
jari |
the Free Software Foundation; either version 3 of the License, or |
784 |
18 Sep 08 |
jari |
(at your option) any later version. |
784 |
18 Sep 08 |
jari |
15 |
|
784 |
18 Sep 08 |
jari |
The software is distributed in the hope that it will be useful, but |
784 |
18 Sep 08 |
jari |
WITHOUT ANY WARRANTY; without even the implied warranty of |
784 |
18 Sep 08 |
jari |
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
784 |
18 Sep 08 |
jari |
General Public License for more details. |
784 |
18 Sep 08 |
jari |
20 |
|
784 |
18 Sep 08 |
jari |
You should have received a copy of the GNU General Public License |
784 |
18 Sep 08 |
jari |
along with BASE. If not, see <http://www.gnu.org/licenses/>. |
784 |
18 Sep 08 |
jari |
23 |
*/ |
43 |
21 Nov 05 |
enell |
24 |
package replicateerror.WithinAssays; |
43 |
21 Nov 05 |
enell |
25 |
|
43 |
21 Nov 05 |
enell |
26 |
import java.io.File; |
43 |
21 Nov 05 |
enell |
27 |
import java.io.IOException; |
43 |
21 Nov 05 |
enell |
28 |
import java.io.PrintStream; |
43 |
21 Nov 05 |
enell |
29 |
import java.util.HashMap; |
43 |
21 Nov 05 |
enell |
30 |
import java.util.Iterator; |
43 |
21 Nov 05 |
enell |
31 |
import java.util.List; |
43 |
21 Nov 05 |
enell |
32 |
|
43 |
21 Nov 05 |
enell |
33 |
import basefile.BASEFileReader; |
43 |
21 Nov 05 |
enell |
34 |
import basefile.BASEFileSection; |
43 |
21 Nov 05 |
enell |
35 |
import basefile.PluginException; |
52 |
25 Nov 05 |
enell |
36 |
import replicateerror.R; |
43 |
21 Nov 05 |
enell |
37 |
import replicateerror.Replicate; |
43 |
21 Nov 05 |
enell |
38 |
import replicateerror.ReplicateError; |
80 |
24 Mar 06 |
enell |
39 |
import replicateerror.Spot; |
43 |
21 Nov 05 |
enell |
40 |
import replicateerror.AcrossAssays.HTML; |
43 |
21 Nov 05 |
enell |
41 |
|
43 |
21 Nov 05 |
enell |
42 |
public class ReplicateErrorWA extends ReplicateError |
43 |
21 Nov 05 |
enell |
43 |
{ |
80 |
24 Mar 06 |
enell |
44 |
private HashMap<Integer, String> assayName; |
43 |
21 Nov 05 |
enell |
45 |
|
43 |
21 Nov 05 |
enell |
46 |
|
43 |
21 Nov 05 |
enell |
47 |
public ReplicateErrorWA() |
43 |
21 Nov 05 |
enell |
48 |
{ |
43 |
21 Nov 05 |
enell |
49 |
super(); |
80 |
24 Mar 06 |
enell |
50 |
assayName = new HashMap<Integer, String>(); |
43 |
21 Nov 05 |
enell |
51 |
File dataDir = new File("data"); |
43 |
21 Nov 05 |
enell |
52 |
dataDir.mkdir(); |
43 |
21 Nov 05 |
enell |
53 |
} |
43 |
21 Nov 05 |
enell |
54 |
|
43 |
21 Nov 05 |
enell |
55 |
@Override |
43 |
21 Nov 05 |
enell |
56 |
public void extractAssays(BASEFileSection section, BASEFileReader bfr) throws PluginException |
43 |
21 Nov 05 |
enell |
57 |
{ |
43 |
21 Nov 05 |
enell |
58 |
int nameCol = section.findFieldList("columns").indexOf("name"); |
43 |
21 Nov 05 |
enell |
59 |
int idCol = section.findFieldList("columns").indexOf("id"); |
43 |
21 Nov 05 |
