Priv.-Doz. Dr. Michael Hutter
Telephone: ++49 +681 302 70703
Fax: ++49 +681 302 70702
Postal address: Center for Bioinformatics
Campus Gebäude E2 1, R. 313
Dipl. Chem. Univ. (1993) University of Erlangen (Germany)
Dr. rer. nat. (1997) University of Erlangen (Germany)
Australian Research Council Postdoctoral Fellow (1997-1998)
University of Sydney (Australia)
Max-Planck Institute of Biophysics, Frankfurt (Germany) (1998-2003)
Center of Bioinformatics, Saarland University (since 2003)
PD (2007) Saarland University
list of publications
teachingcurrent and past terms
special lecture (master program): Modern Methods in Drug Discovery (winter term)
Softwarewerkzeuge der Bioinformatik (Wintersemester)
Ringvorlesung: Einführung in die Bioinformatik (Wintersemester)
Computational Chemistry (Sommersemester)
Seminar zu "Molecular Modelling"
Schlüsselqualifikationen: Wissenschaftliches Publizieren
Current research areas:
Bioisosteric similarity for virtual screening
Along the various concepts of similarity of molecules, e.g. those based on maximum common substructure or fingerprints, this approach makes use of reoccurring chemical modification in successful drugs. The method transfers the concept of homology of protein sequences to the SMILES notation of molecules.
M. Krier, M.C. Hutter
Bioisosteric Similarity of Molecules Based on Structural Alignment and Observed Chemical Replacements in Drugs
J. Chem. Inf. Model. 49 (2009) 1280-1297.
In silico prediction of molecular properties for virtual screening
Virtual screening is a prominent problem in the preclinical development of new pharmaceutical substances. Since it is not practicable to synthesize and test all compounds of virtual libraries, fast computer-based methods for screening are wanted. Ideally these should filter out unsuitable molecules and predict drug-like compounds.
In this context a drug-likeliness index has been developed that assigns a numerical score to a given molecule represented as 2D-representation. This index was derived on the basis of the statistical distribution of atom-pair combinations in known drugs and nondrugs.
O. Frings, M.C. Hutter
Publication in preparation
Drug-likeness: Drug/Nondrug separation
Besides the computation of the individual ADMETox properties of substances, the challenge remains to predict the suitability of a molecule being a drug-like compound pre se.
Therefore the statistical distribution of atom-pair combinations in known drugs and nondrugs has been used to derive a so-called drug-likeliness index. Using decision trees to elucidate the most significant descriptors for partitioning drugs from nondrugs, it has been shown that this index supersedes other common drug-likeness criteria.
Separating Drugs from Nondrugs: A Statistical Approach Using Atom Pair Distributions
J. Chem. Inf. Model. 47 (2007) 186-194.
Determination of hERG channel blockade and QT-interval prolongation
Cardiac arrhythmia due to drug induced QT-interval prolongation is a frequent side effect that has let to the withdrawal of many pharmaceutical drugs. Most medications that lead to QT-interval prolongation are also inhibitors of the hERG-potassium channel. Prediction of hERG-channel binding affinity is thus also addressed in the preclinical development. We have derived a so-called pharmacophoric SMARTS-string whereby 71% of all compounds can be classified according to their potential risk.
M. Gepp, M.C. Hutter
Determination of hERG Channel Blockers Using a Decision Tree
Bioorg. Med. Chem. 14 (2006) 5325-5332.
Prediction of blood-brain barrier permeation
The blood-brain barrier (BBB) is a membranic structure that separates the brain and the nervous system from the systemic blood flow. On top of the intestinal absorption the blood-brain permeability of potential drugs must be addressed during development. In silico methods replacing experimental models are particularly wanted here.
We have presented quantitative and qualitative methods for the prediction of the blood-brain barrier permeability.
C. Andres, M.C. Hutter
CNS Permeability of Drugs Predicted by a Decision Tree
QSAR Comb. Sci. 25 (2006) 305-309.
Quantum chemically derived descriptors for QSAR applications and virtual screening
Semi-empirical quantum mechanics allows the fast computation of molecular descriptors derived from the electronic wave function. Thus molecular properties such as the ionization potential, the reactivity and the electrostatic potential can be used in QSAR or machine learning algorithms.
According descriptor from AM1 calculations have been used by us for the prediction of enzyme inhibitors and the blood-brain barrier permeability.
Mg parameters for AM1
More: Mg parameters for AM1
Former research areas:
photolytic cleavage of "caged" compounds
catalytic phosphoryl transfer in kinases
cAMP-dependent protein kinase
Photosynthetic reaction centres in bacteria
DNA and Mutagens
Molecular orbital methods for solvation effects
Combined QM/MM method for environment effects and drug design