2 |
26 Feb 07 |
jari |
1 |
<html> |
2 |
26 Feb 07 |
jari |
2 |
<body bgcolor="#ffffcc" lang=EN-US style='tab-interval:.5in'> |
2 |
26 Feb 07 |
jari |
3 |
|
2 |
26 Feb 07 |
jari |
4 |
<div class=Section1> |
2 |
26 Feb 07 |
jari |
5 |
|
2 |
26 Feb 07 |
jari |
6 |
<h1>DAM: Discriminant Analysis Module</h1> |
2 |
26 Feb 07 |
jari |
7 |
|
2 |
26 Feb 07 |
jari |
8 |
<h1>Initialization Dialog</h1> |
2 |
26 Feb 07 |
jari |
9 |
|
2 |
26 Feb 07 |
jari |
10 |
<h2>Parameter Information</h2> |
2 |
26 Feb 07 |
jari |
11 |
|
2 |
26 Feb 07 |
jari |
12 |
|
2 |
26 Feb 07 |
jari |
13 |
<div class=MsoNormal align=center style='text-align:center'> |
2 |
26 Feb 07 |
jari |
14 |
|
2 |
26 Feb 07 |
jari |
15 |
<hr size=10 width="100%" align=center> |
2 |
26 Feb 07 |
jari |
16 |
|
2 |
26 Feb 07 |
jari |
17 |
</div> |
2 |
26 Feb 07 |
jari |
18 |
|
2 |
26 Feb 07 |
jari |
19 |
|
2 |
26 Feb 07 |
jari |
20 |
<h2>General DAM Terminology</h2> |
2 |
26 Feb 07 |
jari |
21 |
|
2 |
26 Feb 07 |
jari |
22 |
<p class=MsoNormal>The Primary function of the DAM is to serve as a method for |
2 |
26 Feb 07 |
jari |
23 |
multi-class classification. It incorporates the dimensional reduction method <b><i>Multivariate |
2 |
26 Feb 07 |
jari |
24 |
Partial Least Squares (MPLS)</i></b> and two classification analysis methods <b><i>Polychotomous |
2 |
26 Feb 07 |
jari |
25 |
Discrimination (PDA)</i></b> and <b><i>Quadratic Discriminant Analysis (QDA)</i></b>. |
2 |
26 Feb 07 |
jari |
26 |
Either PDA or QDA are performed after the starting data has been reduced by |
2 |
26 Feb 07 |
jari |
27 |
MPLS. </p> |
2 |
26 Feb 07 |
jari |
28 |
|
2 |
26 Feb 07 |
jari |
29 |
<h2>Classification Selection </h2> |
2 |
26 Feb 07 |
jari |
30 |
|
2 |
26 Feb 07 |
jari |
31 |
<p class=MsoNormal>The classification selection option indicates whether to |
2 |
26 Feb 07 |
jari |
32 |
classify genes or experiments. </p> |
2 |
26 Feb 07 |
jari |
33 |
|
2 |
26 Feb 07 |
jari |
34 |
<h2>Parameters </h2> |
2 |
26 Feb 07 |
jari |
35 |
|
2 |
26 Feb 07 |
jari |
36 |
<p class=MsoNormal><b>Number Of Classes<span style="mso-spacerun: yes"> |
2 |
26 Feb 07 |
jari |
37 |
</span></b><span style="mso-spacerun: yes"> </span>The number of classes.</p> |
2 |
26 Feb 07 |
jari |
38 |
|
2 |
26 Feb 07 |
jari |
39 |
<p class=MsoNormal><![if !supportEmptyParas]> <![endif]><o:p></o:p></p> |
2 |
26 Feb 07 |
jari |
40 |
|
2 |
26 Feb 07 |
jari |
41 |
<p class=MsoNormal><b>Number Of Gene Components </b><span style="mso-spacerun: |
2 |
26 Feb 07 |
jari |
42 |
yes"> </span>The number of gene components for Dimension Reduction.</p> |
2 |
26 Feb 07 |
jari |
43 |
|
2 |
26 Feb 07 |
jari |
44 |
<p class=MsoNormal><![if !supportEmptyParas]> <![endif]><o:p></o:p></p> |
2 |
26 Feb 07 |
jari |
45 |
|
2 |
26 Feb 07 |
jari |
46 |
<p class=MsoNormal><b>Alpha<span style="mso-spacerun: yes"> </span></b><span |
2 |
26 Feb 07 |
jari |
47 |
style="mso-spacerun: yes"> </span>The alpha value chosen for the |
2 |
26 Feb 07 |
jari |
48 |
t-distribution in preliminary gene screening.</p> |
2 |
26 Feb 07 |
jari |
49 |
|
2 |
26 Feb 07 |
jari |
50 |
<h3><span style='font-size:12.0pt;mso-bidi-font-size:13.5pt'>Skip Gene |
2 |
26 Feb 07 |
