With MTrees, you can combine several feature trees to form a so-called "model". You may think of it as the overlay of a bunch of molecular sketches on overhead transparencies on photographic paper, so that the most common parts of the compounds are reflected in lighter greys on the paper. If you combine this information with activitiy data, multiple feature tree "models" become connectivitiy-information preserving "2D-pharmacophore" models. The particular advantage of models is that they are themselves feature trees and can thus directly be used for screening, searching, and more. "Model" generation and handling is facilitated with HTSview, our data analysis toolbox and graphical user front-end to feature tree based methods.