OPUS 4 | Universitätspublikationen

Molecular similarity for machine learning in drug development : poster presentation (2008)

Matthias Rupp Ewgenij Proschak Gisbert Schneider

Poster presentation In pharmaceutical research and drug development, machine learning methods play an important role in virtual screening and ADME/Tox prediction. For the application of such methods, a formal measure of similarity between molecules is essential. Such a measure, in turn, depends on the underlying molecular representation. Input samples have traditionally been modeled as vectors. Consequently, molecules are represented to machine learning algorithms in a vectorized form using molecular descriptors. While this approach is straightforward, it has its shortcomings. Amongst others, the interpretation of the learned model can be difficult, e.g. when using fingerprints or hashing. Structured representations of the input constitute an alternative to vector based representations, a trend in machine learning over the last years. For molecules, there is a rich choice of such representations. Popular examples include the molecular graph, molecular shape and the electrostatic field. We have developed a molecular similarity measure defined directly on the (annotated) molecular graph, a long-standing established topological model for molecules. It is based on the concepts of optimal atom assignments and iterative graph similarity. In the latter, two atoms are considered similar if their neighbors are similar. This recursive definition leads to a non-linear system of equations. We show how to iteratively solve these equations and give bounds on the computational complexity of the procedure. Advantages of our similarity measure include interpretability (atoms of two molecules are assigned to each other, each pair with a score expressing local similarity; this can be visualized to show similar regions of two molecules and the degree of their similarity) and the possibility to introduce knowledge about the target where available. We retrospectively tested our similarity measure using support vector machines for virtual screening on several pharmaceutical and toxicological datasets, with encouraging results. Prospective studies are under way.

Ideenschmiede mit Praxisbezug : fünf Jahre Beilstein-Stiftungsprofessur für Chemieinformatik (2007)

Gisbert Schneider

Eine Stiftungsprofessur ermöglicht die konzentrierte Forschung auf einem speziellen Fachgebiet und schafft den notwendigen Freiraum, Neues zu erproben. Insbesondere kann sie dazu dienen, Brücken zwischen Disziplinen zu errichten. Mit diesem Ziel wurde vor fünf Jahren die Beilstein-Stiftungsprofessur für Chemieinformatik an der Johann Wolfgang Goethe-Universität eingerichtet. Gefördert von dem in Frankfurt am Main ansässigen Beilstein-Institut zur Förderung der Chemischen Wissenschaften, wurde sie in enger Zusammenarbeit mit dem Institut für Organische Chemie und Chemische Biologie unter der Federführung von Prof. Dr. Michael Göbel konzipiert. Nachdem die Förderperiode von fünf Jahren im März 2007 ausgelaufen war, ist die Stiftungsprofessur nahtlos in den ordentlichen Universitätsbetrieb übernommen worden. Dies gibt Anlass, ein Fazit zu ziehen.

Kernel learning for ligand-based virtual screening:discovery of a new PPARgamma agonist (2010)

Matthias Rupp Timon Schroeter Ramona Steri Ewgenij Proschak Katja Hansen Heiko Zettl Oliver Rau Manfred Schubert-Zsilavecz Klaus-Robert Müller Gisbert Schneider

Poster presentation at 5th German Conference on Cheminformatics: 23. CIC-Workshop Goslar, Germany. 8-10 November 2009 We demonstrate the theoretical and practical application of modern kernel-based machine learning methods to ligand-based virtual screening by successful prospective screening for novel agonists of the peroxisome proliferator-activated receptor gamma (PPARgamma) [1]. PPARgamma is a nuclear receptor involved in lipid and glucose metabolism, and related to type-2 diabetes and dyslipidemia. Applied methods included a graph kernel designed for molecular similarity analysis [2], kernel principle component analysis [3], multiple kernel learning [4], and, Gaussian process regression [5]. In the machine learning approach to ligand-based virtual screening, one uses the similarity principle [6] to identify potentially active compounds based on their similarity to known reference ligands. Kernel-based machine learning [7] uses the "kernel trick", a systematic approach to the derivation of non-linear versions of linear algorithms like separating hyperplanes and regression. Prerequisites for kernel learning are similarity measures with the mathematical property of positive semidefiniteness (kernels). The iterative similarity optimal assignment graph kernel (ISOAK) [2] is defined directly on the annotated structure graph, and was designed specifically for the comparison of small molecules. In our virtual screening study, its use improved results, e.g., in principle component analysis-based visualization and Gaussian process regression. Following a thorough retrospective validation using a data set of 176 published PPARgamma agonists [8], we screened a vendor library for novel agonists. Subsequent testing of 15 compounds in a cell-based transactivation assay [9] yielded four active compounds. The most interesting hit, a natural product derivative with cyclobutane scaffold, is a full selective PPARgamma agonist (EC50 = 10 ± 0.2 microM, inactive on PPARalpha and PPARbeta/delta at 10 microM). We demonstrate how the interplay of several modern kernel-based machine learning approaches can successfully improve ligand-based virtual screening results.

