A5015053198- Computational Drug Discovery Methods
- Machine Learning in Materials Science
- Machine Learning and Algorithms
- Advanced Graph Neural Networks
- Analytical Chemistry and Chromatography
- Click Chemistry and Applications
- Multiple Myeloma Research and Treatments
- Biosimilars and Bioanalytical Methods
- Protein Degradation and Inhibitors
- Peptidase Inhibition and Analysis
- Chemical Synthesis and Analysis
- Microbial Natural Products and Biosynthesis
- Pharmacogenetics and Drug Metabolism
- Cell Image Analysis Techniques
- SARS-CoV-2 detection and testing
- Analytical chemistry methods development
- Environmental Toxicology and Ecotoxicology
- Enzyme function and inhibition
- Machine Learning and Data Classification
- Student Assessment and Feedback
- History and advancements in chemistry
- Bioinformatics and Genomic Networks
- Immune cells in cancer
- Graph Theory and Algorithms
- Data Visualization and Analytics
Boehringer Ingelheim (Germany)
2021-2025
Boehringer Ingelheim (United States)
2022
TU Dortmund University
2016-2021
Boehringer Ingelheim (India)
2021
Federated multipartner machine learning has been touted as an appealing and efficient method to increase the effective training data volume thereby predictivity of models, particularly when generation is resource-intensive. In landmark MELLODDY project, indeed, each ten pharmaceutical companies realized aggregated improvements on its own classification or regression models through federated learning. To this end, they leveraged a novel implementation extending multitask across partners,...
The era of big data is influencing the way how rational drug discovery and development bioactive molecules performed versatile tools are needed to assist in molecular design workflows. Scaffold Hunter a flexible visual analytics framework for analysis chemical compound combines techniques from several fields such as mining information visualization. allows analyzing high-dimensional an interactive fashion, combining intuitive visualizations with automated methods including clustering...
With the increase in applications of machine learning methods drug design and related fields, challenge designing sound test sets becomes more prominent. The goal this is to have a realistic split chemical structures (compounds) between training, validation set such that performance on meaningful infer prospective application. This by its own very interesting relevant, but even complex federated approach where multiple partners jointly train model under privacy-preserving conditions must not...
Protein ligand interaction fingerprints are a powerful approach for the analysis and assessment of docking poses to improve performance in virtual screening. In this study, novel fingerprint (PADIF, protein per atom score contributions derived fingerprint) is presented which was specifically designed utilising GOLD scoring functions' together with specific scheme. This allows incorporation known protein-ligand complex structures target-specific scoring. Unlike many other methods, uses...
ADME (Absorption, Distribution, Metabolism, Excretion) properties are key parameters to judge whether a drug candidate exhibits desired pharmacokinetic (PK) profile. In this study, we tested multi-task machine learning (ML) models predict and animal PK endpoints trained on in-house data generated at Boehringer Ingelheim. Models were evaluated both the design stage of compound (i.e., no experimental test compounds available) testing when particular assay would be conducted earlier assays may...
ADME (Absorption, Distribution, Metabolism, Excretion) properties are key parameters to judge whether a drug candidate exhibits desired pharmacokinetic (PK) profile. In this study, we tested multi-task machine learning (ML) models predict and animal PK endpoints trained on in-house data generated at Boehringer Ingelheim. Models were evaluated both the design stage of compound (i. e., no experimental test compounds available) testing when particular assay would be conducted earlier assays may...
ABSTRACT A successful drug needs to combine several properties including high potency and good pharmacokinetic (PK) sustain efficacious plasma concentration over time. To estimate required doses for preclinical animal efficacy models or the clinics, in vivo PK studies need be conducted. Although prediction of ADME compounds using machine learning (ML) based on chemical structures is well established discovery, complete concentration–time profiles has only recently gained attention. In this...
Federated multi-partner machine learning can be an appealing and efficient method to increase the effective training data volume thereby predictivity of models, particularly when generation is resource intensive. In landmark MELLODDY project, each ten pharmaceutical companies realized aggregated improvements on its own classification and/or regression models through federated learning. To this end, they leveraged a novel implementation extending multi-task across partners, platform audited...
