Removal of the need for Data Reduction

Traditional data analysis technologies require a 'one vector per example' approach (single table approach). Thereby all information about a particular example (e.g. a molecule) has to be reduced to a single row of a table. This reduction step requires 'manual data mining' by the researcher who has to decide upfront which information is relevant and which is not. This is especially problematic if

- there is a variable number of descriptors in examples, e.g.

    - number of moieties in a compound
    - secondary structure elements in ORFs
    - ...
- relations among descriptors are important, e.g.

    - order of secondary structure elements
    - links between compound substructures
    - ...

PharmaDM solutions generate descriptors automatically. A priori data reduction is no longer required.