Predicting olfactory receptor neuron responses from odorant structure (2007)

Michael Schmuker Marien De Bruyne Melanie Hähnel Gisbert Schneider

Background Olfactory receptors work at the interface between the chemical world of volatile molecules and the perception of scent in the brain. Their main purpose is to translate chemical space into information that can be processed by neural circuits. Assuming that these receptors have evolved to cope with this task, the analysis of their coding strategy promises to yield valuable insight in how to encode chemical information in an efficient way. Results We mimicked olfactory coding by modeling responses of primary olfactory neurons to small molecules using a large set of physicochemical molecular descriptors and artificial neural networks. We then tested these models by recording in vivo receptor neuron responses to a new set of odorants and successfully predicted the responses of five out of seven receptor neurons. Correlation coefficients ranged from 0.66 to 0.85, demonstrating the applicability of our approach for the analysis of olfactory receptor activation data. The molecular descriptors that are best-suited for response prediction vary for different receptor neurons, implying that each receptor neuron detects a different aspect of chemical space. Finally, we demonstrate that receptor responses themselves can be used as descriptors in a predictive model of neuron activation. Conclusions The chemical meaning of molecular descriptors helps understand structure-response relationships for olfactory receptors and their 'receptive fields'. Moreover, it is possible to predict receptor neuron activation from chemical structure using machine-learning techniques, although this is still complicated by a lack of training data.

Optimized Particle Swarm Optimization (OPSO) and its application to artificial neural network training (2006)

Michael Meissner Michael Schmuker Gisbert Schneider

Background: Particle Swarm Optimization (PSO) is an established method for parameter optimization. It represents a population-based adaptive optimization technique that is influenced by several "strategy parameters". Choosing reasonable parameter values for the PSO is crucial for its convergence behavior, and depends on the optimization task. We present a method for parameter meta-optimization based on PSO and its application to neural network training. The concept of the Optimized Particle Swarm Optimization (OPSO) is to optimize the free parameters of the PSO by having swarms within a swarm. We assessed the performance of the OPSO method on a set of five artificial fitness functions and compared it to the performance of two popular PSO implementations. Results: Our results indicate that PSO performance can be improved if meta-optimized parameter sets are applied. In addition, we could improve optimization speed and quality on the other PSO methods in the majority of our experiments. We applied the OPSO method to neural network training with the aim to build a quantitative model for predicting blood-brain barrier permeation of small organic molecules. On average, training time decreased by a factor of four and two in comparison to the other PSO methods, respectively. By applying the OPSO method, a prediction model showing good correlation with training-, test- and validation data was obtained. Conclusion: Optimizing the free parameters of the PSO method can result in performance gain. The OPSO approach yields parameter combinations improving overall optimization performance. Its conceptual simplicity makes implementing the method a straightforward task.

Distance phenomena in high-dimensional chemical descriptor spaces : consequences for similarity-based approaches (2009)

Matthias Rupp Gisbert Schneider

SQUIRRELnovo : de novo design of a PPARalpha agonist by bioisosteric replacement (2009)

Ewgenij Proschak Kerstin Sander Heiko Zettl Yusuf Tanrikulu Petra Schneider Oliver Rau Holger Stark Manfred Schubert-Zsilavecz Gisbert Schneider

Shape complementarity is a compulsory condition for molecular recognition [1]. In our 3D ligand-based virtual screening approach called SQUIRREL, we combine shape-based rigid body alignment [2] with fuzzy pharmacophore scoring [3]. Retrospective validation studies demonstrate the superiority of methods which combine both shape and pharmacophore information on the family of peroxisome proliferator-activated receptors (PPARs). We demonstrate the real-life applicability of SQUIRREL by a prospective virtual screening study, where a potent PPARalpha agonist with an EC50 of 44 nM and 100-fold selectivity against PPARgamma has been identified. SQUIRREL molecular superposition is based on a graph-matching routine [4] and allows partial matching. We used this advantage for searching for bioisosteric replacement suggestions in a database of molecular fragments derived from a collection of drug-like compounds [5]. The bioisosteric groups suggested by our tool SQURRELnovo, can be used for ligand-based de novo design by a human expert. Using the fibrate derivative GW590735 [6] as query, we designed a novel lead structure by substitution of the acidic head group and hydrophobic tail. The synthesis and following testing in a cell-based reporter gene assay [7,8] revealed that the designed structure activates PPARalpha with an EC50 of 510 nM.

