A5047159788- Computational Drug Discovery Methods
- Machine Learning in Materials Science
- Genetics, Bioinformatics, and Biomedical Research
- Scientific Computing and Data Management
- Metabolomics and Mass Spectrometry Studies
- Protein Structure and Dynamics
- Cell Image Analysis Techniques
- Mass Spectrometry Techniques and Applications
- Bioinformatics and Genomic Networks
- Advanced Proteomics Techniques and Applications
- Chemistry and Chemical Engineering
Charité - Universitätsmedizin Berlin
2020-2023
Humboldt-Universität zu Berlin
2021-2022
Freie Universität Berlin
2021-2022
Firmenich (Switzerland)
2019
A chemical language model for molecular property prediction: it outperforms prior art, is validated on a large, proprietary toxicity dataset, reveals cytotoxic motifs through attention & uses two uncertainty techniques to improve reliability.
Abstract In drug development, late stage toxicity issues of a compound are the main cause failure in clinical trials. silico methods therefore high importance to guide early design process reduce time, costs and animal testing. Technical advances ever growing amount available data enabled machine learning, especially neural networks, impact field predictive toxicology. this study, cytotoxicity prediction, one earliest handles discovery, is investigated using deep learning approach trained on...
In our study, we demonstrate the synergy effect between convolutional neural networks and multiplicity of SMILES. The model propose, so-called Convolutional Neural Fingerprint (CNF) model, reaches accuracy traditional descriptors such as Dragon (Mauri et al. [22]), RDKit (Landrum [18]), CDK2 (Willighagen [43]) PyDescriptor (Masand Rastija [20]). Moreover CNF generally performs better than highly fine-tuned descriptors, especially on small data sets, which is great interest for chemical field...
Accurate molecular property or activity prediction is one of the main goals in computer-aided drug design. Quantitative structure-activity relationship (QSAR) modeling and machine learning, more recently deep have become an integral part this process. Such algorithms require lots data for training which, case physico-chemical bioactivity sets, remains scarce. To address lack data, augmentation techniques are increasingly applied learning. Here, we exploit that compound can be represented by...
Recent advances in machine learning (ML) are reshaping drug discovery. Structure-based ML methods use physically-inspired models to predict binding affinities from protein:ligand complexes. These promise enable the integration of data for many related targets, which addresses issues scarcity single targets and could generalizable predictions a broad range including mutants. In this work, we report our experiences building KinoML, novel framework target-based small molecule discovery with an...
Computational pipelines have become a crucial part of modern drug discovery campaigns. Setting up and maintaining such pipelines, however, can be challenging time-consuming-especially for novice scientists in this domain. TeachOpenCADD is platform that aims to teach domain-specific skills provide pipeline templates as starting points research projects. We offer Python-based solutions common tasks cheminformatics structural bioinformatics the form Jupyter notebooks, based on open source...
Computational pipelines have become a crucial part of modern drug discovery campaigns. Setting up and maintaining such pipelines, however, can be challenging time-consuming --- especially for novice scientists in this domain. TeachOpenCADD is platform that aims to teach domain-specific skills provide pipeline templates as starting points research projects. We offer Python-based solutions common tasks cheminformatics structural bioinformatics the form Jupyter notebooks based on open source...