Supplementary material for the paper on EvDTree:

The 21 possible amino acid substitutions at any position of a given protein structure can be scored using the decision trees calculated by EvDTree at the follwoing URL: http://abcis.cbs.cnrs.fr/evdtree/evdtree.html. The input protein structure should be formatted as a PDB file. Scores are obtained from decision trees optimized either on a general data set or on a limited set of small disulfide rich proteins.

The structure-dependent substitution profiles obtained by the EvDTree algorithm based on decision tree induction are available in MS Word format EvDTree_profiles.doc or as a PDF document EvDTree_profiles.PDF. The table inside this file has the following format:

The first column indicates the native amino acid type, tree node depth and tree node index.

The next 9 columns describe the structural environment associated to the considered node of the decision tree. The meaning of each structural variable is given in the manuscript.

The last 21 columns indicate each substitution score associated with the 21 amino acid types (C refers to half-cystines while J refers to free cysteines).

Note that the root nodes with depth 0 and index 0 correspond to the substitution scores averaged over all possible structural environments for given native and mutant amino acid types. These "root" substitution scores are called EvDTree0 in the manuscript.

Structure-dependent substitution profiles optimized on a limited set of small disulfide rich proteins are also available in MS Word format EvDTreeDS_profiles.doc or as a PDF document EvDTreeDS_profiles.PDF.

Last modified: Tue Nov 23 12:28:09 CET 2004