ADFR v 1.0 March 1026 (Michel Sanner)

This is the first release of ADFR with a C++ implementation of the Flexibility
Tree and the Solis Wets local search.

Main differences with the version described in the PlOS paper include:
 - new C++ implementation of the scoring function
   - new directional Hydrogen bonding term
   - distance cut off for vdw:8, hb:5, estat:10000, desolv:8

   NOTE: given that the Hydrogen-bond term of the AutoDock energy function is
   	 different form the the one in AutoDock4.2 (which was the one used in
	 the Python implementation reported in the paper), docking results can
	 not be directly compared in terms of energy between the C++ and the
	 Python implementation as small variations in energy are expected in
	 this term.
	 
 - The GA uses 1to1 atom matching rather tha Hungarian matching to compute RMSD
    while clustering the population. This speeds up run time bu a factor of 7
    on the Astex Diverse Set. The default method is now set to 1to1. It can be
    change on the command line using --RMSDMatching hungarian

 - Support has been added for docking ligands covalently bound to the receptor.
   This feature is not yet fully tested and not documented yet.
 
This version performs similarly as the published version on the Astex Diverse 
Set with  a success rate of 70% (61/85) even though the weights should be 
re-calibrated as the HB term changed. (TODO)

This version takes an average of 51 minutes and 31 second to dock a ligand (using the default 50 GA evolutions). Docking time vary greatly though (from 282 second for 1hq2 to 20408(s) for 1ygc) with an average of 3020(s) but a median of 1955(s)making it about 280 times faster than the originally published Python version.

