First: Qlucore Omics Explorer has extensive inbuilt support for building and using classifiers, see for instance Supported classifier methods.
Additionally you can also use the statistical filtering in combination with PCA to create a variable subset that well models data.
You need to have an annotation indicating which groups that you would like to classify into. You also need an annotation separating training samples from the ones you would like to classify (called new). Then you use the tools available in the statistics dialog and select multiple or two group comparison, depending on how many groups you would like to classify into. By sliding the p-value slider you will select those variables that are best on separating the groups. You can see which the variables are in either a heat map or a Variable PCA plot.
In the variable panel a list is automatically created with all the active variables. You can create your classifier by copying this list. If you prefer you can use the list tool to select which variables to include in the list. This list of selected variables is then your classifier.
The second step is to involve the new samples (the ones that shall be classified using the new classifier). In the Sample tab you can select (the icon with two circles) which of the samples that shall be active versus visible. Select the training samples as active and then select the new samples as visible. In the variable tab you select the list with variables as input. |
