A docking system usually consists of three main parts, the re presentation of the molecules (e.g. proteins), conformati on l space sampling and last but not least ranking of potential so lutions. Whereas the first two parts can be handled by several different methods the ranking of docking hypotheses is stil l not satisfactory. The main problem is to distinguish between correct and false positive solutions. According to Nussinov et. al[2] some algorithms place correct results on high ranks, but no algor ithms reliably discriminate between correct solutions and false positives. We introduce the IPHE X (IntelligentProteinHypothesisExplorer) system to improve on the scoring problem. It uses rel evance feedback techniques adapted from Query-by-Content ( QbC) systems, especially from the INDI ( IntelligentNavigation in Digital Image databases) system[3]. Fully automatic classification of docking hypotheses is difficult, but a number of algorithms can calculate geometric and physico-chemical fea tur s. They are combined in the scoring function. An optimal set of weights for the individual components can vary between di fferent query proteins. During development those weights c an easily be modified explicitely, but this needs knowledge abo ut the implementation details. Providing an easy to use inte rfac hiding the complex scoring functions from the user a set of hy pothesis can be evaluated and scored easily. The hypotheses are scored from highly non-relevant to highly relevant. Hav ing scored a set of hypothesis the system modifies the weights within the scoring function according to the users feedback . IPHEX is used to improve the scoring of our docking system E LMAR [1, 5]. It also allows to study the influence of e.g. charge, hydrophobicity and geometrical constraints, refle cted by the users feedback to our algorithm. First results sh ow t at this approach can be used to improve the scoring of the E LMAR system. Of course intensive testing and evaluation has to b e done.
[1]
S. Posch,et al.
viwish: a visualization server for protein modelling and docking.
,
1996,
Gene.
[2]
Gerhard Sagerer,et al.
Estimation and filtering of potential protein-protein docking positions
,
1998,
Bioinform..
[3]
Ruth Nussinov,et al.
Principles of docking: An overview of search algorithms and a guide to scoring functions
,
2002,
Proteins.
[4]
F. Kummert,et al.
ElMaR: A Protein Docking System using Flexibility Information
,
2002
.
[5]
Helge J. Ritter,et al.
INDI - intelligent database navigation by interactive and intuitive content-based image retrieval
,
2002,
Proceedings. International Conference on Image Processing.