enell |
60 |
|
43 |
21 Nov 05 |
enell |
61 |
try |
43 |
21 Nov 05 |
enell |
62 |
{ |
43 |
21 Nov 05 |
enell |
63 |
String[] vals; |
43 |
21 Nov 05 |
enell |
64 |
while ((vals = bfr.readDataRow()) != null) |
43 |
21 Nov 05 |
enell |
65 |
{ |
80 |
24 Mar 06 |
enell |
66 |
assayName.put(new Integer(vals[idCol]), vals[nameCol]); |
43 |
21 Nov 05 |
enell |
67 |
} |
43 |
21 Nov 05 |
enell |
68 |
} |
43 |
21 Nov 05 |
enell |
69 |
catch (IOException e) |
43 |
21 Nov 05 |
enell |
70 |
{ |
43 |
21 Nov 05 |
enell |
71 |
e.printStackTrace(); |
43 |
21 Nov 05 |
enell |
72 |
} |
43 |
21 Nov 05 |
enell |
73 |
} |
43 |
21 Nov 05 |
enell |
74 |
|
43 |
21 Nov 05 |
enell |
75 |
@Override |
43 |
21 Nov 05 |
enell |
76 |
public void extractSpots(BASEFileSection section, BASEFileReader bfr) |
43 |
21 Nov 05 |
enell |
77 |
{ |
43 |
21 Nov 05 |
enell |
78 |
List<String> columns = section.findFieldList("columns"); |
43 |
21 Nov 05 |
enell |
79 |
List<String> assayFields = section.findFieldList("assayFields"); |
80 |
24 Mar 06 |
enell |
80 |
List<Integer> assays = section.findFieldIntList("assays"); |
43 |
21 Nov 05 |
enell |
81 |
|
43 |
21 Nov 05 |
enell |
82 |
int posCol = columns.indexOf("position"); |
43 |
21 Nov 05 |
enell |
83 |
int repCol = columns.indexOf("reporter"); |
43 |
21 Nov 05 |
enell |
84 |
int assayData = columns.indexOf("assayData"); |
43 |
21 Nov 05 |
enell |
85 |
int int1Col = assayFields.indexOf("intensity1"); |
43 |
21 Nov 05 |
enell |
86 |
int int2Col = assayFields.indexOf("intensity2"); |
43 |
21 Nov 05 |
enell |
87 |
|
80 |
24 Mar 06 |
enell |
88 |
HashMap<Integer, Integer> assayColumns = new HashMap<Integer, Integer>(); |
43 |
21 Nov 05 |
enell |
89 |
for (int i = 0; i < assays.size(); i++) |
43 |
21 Nov 05 |
enell |
90 |
{ |
43 |
21 Nov 05 |
enell |
91 |
assayColumns.put(assays.get(i), assayData + assayFields.size() * i); |
43 |
21 Nov 05 |
enell |
92 |
} |
43 |
21 Nov 05 |
enell |
93 |
|
43 |
21 Nov 05 |
enell |
94 |
BASEFileSection spotSection = new BASEFileSection(); |
43 |
21 Nov 05 |
enell |
95 |
spotSection.setHeader("section", "spots"); |
43 |
21 Nov 05 |
enell |
96 |
spotSection.setHeader("channels", "2"); |
43 |
21 Nov 05 |
enell |
97 |
spotSection.setHeader("assayFields", "intensity1\tintensity2"); |
43 |
21 Nov 05 |
enell |
98 |
spotSection.setHeader("columns", "position\treporter\tassayData"); |
43 |
21 Nov 05 |
enell |
99 |
spotSection.setHeader("assays", ""); |
43 |
21 Nov 05 |
enell |
100 |
|
43 |
21 Nov 05 |
enell |
101 |
try |
43 |
21 Nov 05 |
enell |
102 |
{ |
43 |
21 Nov 05 |
enell |
103 |
HTML.openPlotsTable(center.toString()); |
80 |
24 Mar 06 |
enell |
104 |
for (Integer assay : assays) |
43 |
21 Nov 05 |
enell |
105 |
{ |
43 |
21 Nov 05 |
enell |
106 |
HashMap<Integer, Replicate> replicates = new HashMap<Integer, Replicate>(); |
80 |
24 Mar 06 |
