Descriptor selection for PCA

In the 'PCA descriptor selection' form, all the Dragon molecular descriptors are listed on the basis of their logical blocks and sub-blocks. The user can select the blocks, sub-blocks and single descriptors to be used for PCA calculation by checking the corresponding check boxes. Note that check boxes corresponding to descriptor blocks (or sub-blocks or single descriptors) previously excluded from the calculation are disabled.

 

In the 'Calculation options', the user can decide:

to calculate PCA on all the calculated descriptors together;
to calculate PCA separately on each selected descriptor block;
to calculate PCA separately on each selected descriptor sub-block.

 

The number of descriptors selected for the PCA calculation is displayed, together with the number of descriptors excluded on the basis of their standard deviations. Note that descriptors with standard deviation lower than 0.0001 are automatically excluded from the PCA calculation.

 

If external variables were loaded, the user can decide to use the imported external variables in the PCA calculation by clicking the 'select external variable' button and then selecting the desired external variables. If PCA is calculated separately on different descriptor blocks or sub-blocks, external variables are added to each descriptor block (or sub-block).

 

By clicking on the 'calculate' button, PCA is performed and the window for analyzing the PCA results is opened.