Robert Preissner

SuperTarget and Matador: resources for exploring drug-target relationships

The molecular basis of drug action is often not well understood. This is partly because the very abundant and diverse information generated in the past decades on drugs is hidden in millions of medical articles or textbooks. Therefore, we developed a one-stop data warehouse, SuperTarget that integrates drug-related information about medical indication areas, adverse drug effects, drug metabolization, pathways and Gene Ontology terms of the target proteins. An easy-to-use query interface enables the user to pose complex queries, for example to find drugs that target a certain pathway, interacting drugs that are metabolized by the same cytochrome P450 or drugs that target the same protein but are metabolized by different enzymes. Furthermore, we provide tools for 2D drug screening and sequence comparison of the targets. The database contains more than 2500 target proteins, which are annotated with about 7300 relations to 1500 drugs; the vast majority of entries have pointers to the respective literature source. A subset of these drugs has been annotated with additional binding information and indirect interactions and is available as a separate resource called Matador. SuperTarget and Matador are available at http://insilico.charite.de/supertarget and http://matador.embl.de

PROMISCUOUS: a database for network-based drug-repositioning

The procedure of drug approval is time-consuming, costly and risky. Accidental findings regarding multi-specificity of approved drugs led to block-busters in new indication areas. Therefore, the interest in systematically elucidating new areas of application for known drugs is rising. Furthermore, the knowledge, understanding and prediction of so-called off-target effects allow a rational approach to the understanding of side-effects. With PROMISCUOUS we provide an exhaustive set of drugs (25 000), including withdrawn or experimental drugs, annotated with drug–protein and protein–protein relationships (21 500/104 000) compiled from public resources via text and data mining including manual curation. Measures of structural similarity for drugs as well as known side-effects can be easily connected to protein–protein interactions to establish and analyse networks responsible for multi-pharmacology. This network-based approach can provide a starting point for drug-repositioning. PROMISCUOUS is publicly available at http://bioinformatics.charite.de/promiscuous.

Polymorphic Cytochrome P450 Enzymes (CYPs) and Their Role in Personalized Therapy

The cytochrome P450 (CYP) enzymes are major players in drug metabolism. More than 2,000 mutations have been described, and certain single nucleotide polymorphisms (SNPs) have been shown to have a large impact on CYP activity. Therefore, CYPs play an important role in inter-individual drug response and their genetic variability should be factored into personalized medicine. To identify the most relevant polymorphisms in human CYPs, a text mining approach was used. We investigated their frequencies in different ethnic groups, the number of drugs that are metabolized by each CYP, the impact of CYP SNPs, as well as CYP expression patterns in different tissues. The most important polymorphic CYPs were found to be 1A2, 2D6, 2C9 and 2C19. Thirty-four common allele variants in Caucasians led to altered enzyme activity. To compare the relevant Caucasian SNPs with those of other ethnicities a search in 1,000 individual genomes was undertaken. We found 199 non-synonymous SNPs with frequencies over one percent in the 1,000 genomes, many of them not described so far. With knowledge of frequent mutations and their impact on CYP activities, it may be possible to predict patient response to certain drugs, as well as adverse side effects. With improved availability of genotyping, our data may provide a resource for an understanding of the effects of specific SNPs in CYPs, enabling the selection of a more personalized treatment regimen.

SuperTarget goes quantitative: update on drug–target interactions

There are at least two good reasons for the on-going interest in drug–target interactions: first, drug-effects can only be fully understood by considering a complex network of interactions to multiple targets (so-called off-target effects) including metabolic and signaling pathways; second, it is crucial to consider drug-target-pathway relations for the identification of novel targets for drug development. To address this on-going need, we have developed a web-based data warehouse named SuperTarget, which integrates drug-related information associated with medical indications, adverse drug effects, drug metabolism, pathways and Gene Ontology (GO) terms for target proteins. At present, the updated database contains >6000 target proteins, which are annotated with >330 000 relations to 196 000 compounds (including approved drugs); the vast majority of interactions include binding affinities and pointers to the respective literature sources. The user interface provides tools for drug screening and target similarity inclusion. A query interface enables the user to pose complex queries, for example, to find drugs that target a certain pathway, interacting drugs that are metabolized by the same cytochrome P450 or drugs that target proteins within a certain affinity range. SuperTarget is available at http://bioinformatics.charite.de/supertarget.