enell |
107 |
HashMap<Integer, Replicate> singletons = new HashMap<Integer, Replicate>(); |
43 |
21 Nov 05 |
enell |
108 |
|
43 |
21 Nov 05 |
enell |
109 |
String[] vals; |
43 |
21 Nov 05 |
enell |
110 |
while ((vals = bfr.readDataRow()) != null) |
43 |
21 Nov 05 |
enell |
111 |
{ |
80 |
24 Mar 06 |
enell |
112 |
Integer reporter = new Integer(vals[repCol]); |
43 |
21 Nov 05 |
enell |
113 |
|
43 |
21 Nov 05 |
enell |
114 |
Replicate replicate = replicates.get(reporter); |
43 |
21 Nov 05 |
enell |
115 |
if (replicate == null) |
43 |
21 Nov 05 |
enell |
116 |
{ |
80 |
24 Mar 06 |
enell |
117 |
replicate = new Replicate(); |
43 |
21 Nov 05 |
enell |
118 |
} |
43 |
21 Nov 05 |
enell |
119 |
try |
43 |
21 Nov 05 |
enell |
120 |
{ |
80 |
24 Mar 06 |
enell |
121 |
int position = new Integer(vals[posCol]); |
43 |
21 Nov 05 |
enell |
122 |
float int1 = new Float(vals[assayColumns.get(assay) + int1Col]); |
43 |
21 Nov 05 |
enell |
123 |
float int2 = new Float(vals[assayColumns.get(assay) + int2Col]); |
43 |
21 Nov 05 |
enell |
124 |
if (int1 > 0 && int2 > 0) |
43 |
21 Nov 05 |
enell |
125 |
{ |
80 |
24 Mar 06 |
enell |
126 |
Spot s = new Spot(new Integer(assay)); |
80 |
24 Mar 06 |
enell |
127 |
s.add(position, int1, int2); |
80 |
24 Mar 06 |
enell |
128 |
replicate.add(s); |
43 |
21 Nov 05 |
enell |
129 |
replicates.put(reporter, replicate); |
43 |
21 Nov 05 |
enell |
130 |
} |
43 |
21 Nov 05 |
enell |
131 |
} |
43 |
21 Nov 05 |
enell |
132 |
catch (NumberFormatException e) |
43 |
21 Nov 05 |
enell |
133 |
{} |
43 |
21 Nov 05 |
enell |
134 |
} |
43 |
21 Nov 05 |
enell |
135 |
|
52 |
25 Nov 05 |
enell |
// remove unusable replicates |
791 |
24 Sep 08 |
jari |
137 |
purgeUnusableReplicates(replicates, singletons); |
43 |
21 Nov 05 |
enell |
138 |
|
795 |
01 Oct 08 |
jari |
// calculate mean or median before filter |
795 |
01 Oct 08 |
jari |
140 |
List<Float> mList = getMList(replicates.values()); |
43 |
21 Nov 05 |
enell |
141 |
float averageBefore; |
795 |
01 Oct 08 |
jari |
142 |
int nbrBefore = mList.size(); |
43 |
21 Nov 05 |
enell |
143 |
if (center == Center.median) |
43 |
21 Nov 05 |
enell |
144 |
{ |
43 |
21 Nov 05 |
enell |
145 |
averageBefore = median(mList); |
43 |
21 Nov 05 |
enell |
146 |
} |
43 |
21 Nov 05 |
enell |
147 |
else |
43 |
21 Nov 05 |
enell |
148 |
{ |
43 |
21 Nov 05 |
enell |
149 |
averageBefore = mean(mList); |
43 |
21 Nov 05 |
enell |
150 |
} |
43 |
21 Nov 05 |
enell |
// Calculate standard deviation |
43 |
21 Nov 05 |
enell |
152 |
float sdBefore = standardDiviation(mList, mean(mList)); |
43 |
21 Nov 05 |
enell |
153 |
|
80 |
24 Mar 06 |
enell |
154 |
R.setABRBefore(assayName.get(assay), caluculateABR(replicates.values())); |
52 |
25 Nov 05 |
enell |
155 |
|
791 |
24 Sep 08 |
jari |
// print result to file for later use in R |
80 |
24 Mar 06 |
enell |
157 |