Abstract A common issue during drug design and development is the discovery of novel scaffolds for protein targets. On one hand chemical space purchasable compounds rather limited; on other artificially generated molecules suffer from a grave lack accessibility in practice. Therefore, we virtual library small which are synthesizable educts, called CH I PMUNK (CHemically feasible In silico Public Molecular UNiverse Knowledge base). Altogether, covers over 95 million encompasses regions that...
ABSTRACT A successful drug needs to combine several properties including high potency and good pharmacokinetic (PK) sustain efficacious plasma concentration over time. To estimate required doses for preclinical animal efficacy models or the clinics, in vivo PK studies need be conducted. While prediction of ADME compounds using Machine Learning (ML) based on chemical structures is well established discovery, complete concentration-time profiles has only recently gained attention. In this...
Machine learning models predicting the bioactivity of chemical compounds belong nowadays to standard tools cheminformaticians and computational medicinal chemists. Multi-task federated are promising machine approaches that allow privacy-preserving usage large amounts data from diverse sources, which is crucial for achieving good generalization high-performance results. Using large, real world sets six pharmaceutical companies, here we investigate different strategies averaging weighted task...
As training volume increases predictive model quality, leveraging existing external data sources holds the promise of time- and cost-efficiency. In a drug discovery setting, pharmaceutical companies all own substantial but confidential datasets. The MELLODDY project develops privacy-preserving federated machine learning solution deploys it at an unprecedented scale (more than 100,000 tasks across ten major companies), while ensuring security privacy each partner’s sensitive data. Each...
Chemical similarity between two molecules is a fundamental concept in cheminformatics and structure comparison therefore an often applied important task. Structural used, e.g., to predict biological activities or analyze molecular datasets. One approach for the identification of chemical based on graph representation molecules, because molecule can intuitively be interpreted as structure. In this article we focus algorithms calculation representation, which expressed maximum common...
Knowledge about interrelationships between different proteins is crucial in fundamental research for the elucidation of protein networks and pathways. Furthermore, it especially critical chemical biology to identify further key regulators a disease take advantage polypharmacology effects. Here, we present new concept that combines scaffold-based analysis bioactivity data with subsequent screening novel inhibitors target interest. The initial revealed flavone-like scaffold can be found...
Macrophage polarization critically contributes to a multitude of human pathologies. Hence, modulating macrophage is promising approach with enormous therapeutic potential. Macrophages are characterized by remarkable functional and phenotypic plasticity, pro-inflammatory (M1) anti-inflammatory (M2) states at the extremes multidimensional spectrum. Cell morphology major indicator for activation, describing M1(-like) (rounded) M2(-like) (elongated) different cell shapes. Here, we introduced...
With the increase in applications of machine learning methods drug design and related fields, challenge designing sound test sets becomes more prominent. The goal this is to have a realistic split chemical structures (compounds) between training, validation set such that performance on meaningful infer prospective application. This by its own very interesting relevant,but even complex federated approach where multiple partners jointly train model under privacy-preserving conditions must not...
With the increase in applications of machine learning methods drug design and related fields, challenge designing sound test sets becomes more prominent. The goal this is to have a realistic split chemical structures (compounds) between training, validation set such that performance on meaningful infer prospective application. This by its own very interesting relevant,but even complex federated approach where multiple partners jointly train model under privacy-preserving conditions must not...
With the increase in applications of machine learning methods drug design and related fields, challenge designing sound test sets becomes more prominent. The goal this is to have a realistic split chemical structures (compounds) between training, validation set such that performance on meaningful infer prospective application. This by its own very interesting relevant,but even complex federated approach where multiple partners jointly train model under privacy-preserving conditions must not...
The Cover Feature shows three chipmunks involved in the creation, analysis, and clustering of synthesizable virtual molecule library CHIPMUNK. Nearly 100 million compounds were generated with silico reactions on accessible building blocks, their descriptor profile was analysed. clustered together molecules from other public libraries order to relate it known chemical space divide huge into manageable subsets. It serves as an idea generator covers beyond rule five protein–protein well...
With the increase in applications of machine learning methods drug design and related fields, challenge designing sound test sets becomes more prominent. The goal this is to have a realistic split chemical structures (compounds) between training, validation set such that performance on meaningful infer prospective application. This by its own very interesting relevant,but even complex federated approach where multiple partners jointly train model under privacy-preserving conditions must not...