Jun 17 2019
The MCS method is K-greedy algorithm which simultaneously grows K different common substructures applying scoring and resorting on each step.
To display the interactive MCS dendrogram:
You can display and undisplay the points and labels by using the check boxes in the columns "visible", "showName" and "showStr". You can right click on the column and choose (un)check all. Points in the dendrogram are fully interactive and linked to the table. You can zoom and translate the plot around the space. The points are colored by > blue = common substructure, green = original compounds pink = root (most common)
A Self Organized Map (SOM) is a way to represent higher dimensional data in 2D (or 3D) such that similar data is grouped together. Nodes in 2D map are called neurons. Each neuron is assigned a weight vector of the same dimensionality as input space - you can read more here.
In ICM the initial placement can be either 1). random or 2). find four mutually remote points in the input dataset and put in the corners and uniformly distribute the rest. The training is undertaken in the following way: a vector is chosen at random from the training set, then ICM finds the node with the closest distance to the chosen vector. Next the radius of the Best Matching Unit (BMU) neighborhood is calculated and then the weights of the BMU neighborhood nodes are adjusted - this is repeated for N iterations. The final mapping stage assigns input vectors to their closest nodes.
To run the Self Orgnanized Map method in ICM:
A map containing the nodes be displayed as shown below. Each node contains a stacked cluster of input compounds. Sequential clicking on the node loops through the representatives.
|Copyright© 1989-2019, Molsoft,LLC - All Rights Reserved.|
This document contains proprietary and confidential information of
The content of this document may not be disclosed to third parties, copied or duplicated in any form,
in whole or in part, without the prior written permission from Molsoft, LLC.