Abstract
Since the last published update in 2014, the SuperPred webserver has been continuously developed to offer state-of-the-art models for drug classification according to ATC classes and target ...prediction. For the first time, a thoroughly filtered ATC dataset, that is suitable for accurate predictions, is provided along with detailed information on the achieved predictions. This aims to overcome the challenges in comparing different published prediction methods, since performance can vary greatly depending on the training dataset used. Additionally, both ATC and target prediction have been reworked and are now based on machine learning models instead of overall structural similarity, stressing the importance of functional groups for the mechanism of action of small molecule substances. Additionally, the dataset for the target prediction has been extensively filtered and is no longer only based on confirmed binders but also includes non-binding substances to reduce false positives. Using these methods, accuracy for the ATC prediction could be increased by almost 5% to 80.5% compared to the previous version, and additionally the scoring function now offers values which are easily assessable at first glance. SuperPred 3.0 is publicly available without the need for registration at: https://prediction.charite.de/index.php.
Graphical Abstract
Graphical Abstract
SuperPred 3.0: drug classification and target prediction—a machine learning approach.
Kinase inhibitors are important cancer therapeutics. Polypharmacology is commonly observed, requiring thorough target deconvolution to understand drug mechanism of action. Using chemical proteomics, ...we analyzed the target spectrum of 243 clinically evaluated kinase drugs. The data revealed previously unknown targets for established drugs, offered a perspective on the "druggable" kinome, highlighted (non)kinase off-targets, and suggested potential therapeutic applications. Integration of phosphoproteomic data refined drug-affected pathways, identified response markers, and strengthened rationale for combination treatments. We exemplify translational value by discovering SIK2 (salt-inducible kinase 2) inhibitors that modulate cytokine production in primary cells, by identifying drugs against the lung cancer survival marker MELK (maternal embryonic leucine zipper kinase), and by repurposing cabozantinib to treat FLT3-ITD-positive acute myeloid leukemia. This resource, available via the ProteomicsDB database, should facilitate basic, clinical, and drug discovery research and aid clinical decision-making.
The SuperPred web server connects chemical similarity of drug-like compounds with molecular targets and the therapeutic approach based on the similar property principle. Since the first release of ...this server, the number of known compound-target interactions has increased from 7000 to 665,000, which allows not only a better prediction quality but also the estimation of a confidence. Apart from the addition of quantitative binding data and the statistical consideration of the similarity distribution in all drug classes, new approaches were implemented to improve the target prediction. The 3D similarity as well as the occurrence of fragments and the concordance of physico-chemical properties is also taken into account. In addition, the effect of different fingerprints on the prediction was examined. The retrospective prediction of a drug class (ATC code of the WHO) allows the evaluation of methods and descriptors for a well-characterized set of approved drugs. The prediction is improved by 7.5% to a total accuracy of 75.1%. For query compounds with sufficient structural similarity, the web server allows prognoses about the medical indication area of novel compounds and to find new leads for known targets. SuperPred is publicly available without registration at: http://prediction.charite.de.
Here, we present an updated version of CancerResource, freely available without registration at http://bioinformatics.charite.de/care. With upcoming information on target expression and mutations in ...patients' tumors, the need for systems supporting decisions on individual therapy is growing. This knowledge is based on numerous, experimentally validated drug-target interactions and supporting analyses such as measuring changes in gene expression using microarrays and HTS-efforts on cell lines. To enable a better overview about similar drug-target data and supporting information, a series of novel information connections are established and made available as described in the following. CancerResource contains about 91,000 drug-target relations, more than 2000 cancer cell lines and drug sensitivity data for about 50,000 drugs. CancerResource enables the capability of uploading external expression and mutation data and comparing them to the database's cell lines. Target genes and compounds are projected onto cancer-related pathways to get a better overview about how drug-target interactions benefit the treatment of cancer. Features like cellular fingerprints comprising of mutations, expression values and drug-sensitivity data can promote the understanding of genotype to drug sensitivity associations. Ultimately, these profiles can also be used to determine the most effective drug treatment for a cancer cell line most similar to a patient's tumor cells.
Assay interference is an acknowledged problem in high-throughput screening, and pan-assay interference compounds (PAINS) filters are one of a number of approaches that have been suggested for ...identification of potential screening artifacts or frequent hitters. Many studies have highlighted that the unwary usage of these structural alerts should be reconsidered and criticized their extrapolation beyond the applicability domain. A large-scale investigation of the activity profiles and the structural context of PAINS might provide a better assessment of whether this extrapolation is valid. To this end, multiple publicly accessible compound collections were screened, and the PAINS statistics are comprehensively presented and discussed. Next, the promiscuity trends and activity profiles of PAINS were compared with those compounds not matching any PAINS substructures. Overall, PAINS demonstrated higher promiscuity and relatively higher assay hit rates compared with the other compounds. Furthermore, nearly 2000 distinct target–ligand complexes containing PAINS were analyzed, and the interactions were quantified and compared. In more than 50% of the instances, the PAINS atoms participated in interactions more frequently compared with the remaining atoms of the ligand structure. Many PAINS participated in crucial interactions that were often responsible for binding of the ligand, which reaffirms their distinction from those responsible for assay interference. In conclusion, we reinforce that while it is important to employ compound filters to eliminate nonspecific hits, establishing a set of statistically significant and validated PAINS filters is essential to restrain the black-box practice of triaging screening hits matching any of the proposed 480 alerts.