jari |
51 |
Screening step (ANOVA)</span> <span style='font-size:12.0pt;mso-bidi-font-size: |
2 |
26 Feb 07 |
jari |
52 |
13.5pt;font-weight:normal'><span style="mso-spacerun: yes"> </span><span |
2 |
26 Feb 07 |
jari |
53 |
style="mso-spacerun: yes"> </span>Skips the Gene Screening / Selection |
2 |
26 Feb 07 |
jari |
54 |
step when executing the Initial Classification and Assessment algorithms(A0, A1 |
2 |
26 Feb 07 |
jari |
55 |
or A2) <o:p></o:p></span></h3> |
2 |
26 Feb 07 |
jari |
56 |
|
2 |
26 Feb 07 |
jari |
57 |
<h2><span style='font-size:12.0pt;mso-bidi-font-size:18.0pt'><span |
2 |
26 Feb 07 |
jari |
58 |
style="mso-spacerun: yes"> </span><br> |
2 |
26 Feb 07 |
jari |
59 |
</span>Assessment Algorithm Selection</h2> |
2 |
26 Feb 07 |
jari |
60 |
|
2 |
26 Feb 07 |
jari |
61 |
<p class=MsoNormal>The Initial Classification Algorithm and assessment |
2 |
26 Feb 07 |
jari |
62 |
algorithms A0, A1 & A2 essentially contain the same fundamental stages |
2 |
26 Feb 07 |
jari |
63 |
(i.e. Preliminary Gene Selection, Dimension Reduction(MPLS) & |
2 |
26 Feb 07 |
jari |
64 |
Classification/Prediction based on Leave One Out Cross Validation (LOOCV)) but |
2 |
26 Feb 07 |
jari |
65 |
they are ordered in a different sequence. </p> |
2 |
26 Feb 07 |
jari |
66 |
|
2 |
26 Feb 07 |
jari |
67 |
<h3>Initial Classification algorithm</h3> |
2 |
26 Feb 07 |
jari |
68 |
|
2 |
26 Feb 07 |
jari |
69 |
<ol style='margin-top:0in' start=1 type=1> |
2 |
26 Feb 07 |
jari |
70 |
<li class=MsoNormal style='mso-list:l3 level1 lfo3;tab-stops:list .5in'><i>Preliminary |
2 |
26 Feb 07 |
jari |
71 |
Gene Selection: </i>Use ANOVA to select genes from given gene expression |
2 |
26 Feb 07 |
jari |
72 |
matrix.</li> |
2 |
26 Feb 07 |
jari |
73 |
<li class=MsoNormal style='mso-list:l3 level1 lfo3;tab-stops:list .5in left 4.5in'><i>Dimension |
2 |
26 Feb 07 |
jari |
74 |
Reduction</i>: Use MPLS algorithm to obtain gene components matrix from |
2 |
26 Feb 07 |
jari |
75 |
gene expression matrix</li> |
2 |
26 Feb 07 |
jari |
76 |
<li class=MsoNormal style='mso-list:l3 level1 lfo3;tab-stops:list .5in'><i>Classification</i>:.<br> |
2 |
26 Feb 07 |
jari |
77 |
Use the fitted classifier from training to predict all the test samples in |
2 |
26 Feb 07 |
jari |
78 |
the gene components matrix obtained from step 2. <span |
2 |
26 Feb 07 |
jari |
79 |
style="mso-spacerun: yes"> </span></li> |
2 |
26 Feb 07 |
jari |
80 |
</ol> |
2 |
26 Feb 07 |
jari |
81 |
|
2 |
26 Feb 07 |
jari |
82 |
<h3>A0 algorithm</h3> |
2 |
26 Feb 07 |
jari |
83 |
|
2 |
26 Feb 07 |
jari |
84 |
<ol style='margin-top:0in' start=1 type=1> |
2 |
26 Feb 07 |
jari |
85 |
<li class=MsoNormal style='mso-list:l1 level1 lfo9;tab-stops:list .5in'><i>Preliminary |
2 |
26 Feb 07 |
jari |
86 |
Gene Selection: </i>Use ANOVA to select genes from given gene expression |
2 |
26 Feb 07 |
jari |
87 |
matrix.</li> |
2 |
26 Feb 07 |
jari |
88 |
<li class=MsoNormal style='mso-list:l1 level1 lfo9;tab-stops:list .5in left 4.5in'><i>Dimension |
2 |
26 Feb 07 |
jari |
89 |
Reduction</i>: Use MPLS algorithm to obtain gene components matrix from |
2 |
26 Feb 07 |
jari |
90 |
gene expression matrix</li> |