printDuplicates(replicates.values(), assayName.get(assay) + ".data"); |
43 |
21 Nov 05 |
enell |
158 |
|
80 |
24 Mar 06 |
enell |
// filter replicates abs(M - mean) against 'sdLimit' number of sd's |
2011 |
09 Jul 13 |
olle |
// A negative 'sdLimit' indicates no filtering |
2011 |
09 Jul 13 |
olle |
161 |
if (sdLimit > 0f) |
43 |
21 Nov 05 |
enell |
162 |
{ |
2011 |
09 Jul 13 |
olle |
163 |
for (Replicate r : replicates.values()) |
2011 |
09 Jul 13 |
olle |
164 |
{ |
2011 |
09 Jul 13 |
olle |
165 |
r.filter(averageBefore, sdLimit * sdBefore); |
2011 |
09 Jul 13 |
olle |
166 |
} |
43 |
21 Nov 05 |
enell |
167 |
} |
791 |
24 Sep 08 |
jari |
// print result to file for later use in R |
80 |
24 Mar 06 |
enell |
169 |
printDuplicates(replicates.values(), assayName.get(assay) + "good.data"); |
43 |
21 Nov 05 |
enell |
170 |
|
43 |
21 Nov 05 |
enell |
// calculate mean |
80 |
24 Mar 06 |
enell |
172 |
mList = getMList(replicates.values()); |
43 |
21 Nov 05 |
enell |
173 |
float averageAfter; |
43 |
21 Nov 05 |
enell |
174 |
if (center == Center.median) |
43 |
21 Nov 05 |
enell |
175 |
{ |
43 |
21 Nov 05 |
enell |
176 |
averageAfter = median(mList); |
43 |
21 Nov 05 |
enell |
177 |
} |
43 |
21 Nov 05 |
enell |
178 |
else |
43 |
21 Nov 05 |
enell |
179 |
{ |
43 |
21 Nov 05 |
enell |
180 |
averageAfter = mean(mList); |
43 |
21 Nov 05 |
enell |
181 |
} |
43 |
21 Nov 05 |
enell |
// Calculate standard deviation |
43 |
21 Nov 05 |
enell |
183 |
float sdAfter = standardDiviation(mList, mean(mList)); |
43 |
21 Nov 05 |
enell |
184 |
|
80 |
24 Mar 06 |
enell |
185 |
R.setABRAfter(assayName.get(assay), caluculateABR(replicates.values())); |
80 |
24 Mar 06 |
enell |
186 |
|
80 |
24 Mar 06 |
enell |
// print to stdout |
80 |
24 Mar 06 |
enell |
188 |
spotSection.setHeader("assays", assay); |
80 |
24 Mar 06 |
enell |
189 |
System.out.println(spotSection); |
80 |
24 Mar 06 |
enell |
190 |
printSpots(assay, replicates); |
80 |
24 Mar 06 |
enell |
191 |
printSpots(assay, singletons); |
80 |
24 Mar 06 |
enell |
192 |
System.out.println(); |
43 |
21 Nov 05 |
enell |
193 |
|
795 |
01 Oct 08 |
jari |
194 |
HTML.addPlotsTable(assayName.get(assay), averageBefore, sdBefore, |
795 |
01 Oct 08 |
jari |
195 |
nbrBefore, averageAfter, sdAfter, mList.size()); |
795 |
01 Oct 08 |
jari |
196 |
|
43 |
21 Nov 05 |
enell |
197 |
bfr.resetDataPosition(section); |
43 |
21 Nov 05 |
enell |
198 |
} |
43 |
21 Nov 05 |
enell |
199 |
HTML.closePlotsTable(); |
52 |
25 Nov 05 |
enell |
200 |
|
52 |
25 Nov 05 |
enell |
201 |
R.printPlotFile(); |
43 |
21 Nov 05 |
enell |
202 |
} |
43 |
21 Nov 05 |
enell |
203 |
catch (IOException e) |
43 |
21 Nov 05 |
enell |
204 |
{ |
43 |
21 Nov 05 |
enell |
205 |
e.printStackTrace(); |
43 |
21 Nov 05 |
enell |
206 |
} |
43 |
21 Nov 05 |
enell |
207 |
} |
43 |
21 Nov 05 |
enell |
208 |
} |