PocketGraph : graph representation of binding site volumes (2009)

Martin Weisel Jan M. Kriegl Gisbert Schneider

The representation of small molecules as molecular graphs [1] is a common technique in various fields of cheminformatics. This approach employs abstract descriptions of topology and properties for rapid analyses and comparison. Receptor-based methods in contrast mostly depend on more complex representations impeding simplified analysis and limiting the possibilities of property assignment. In this study we demonstrate that ligand-based methods can be applied to receptor-derived binding site analysis. We introduce the new method PocketGraph that translates representations of binding site volumes into linear graphs and enables the application of graph-based methods to the world of protein pockets. The method uses the PocketPicker [2] algorithm for characterization of binding site volumes and employs a Growing Neural Gas [3] procedure to derive graph representations of pocket topologies. Self-organizing map (SOM) projections revealed a limited number of pocket topologies. We argue that there is only a small set of pocket shapes realized in the known ligand-receptor complexes.

PhAST : pharmacophore alignment search tool (2009)

Volker Hähnke Bettina Petra Hofmann Ewgenij Proschak Dieter Steinhilber Gisbert Schneider

We developed the Pharmacophore Alignment Search Tool (PhAST), a text-based technique for rapid hit and lead structure searching in large compound databases. For each molecule, a two-dimensional graph of potential pharmacophoric points (PPPs) is created, which has an identical topology as the original molecule with implicit hydrogen atoms. Each vertex is coloured by a symbol representing the corresponding PPP. The vertices of the graph are canonically labelled [1]. The symbols associated with the vertices are combined to a so-called PhAST-Sequence beginning with the vertex with the lowest canonical label. Due to the canonical labelling the created PhAST-Sequence is characteristic for each molecule. For similarity assessment, PhAST-Sequences are compared using the sequence identity in their global pairwise alignment [2]. The alignment score lies between 0 (no similarity) and 1 (identical PhAST-Sequences). In order to use global pairwise sequence alignment, a score matrix for pharmacophoric symbols was developed and gap penalties were optimized. PhAST performed comparably and sometimes superior to other similarity search tools (CATS2D [3], MOE pharmacophore quadruples [4]) in retrospective virtual screenings using the COBRA [5] collection of drugs and lead structures. Most importantly, the PhAST alignment technique allows for the computation of significance estimates that help prioritize a virtual hit list.

Pseudoreceptor-based pocket selection in a molecular dynamics simulation of the histamine H4 receptor (2009)

Tim Werner Tim Geppert Yusuf Tanrikulu Ewgenij Proschak Gisbert Schneider

There is a renewed interest in pseudoreceptor models which enable computational chemists to bridge the gap of ligand- and receptor-based drug design [1]. We developed a pseudoreceptor model for the histamine H4 receptor (H4R) based on five potent antagonists representing different chemotypes. Here we present the selection of potential ligand binding pockets that occur during molecular dynamics (MD) simulations of a homology-based receptor model. We present a method for prioritizing receptor models according to their match with the consensus ligand-binding mode represented by the pseudoreceptor. In this way, ligand information can be transferred to receptor-based modelling. We use Geometric Hashing to match three-dimensional points in Cartesion space [2]. This allows for the rapid translation- and rotation-free comparison of atom coordinates, which also permits partial matching. The only prerequisite is a hash table, which uses distance triplets as hash keys. Each time a distance triplet occurring in the candidate point set which corresponds to an existing key, the match is represented by a vote of the respective key. Finally, the global match of both point sets can be easily extracted by selection of voted distance triplets. The results revealed a preferred ligand-binding pocket in H4R, which would not have been identified using an unrefined homology model of the protein. The key idea was to rely on ligand information by pseudoreceptor modelling.

Universitätspublikationen

Refine

Author

Year of publication

Language

Institute

20 search hits