Searching for two-dimensional (2D) structural similarities is a useful tool to identify new active compounds in drug-discovery programs. However, as 2D similarity measures neglect important ...structural and functional features, similarity by 2D might be underestimated. In the present study, we used combined 2D and three-dimensional (3D) similarity comparisons to reveal possible new functions and/or side-effects of known bioactive compounds.
We utilised more than 10,000 compounds from the SuperTarget database with known inhibition values for twelve different anti-cancer targets. We performed all-against-all comparisons resulting in 2D similarity landscapes. Among the regions with low 2D similarity scores are inhibitors of vascular endothelial growth factor receptor (VEGFR) and inhibitors of poly ADP-ribose polymerase (PARP). To demonstrate that 3D landscape comparison can identify similarities, which are untraceable in 2D similarity comparisons, we analysed this region in more detail. This 3D analysis showed the unexpected structural similarity between inhibitors of VEGFR and inhibitors of PARP. Among the VEGFR inhibitors that show similarities to PARP inhibitors was Vatalanib, an oral "multi-targeted" small molecule protein kinase inhibitor being studied in phase-III clinical trials in cancer therapy. An in silico docking simulation and an in vitro HT universal colorimetric PARP assay confirmed that the VEGFR inhibitor Vatalanib exhibits off-target activity as a PARP inhibitor, broadening its mode of action.
In contrast to the 2D-similarity search, the 3D-similarity landscape comparison identifies new functions and side effects of the known VEGFR inhibitor Vatalanib.
Natural products play a significant role in drug discovery and development. Many topological pharmacophore patterns are common between natural products and commercial drugs. A better understanding of ...the specific physicochemical and structural features of natural products is important for corresponding drug development. Several encyclopedias of natural compounds have been composed, but the information remains scattered or not freely available. The first version of the Supernatural database containing ∼ 50,000 compounds was published in 2006 to face these challenges. Here we present a new, updated and expanded version of natural product database, Super Natural II (http://bioinformatics.charite.de/supernatural), comprising ∼ 326,000 molecules. It provides all corresponding 2D structures, the most important structural and physicochemical properties, the predicted toxicity class for ∼ 170,000 compounds and the vendor information for the vast majority of compounds. The new version allows a template-based search for similar compounds as well as a search for compound names, vendors, specific physical properties or any substructures. Super Natural II also provides information about the pathways associated with synthesis and degradation of the natural products, as well as their mechanism of action with respect to structurally similar drugs and their target proteins.
Immune mediated adverse drug reactions (IM-ADRs) remain a significant source of patient morbidity that have more recently been shown to be associated with specific class I and/or II human leukocyte ...antigen (HLA) alleles. Abacavir-induced hypersensitivity syndrome is a CD8+ T cell dependent IM-ADR that is exclusively mediated by HLA-B*57:01. We and others have previously shown that abacavir can occupy the floor of the peptide binding groove of HLA-B*57:01 molecules, increasing the affinity of certain self peptides resulting in an altered peptide-binding repertoire. Here, we have identified another drug, acyclovir, which appears to act in a similar fashion. As with abacavir, acyclovir showed a dose dependent increase in affinity for peptides with valine and isoleucine at their C-terminus. In agreement with the binding studies, HLA-B*57:01 peptide-elution studies performed in the presence of acyclovir revealed an increased number of endogenously bound peptides with a C-terminal isoleucine. Accordingly, we have hypothesized that acyclovir acts by the same mechanism as abacavir, although our data also suggest the overall effect is much smaller: the largest changes of peptide affinity for acyclovir were 2-5 fold, whereas for abacavir this effect was as much as 1000-fold. Unlike abacavir, acyclovir is not known to cause IM-ADRs. We conclude that the modest effect of acyclovir on HLA binding affinity in contrast to the large effect of abacavir is insufficient to trigger a hypersensitivity syndrome. We further support this by functional in vitro studies where acyclovir, unlike abacavir, was unable to produce an increase in IFN-γ upon expansion of HLA-B*57:01+ PBMCs from healthy donors. Using abacavir and acyclovir as examples we therefore propose an in vitro pre-clinical screening strategy, whereby thresholds can be applied to MHC-peptide binding assays to determine the likelihood that a drug could cause a clinically relevant IM-ADR.