2 |
26 Feb 07 |
jari |
1 |
<html> |
2 |
26 Feb 07 |
jari |
2 |
<body bgcolor = "#FFFFCC"><basefont face = "Arial"> |
2 |
26 Feb 07 |
jari |
3 |
<h1>SOM: Self Organizing Maps</h1><h2>Parameter Information</h2> |
2 |
26 Feb 07 |
jari |
4 |
<hr size = 10> |
2 |
26 Feb 07 |
jari |
5 |
|
2 |
26 Feb 07 |
jari |
6 |
<h2>Basic Terminology</h2> |
2 |
26 Feb 07 |
jari |
7 |
Node: an SOM structure to which expression elements are associated to form clusters. |
2 |
26 Feb 07 |
jari |
8 |
<br><br> |
2 |
26 Feb 07 |
jari |
9 |
SOM Vector: a vector of size n, associated with a node that represents the nodes location |
2 |
26 Feb 07 |
jari |
10 |
in the n dimensional space. Each node has one SOM Vector. |
2 |
26 Feb 07 |
jari |
11 |
<br><br> |
2 |
26 Feb 07 |
jari |
12 |
Training/Adaptation: the process of repositioning the SOM Nodes by altering their associated |
2 |
26 Feb 07 |
jari |
13 |
SOM Vectors. The adaptation process is a result of an expression element being associated |
2 |
26 Feb 07 |
jari |
14 |
with a node. The new position is determined by the distance between the expression element |
2 |
26 Feb 07 |
jari |
15 |
and the SOM Vector, the Alpha value, and the neighborhood convention (see below). |
2 |
26 Feb 07 |
jari |
16 |
<br><br> |
2 |
26 Feb 07 |
jari |
17 |
Topology: a two dimensional topology used to define how node-to-node distances are calculated. |
2 |
26 Feb 07 |
jari |
18 |
<br><br> |
2 |
26 Feb 07 |
jari |
19 |
Note that a cluster is a collection of expression elements associated with a Node. |
2 |
26 Feb 07 |
jari |
20 |
<br> |
2 |
26 Feb 07 |
jari |
21 |
|
2 |
26 Feb 07 |
jari |
22 |
<h2>Sample Selection</h2> |
2 |
26 Feb 07 |
jari |
23 |
The sample selection option indicates whether to cluster genes or experiments. |
2 |
26 Feb 07 |
jari |
24 |
<br> |
2 |
26 Feb 07 |
jari |
25 |
|
2 |
26 Feb 07 |
jari |
26 |
<h2> Distance Metric Selection </h2> |
2 |
26 Feb 07 |
jari |
27 |
This area allows the selection of the metric to be used to assess gene-to-gene |
2 |
26 Feb 07 |
jari |
28 |
or sample-to-sample distances. The initial metric displayed (choosen) corresponds to the global |
2 |
26 Feb 07 |
jari |
29 |
setting in the Multiple Array Viewer's 'Metrics' menu. Alterations to the |
2 |
26 Feb 07 |
jari |
30 |
chosen metric in this dialog will only alter the metric used for the current |
2 |
26 Feb 07 |
jari |
31 |
algorithm run. The global setting in the main 'Metrics' menu will remain unchanged. |
2 |
26 Feb 07 |
jari |
32 |
<br><br> |
2 |
26 Feb 07 |
jari |
33 |
An appendix in the MeV manual describes the distance metrics offered in MeV. |
2 |
26 Feb 07 |
jari |
34 |
<br> |
2 |
26 Feb 07 |
jari |
35 |
|
2 |
26 Feb 07 |
jari |
36 |
<h2>Dimension X</h2> |
2 |
26 Feb 07 |
jari |
37 |
This positive integer value determines the X dimension of the resulting topology. |
2 |
26 Feb 07 |
jari |
38 |
<br> |
2 |
26 Feb 07 |
jari |
39 |
|
2 |
26 Feb 07 |
jari |
40 |
<h2>Dimension Y</h2> |
2 |
26 Feb 07 |
jari |
41 |
This positive integer value determines the Y dimension of the resulting topology. |
2 |
26 Feb 07 |
jari |
42 |
Note that Dimension X times Dimension Y gives the number of clusters that will |
2 |
26 Feb 07 |
jari |
43 |
be produced. |
2 |
26 Feb 07 |
jari |
44 |
<br> |
2 |
26 Feb 07 |
jari |
45 |
|
2 |
26 Feb 07 |
jari |
46 |
<h2>Iterations</h2> |
2 |
26 Feb 07 |
jari |
47 |
This positive integer value indicates the total number of times that the data set |
2 |
26 Feb 07 |
jari |
48 |
will be presented to the network (or Map, Graph). Each expression element |
2 |
26 Feb 07 |
jari |
49 |
will be presented this number of times to train the Nodes. |
2 |
26 Feb 07 |
jari |
50 |
<br> |
2 |
26 Feb 07 |
jari |
51 |
|
2 |
26 Feb 07 |
jari |
52 |