2 |
26 Feb 07 |
jari |
91 |
<li class=MsoNormal style='mso-list:l1 level1 lfo9;tab-stops:list .5in'><i>Classification/Prediction</i>: |
2 |
26 Feb 07 |
jari |
92 |
Classification is based on LOOCV.<br> |
2 |
26 Feb 07 |
jari |
93 |
For each sample in the gene components matrix obtained from step 2, leave |
2 |
26 Feb 07 |
jari |
94 |
out this sample, and fit classifier to the remaining samples. Use the fitted |
2 |
26 Feb 07 |
jari |
95 |
classifier to predict left out sample.<br> |
2 |
26 Feb 07 |
jari |
96 |
<span style="mso-spacerun: yes"> </span><br> |
2 |
26 Feb 07 |
jari |
97 |
<b>NOTE</b>: For a given expression matrix, steps 1 (Preliminary Gene |
2 |
26 Feb 07 |
jari |
98 |
Selection) & 2 (Dimension Reduction) are <i>fixed</i> with respect to |
2 |
26 Feb 07 |
jari |
99 |
LOOCV. Thus, the effect of gene selection & dimension reduction on the |
2 |
26 Feb 07 |
jari |
100 |
classification cannot be assessed. </li> |
2 |
26 Feb 07 |
jari |
101 |
</ol> |
2 |
26 Feb 07 |
jari |
102 |
|
2 |
26 Feb 07 |
jari |
103 |
<h3>A1 algorithm</h3> |
2 |
26 Feb 07 |
jari |
104 |
|
2 |
26 Feb 07 |
jari |
105 |
<p class=MsoNormal style='margin-left:39.0pt;text-indent:-.25in;mso-list:l2 level1 lfo5; |
2 |
26 Feb 07 |
jari |
106 |
tab-stops:list 39.0pt'><![if !supportLists]>1.<span style='font:7.0pt "Times New Roman"'> |
2 |
26 Feb 07 |
jari |
107 |
</span><![endif]><i>Preliminary Gene Selection</i>: Use ANOVA to select genes |
2 |
26 Feb 07 |
jari |
108 |
from given gene expression matrix.</p> |
2 |
26 Feb 07 |
jari |
109 |
|
2 |
26 Feb 07 |
jari |
110 |
<p class=MsoNormal style='margin-left:21.0pt'><i><![if !supportEmptyParas]> <![endif]><o:p></o:p></i></p> |
2 |
26 Feb 07 |
jari |
111 |
|
2 |
26 Feb 07 |
jari |
112 |
<p class=MsoNormal style='margin-left:21.0pt'>For each sample in the gene |
2 |
26 Feb 07 |
jari |
113 |
expression matrix, leave out this sample to obtain a sub expression matrix.</p> |
2 |
26 Feb 07 |
jari |
114 |
|
2 |
26 Feb 07 |
jari |
115 |
<p class=MsoNormal style='margin-left:39.0pt;text-indent:-.25in;mso-list:l2 level1 lfo5; |
2 |
26 Feb 07 |
jari |
116 |
tab-stops:list 39.0pt'><![if !supportLists]>2.<span style='font:7.0pt "Times New Roman"'> |
2 |
26 Feb 07 |
jari |
117 |
</span><![endif]><i>Dimension Reduction</i>: Use MPLS algorithm to obtain gene |
2 |
26 Feb 07 |
jari |
118 |
components matrix from the sub gene expression matrix.</p> |
2 |
26 Feb 07 |
jari |
119 |
|
2 |
26 Feb 07 |
jari |
120 |
<p class=MsoNormal style='margin-left:39.0pt;text-indent:-.25in;mso-list:l2 level1 lfo5; |
2 |
26 Feb 07 |
jari |
121 |
tab-stops:list 39.0pt'><![if !supportLists]>3.<span style='font:7.0pt "Times New Roman"'> |
2 |
26 Feb 07 |
jari |
122 |
</span><![endif]><i>Classification/Prediction</i>: Fit classifier to the |
2 |
26 Feb 07 |
jari |
123 |
remaining samples in gene components matrix. Use the fitted classifier to |
2 |
26 Feb 07 |
jari |
124 |
predict left out sample.</p> |
2 |
26 Feb 07 |
jari |
125 |
|
2 |
26 Feb 07 |
jari |
126 |
<p class=MsoNormal style='margin-left:21.0pt'><i><![if !supportEmptyParas]> <![endif]><o:p></o:p></i></p> |
2 |
26 Feb 07 |
jari |
127 |
|
2 |
26 Feb 07 |
jari |
128 |