Voronoi Cell: New Method for Allocation of Space among Atoms: Elimination of Avoidable Errors in Calculation of Atomic Volume and Density

In computing the volume occupied by atoms and the density in proteins, one is faced with the problem of intersecting spheres. To estimate either, the space between the atoms has to be divided according to the location of the atoms relative to each other. Various methods, based on Voronois idea of approximating the atomic space by polyhedra, have been proposed for this purpose. Comparing procedures concerned with the allocation of space among distinct atoms, we observe different partitionings of space, with deviations of more than 100% for particular atoms. Furthermore, we find that the separating planes of different Voronoi procedures do not meet the intersection circles of covalently linked atoms. This leads to a misallocation of space of up to 7% for atom pairs that largely differ in atomic size (e.g., C—H). Several algorithms are negatively affected by small unallocated polyhedra (“vertex error”). These effects are cumulative for a small protein up to a loss of some 60 Å3 of total volume, which would correspond to the deletion of one complete residue. To overcome these errors, instead of using dividing planes between the atoms, we use curved surfaces, defined as the set of those geometrical loci with equal orthogonal distance to the surfaces of the van der Waals spheres under consideration. The proposed dividing surface meets not only the intersection circle of the two van der Waals spheres but also the intersection circle of the two spheres enlarged by an arbitrary value (e.g., radius of water). This hyperbolic surface enveloping the Voronoi cell can be easily constructed and offers the following advantages: no misallocation of volume for atoms of different size, no vertex error, geometrically reasonable allocation of the volume among atoms, avoidance of discontinuities between neighboring atoms, and improved applicability to water‐accessible protein surfaces.

SuperPred: drug classification and target prediction

The drug classification scheme of the World Health Organization (WHO) [Anatomical Therapeutic Chemical (ATC)-code] connects chemical classification and therapeutic approach. It is generally accepted that compounds with similar physicochemical properties exhibit similar biological activity. If this hypothesis holds true for drugs, then the ATC-code, the putative medical indication area and potentially the medical target should be predictable on the basis of structural similarity. We have validated that the prediction of the drug class is reliable for WHO-classified drugs. The reliability of the predicted medical effects of the compounds increases with a rising number of (physico-) chemical properties similar to a drug with known function. The web-server translates a user-defined molecule into a structural fingerprint that is compared to about 6300 drugs, which are enriched by 7300 links to molecular targets of the drugs, derived through text mining followed by manual curation. Links to the affected pathways are provided. The similarity to the medical compounds is expressed by the Tanimoto coefficient that gives the structural similarity of two compounds. A similarity score higher than 0.85 results in correct ATC prediction for 81% of all cases. As the biological effect is well predictable, if the structural similarity is sufficient, the web-server allows prognoses about the medical indication area of novel compounds and to find new leads for known targets. Availability: the system is freely accessible at http://bioinformatics.charite.de/superpred. SuperPred can be obtained via a Creative Commons Attribution Noncommercial-Share Alike 3.0 License.

SuperLooper—a prediction server for the modeling of loops in globular and membrane proteins

SuperLooper provides the first online interface for the automatic, quick and interactive search and placement of loops in proteins (LIP). A database containing half a billion segments of water-soluble proteins with lengths up to 35 residues can be screened for candidate loops. A specified database containing 180 000 membrane loops in proteins (LIMP) can be searched, alternatively. Loop candidates are scored based on sequence criteria and the root mean square deviation (RMSD) of the stem atoms. Searching LIP, the average global RMSD of the respective top-ranked loops to the original loops is benchmarked to be <2 Å, for loops up to six residues or <3 Å for loops shorter than 10 residues. Other suitable conformations may be selected and directly visualized on the web server from a top-50 list. For user guidance, the sequence homology between the template and the original sequence, proline or glycine exchanges or close contacts between a loop candidate and the remainder of the protein are denoted. For membrane proteins, the expansions of the lipid bilayer are automatically modeled using the TMDET algorithm. This allows the user to select the optimal membrane protein loop concerning its relative orientation to the lipid bilayer. The server is online since October 2007 and can be freely accessed at URL: http://bioinformatics.charite.de/superlooper/