<h2>Alpha</h2> |
2 |
26 Feb 07 |
jari |
53 |
This value is used to scale the alteration of SOM vectors when a new expression |
2 |
26 Feb 07 |
jari |
54 |
vector is associated with a node. |
2 |
26 Feb 07 |
jari |
55 |
<br> |
2 |
26 Feb 07 |
jari |
56 |
|
2 |
26 Feb 07 |
jari |
57 |
|
2 |
26 Feb 07 |
jari |
58 |
<h2>Radius</h2> |
2 |
26 Feb 07 |
jari |
59 |
When using the bubble neighborhood parameter this float value is used to |
2 |
26 Feb 07 |
jari |
60 |
define the extent of the neighborhood. If an SOM vector is within this |
2 |
26 Feb 07 |
jari |
61 |
distance from the winning node (the cluster to which an element has |
2 |
26 Feb 07 |
jari |
62 |
been assigned) then that Node (and SOM vector) is considered to be in the neighborhood |
2 |
26 Feb 07 |
jari |
63 |
and it's SOM vector is adapted. |
2 |
26 Feb 07 |
jari |
64 |
<br> |
2 |
26 Feb 07 |
jari |
65 |
|
2 |
26 Feb 07 |
jari |
66 |
<h2>Initialization</h2> |
2 |
26 Feb 07 |
jari |
67 |
Random Genes or Random Experiments: Indicates that the initial SOM vectors will be selected |
2 |
26 Feb 07 |
jari |
68 |
at random as actual elements in the data. |
2 |
26 Feb 07 |
jari |
69 |
<br><br> |
2 |
26 Feb 07 |
jari |
70 |
Random Vector: Indicates that the initial SOM vectors will be constructed as random vectors |
2 |
26 Feb 07 |
jari |
71 |
generated to reflect the magnitude of the data set. These initial vectors are not actual |
2 |
26 Feb 07 |
jari |
72 |
expression vectors in the data set. |
2 |
26 Feb 07 |
jari |
73 |
<br> |
2 |
26 Feb 07 |
jari |
74 |
|
2 |
26 Feb 07 |
jari |
75 |
<h2>Neighborhood</h2> |
2 |
26 Feb 07 |
jari |
76 |
The neighborhood options indicate the conventions (formulas) used to update (adapt) an SOM vector |
2 |
26 Feb 07 |
jari |
77 |
once an expression vector has been added into a Node's neighborhood. |
2 |
26 Feb 07 |
jari |
78 |
<br><br> |
2 |
26 Feb 07 |
jari |
79 |
Bubble: This option uses the provided radius (see above) to determine which surrounding |
2 |
26 Feb 07 |
jari |
80 |
SOM nodes are in the neighborhood and therefore are candidates for adaptation. |
2 |
26 Feb 07 |
jari |
81 |
When this option is selected the Alpha parameter for scaling the adaptation is used directly |
2 |
26 Feb 07 |
jari |
82 |
as provided from the user. |
2 |
26 Feb 07 |
jari |
83 |
<br><br> |
2 |
26 Feb 07 |
jari |
84 |
Gaussian: This option forces all SOM vectors in the network to be adapted regardless of proximity to the |
2 |
26 Feb 07 |
jari |
85 |
winning node. In this case the Alpha parameter is scaled based on the distance between |
2 |
26 Feb 07 |
jari |
86 |
the SOM vector to be adapted and the winning node's SOM vector. |
2 |
26 Feb 07 |
jari |
87 |
<br> |
2 |
26 Feb 07 |
jari |
88 |
|
2 |
26 Feb 07 |
jari |
89 |
<h2>Topology</h2> |
2 |
26 Feb 07 |
jari |
90 |
Indicates whether the topology should be rectangular or hexagonal. If rectangular topology is selected |
2 |
26 Feb 07 |
jari |
91 |
the node-to-node distance is determined as Euclidean distance within the two dimensional |
2 |
26 Feb 07 |
jari |
92 |
x-y grid. If hexagonal distance is used an appropriate formula is used to determine the distance |
2 |
26 Feb 07 |
jari |
93 |
given the coordinates of the two nodes. |
2 |
26 Feb 07 |
jari |
94 |
<br> |
2 |
26 Feb 07 |
jari |
95 |
|
2 |
26 Feb 07 |
jari |
96 |
<h2> Hierarchical Clustering </h2> |
2 |
26 Feb 07 |
jari |
97 |
This check box selects whether to perform hierarchical clustering on the elements in each cluster |
2 |
26 Feb 07 |
jari |
98 |
created. |
2 |
26 Feb 07 |
jari |
99 |
|
2 |
26 Feb 07 |
jari |
100 |
|
2 |
26 Feb 07 |
jari |
101 |
|
2 |
26 Feb 07 |
jari |
102 |
</basefont> |
2 |
26 Feb 07 |
jari |
103 |
</body> |
2 |
26 Feb 07 |
jari |
104 |
</html> |