<p class=MsoNormal style='margin-left:21.0pt'><b>NOTE</b>: The first modification |
2 |
26 Feb 07 |
jari |
129 |
to <i>A0</i> given as <i>A1</i> assesses the affects of dimension reduction. |
2 |
26 Feb 07 |
jari |
130 |
The dimension reduction as well as the classifier is refitted <i>N</i> times, |
2 |
26 Feb 07 |
jari |
131 |
one for each sample left out. </p> |
2 |
26 Feb 07 |
jari |
132 |
|
2 |
26 Feb 07 |
jari |
133 |
<h3>A2 algorithm</h3> |
2 |
26 Feb 07 |
jari |
134 |
|
2 |
26 Feb 07 |
jari |
135 |
<p class=MsoNormal>For each sample in the gene expression matrix, leave out |
2 |
26 Feb 07 |
jari |
136 |
this sample to obtain a sub expression matrix.</p> |
2 |
26 Feb 07 |
jari |
137 |
|
2 |
26 Feb 07 |
jari |
138 |
<ol style='margin-top:0in' start=1 type=1> |
2 |
26 Feb 07 |
jari |
139 |
<li class=MsoNormal style='mso-list:l0 level1 lfo8;tab-stops:list .5in'><i>Preliminary |
2 |
26 Feb 07 |
jari |
140 |
Gene Selection</i>: Use ANOVA to select genes from the sub expression |
2 |
26 Feb 07 |
jari |
141 |
matrix.</li> |
2 |
26 Feb 07 |
jari |
142 |
<li class=MsoNormal style='mso-list:l0 level1 lfo8;tab-stops:list .5in'><i>Dimension |
2 |
26 Feb 07 |
jari |
143 |
Reduction</i>: Use MPLS algorithm to obtain gene components matrix from |
2 |
26 Feb 07 |
jari |
144 |
the sub expression matrix.</li> |
2 |
26 Feb 07 |
jari |
145 |
<li class=MsoNormal style='mso-list:l0 level1 lfo8;tab-stops:list .5in'><i>Classification/Prediction</i>: |
2 |
26 Feb 07 |
jari |
146 |
Fit classifier to the remaining samples in gene components matrix. Use the |
2 |
26 Feb 07 |
jari |
147 |
fitted classifier to predict left out sample.</li> |
2 |
26 Feb 07 |
jari |
148 |
</ol> |
2 |
26 Feb 07 |
jari |
149 |
|
2 |
26 Feb 07 |
jari |
150 |
<p class=MsoNormal style='margin-left:.25in'><i><![if !supportEmptyParas]> <![endif]><o:p></o:p></i></p> |
2 |
26 Feb 07 |
jari |
151 |
|
2 |
26 Feb 07 |
jari |
152 |
<p class=MsoNormal style='margin-left:.25in'><b>NOTE</b>: The second |
2 |
26 Feb 07 |
jari |
153 |
modification to <i>A0</i> given as <i>A2</i> assesses the affects of gene |
2 |
26 Feb 07 |
jari |
154 |
selection. The gene set is reselected for classification each time a sample is |
2 |
26 Feb 07 |
jari |
155 |
left out. </p> |
2 |
26 Feb 07 |
jari |
156 |
|
2 |
26 Feb 07 |
jari |
157 |
<h2>Classification Method Selection</h2> |
2 |
26 Feb 07 |
jari |
158 |
|
2 |
26 Feb 07 |
jari |
159 |
<p class=MsoNormal>The Classification methods are performed after the data has |
2 |
26 Feb 07 |
jari |
160 |
reduced by dimension reduction (MPLS) </p> |
2 |
26 Feb 07 |
jari |
161 |
|
2 |
26 Feb 07 |
jari |
162 |
<h3>PDA</h3> |
2 |
26 Feb 07 |
jari |
163 |
|
2 |
26 Feb 07 |
jari |
164 |
<p class=MsoNormal>Polychotomous Discrimination </p> |
2 |
26 Feb 07 |
jari |
165 |
|
2 |
26 Feb 07 |
jari |
166 |
<h3>QDA</h3> |
2 |
26 Feb 07 |
jari |
167 |
|
2 |
26 Feb 07 |
jari |
168 |
<p class=MsoNormal>Quadratic Discriminant Analysis </basefont></p> |
2 |
26 Feb 07 |
jari |
169 |
|
2 |
26 Feb 07 |
jari |
170 |
</div> |
2 |
26 Feb 07 |
jari |
171 |
|
2 |
26 Feb 07 |
jari |
172 |
</body> |
2 |
26 Feb 07 |
jari |
173 |
|
2 |
26 Feb 07 |
jari |
174 |
</html> |