ProTox: a web server for the in silico prediction of rodent oral toxicity

Animal trials are currently the major method for determining the possible toxic effects of drug candidates and cosmetics. In silico prediction methods represent an alternative approach and aim to rationalize the preclinical drug development, thus enabling the reduction of the associated time, costs and animal experiments. Here, we present ProTox, a web server for the prediction of rodent oral toxicity. The prediction method is based on the analysis of the similarity of compounds with known median lethal doses (LD50) and incorporates the identification of toxic fragments, therefore representing a novel approach in toxicity prediction. In addition, the web server includes an indication of possible toxicity targets which is based on an in-house collection of protein–ligand-based pharmacophore models (‘toxicophores’) for targets associated with adverse drug reactions. The ProTox web server is open to all users and can be accessed without registration at: http://tox.charite.de/tox. The only requirement for the prediction is the two-dimensional structure of the input compounds. All ProTox methods have been evaluated based on a diverse external validation set and displayed strong performance (sensitivity, specificity and precision of 76, 95 and 75%, respectively) and superiority over other toxicity prediction tools, indicating their possible applicability for other compound classes.

CancerResource: a comprehensive database of cancer-relevant proteins and compound interactions supported by experimental knowledge

During the development of methods for cancer diagnosis and treatment, a vast amount of information is generated. Novel cancer target proteins have been identified and many compounds that activate or inhibit cancer-relevant target genes have been developed. This knowledge is based on an immense number of experimentally validated compound–target interactions in the literature, and excerpts from literature text mining are spread over numerous data sources. Our own analysis shows that the overlap between important existing repositories such as Comparative Toxicogenomics Database (CTD), Therapeutic Target Database (TTD), Pharmacogenomics Knowledge Base (PharmGKB) and DrugBank as well as between our own literature mining for cancer-annotated entries is surprisingly small. In order to provide an easy overview of interaction data, it is essential to integrate this information into a single, comprehensive data repository. Here, we present CancerResource, a database that integrates cancer-relevant relationships of compounds and targets from (i) our own literature mining and (ii) external resources complemented with (iii) essential experimental and supporting information on genes and cellular effects. In order to facilitate an overview of existing and supporting information, a series of novel information connections have been established. CancerResource addresses the spectrum of research on compound–target interactions in natural sciences as well as in individualized medicine; CancerResource is available at: http://bioinformatics.charite.de/cancerresource/.

Voronoia: analyzing packing in protein structures

The packing of protein atoms is an indicator for their stability and functionality, and applied in determining thermostability, in protein design, ligand binding and to identify flexible regions in proteins. Here, we present Voronoia, a database of atomic-scale packing data for protein 3D structures. It is based on an improved Voronoi Cell algorithm using hyperboloid interfaces to construct atomic volumes, and to resolve solvent-accessible and -inaccessible regions of atoms. The database contains atomic volumes, local packing densities and interior cavities calculated for 61 318 biological units from the PDB. A report for each structure summarizes the packing by residue and atom types, and lists the environment of interior cavities. The packing data are compared to a nonredundant set of structures from SCOP superfamilies. Both packing densities and cavities can be visualized in the 3D structures by the Jmol plugin. Additionally, PDB files can be submitted to the Voronoia server for calculation. This service performs calculations for most full-atomic protein structures within a few minutes. For batch jobs, a standalone version of the program with an optional PyMOL plugin is available for download. The database can be freely accessed at: http://bioinformatics.charite.de/voronoia.