A5001435363- Computational Drug Discovery Methods
- Receptor Mechanisms and Signaling
- Machine Learning in Materials Science
- AI in cancer detection
- Protein Structure and Dynamics
- Cell Image Analysis Techniques
- Zebrafish Biomedical Research Applications
- Bioinformatics and Genomic Networks
- Medical Imaging and Analysis
- Metabolomics and Mass Spectrometry Studies
- Chemical Synthesis and Analysis
- Biomedical Text Mining and Ontologies
- Single-cell and spatial transcriptomics
- Explainable Artificial Intelligence (XAI)
- Genomics and Chromatin Dynamics
- Machine Learning in Healthcare
- Pharmacogenetics and Drug Metabolism
- Medical Image Segmentation Techniques
- Microbial Natural Products and Biosynthesis
- RNA and protein synthesis mechanisms
- Advanced Fluorescence Microscopy Techniques
- Neurobiology and Insect Physiology Research
- Biosimilars and Bioanalytical Methods
- Intracerebral and Subarachnoid Hemorrhage Research
- Neuroscience and Neuropharmacology Research
University of California, San Francisco
2016-2025
Institute for Neurodegenerative Disorders
2014-2024
Keiser University
2024
Kavli Institute for Theoretical Sciences
2020
University of California System
2020
Universidad Católica de Santa Fe
2019
Quantitative BioSciences
2018
SeaChange Pharmaceuticals (United States)
2012-2013
University of New Mexico
2008
Biocom
2008
Abstract Neuropathologists assess vast brain areas to identify diverse and subtly-differentiated morphologies. Standard semi-quantitative scoring approaches, however, are coarse-grained lack precise neuroanatomic localization. We report a proof-of-concept deep learning pipeline that identifies specific neuropathologies—amyloid plaques cerebral amyloid angiopathy—in immunohistochemically-stained archival slides. Using automated segmentation of stained objects cloud-based interface, we...
Abstract Motivation Knowledge graphs (KGs) are being adopted in industry, commerce and academia. Biomedical KG presents a challenge due to the complexity, size heterogeneity of underlying information. Results In this work, we present Scalable Precision Medicine Open Engine (SPOKE), biomedical connecting millions concepts via semantically meaningful relationships. SPOKE contains 27 million nodes 21 different types 53 edges 55 downloaded from 41 databases. The graph is built on framework 11...
Virtual and high-throughput screens (HTS) should have complementary strengths weaknesses, but studies that prospectively comprehensively compare them are rare. We undertook a parallel docking HTS screen of 197861 compounds against cruzain, thiol protease target for Chagas disease, looking reversible, competitive inhibitors. On workup, 99% the hits were eliminated as false positives, yielding 146 well-behaved, ligands. These fell into five chemotypes: two prioritized by scoring among top 0.1%...
The similarity of drug targets is typically measured using sequence or structural information. Here, we consider chemo-centric approaches that measure target on the basis their ligands, asking how chemoinformatics similarities differ from those derived bioinformatically, stable ligand networks are to changes in metrics, and which network most reliable for prediction pharmacology. We calculated between hundreds ligands mapped relationship them a formal network. Bioinformatics were based BLAST...
Statistical and machine learning approaches predict drug-to-target relationships from 2D small-molecule topology patterns. One might expect 3D information to improve these calculations. Here we apply the logic of extended connectivity fingerprint (ECFP) develop a rapid, alignment-invariant representation molecular conformers, three-dimensional (E3FP). By integrating E3FP with similarity ensemble approach (SEA), achieve higher precision-recall performance relative SEA ECFP on ChEMBL20...
New machine learning methods to analyze raw chemical and biological data are now widely accessible as open-source toolkits. This positions researchers leverage powerful, predictive models in their own domains. We caution, however, that the application of experimental research merits careful consideration. Machine algorithms readily exploit confounding variables artifacts instead relevant patterns, leading overoptimistic performance poor model generalization. In parallel strong control...
Machine learning-based drug discovery success depends on molecular representation. Yet traditional fingerprints omit both the protein and pointers back to structural information that would enable better model interpretability. Therefore, we propose LUNA, a Python 3 toolkit calculates encodes protein–ligand interactions into new hashed inspired by Extended Connectivity FingerPrint (ECFP): EIFP (Extended Interaction FingerPrint), FIFP (Functional Hybrid (HIFP). LUNA also provides visual...
Natural and experimental genetic variants can modify DNA loops insulating boundaries to tune transcription, but it is unknown how sequence perturbations affect chromatin organization genome wide. We developed a deep-learning strategy quantify the effect of any insertion, deletion, or substitution on contacts systematically scored millions synthetic variants. While most manipulations have little impact, regions with CTCF motifs active transcription are highly sensitive, as expected. Our...
Whereas 400 million distinct compounds are now purchasable within the span of a few weeks, biological activities most unknown. To facilitate access to new chemistry for biology, we have combined Similarity Ensemble Approach (SEA) with maximum Tanimoto similarity nearest bioactive predict activity every commercially available molecule in ZINC. This method, which label SEA+TC, outperforms both SEA and naïve-Bayesian classifier via predictive performance on 5-fold cross-validation ChEMBL's...
Abstract Artificial intelligence models match or exceed dermatologists in melanoma image classification. Less is known about their robustness against real-world variations, and clinicians may incorrectly assume that a model with an acceptable area under the receiver operating characteristic curve related performance metric ready for clinical use. Here, we systematically assessed of dermatologist-level convolutional neural networks (CNNs) on non-curated images by applying computational...
Message passing neural networks (MPNNs) on molecular graphs generate continuous and differentiable encodings of small molecules with state-of-the-art performance protein-ligand complex scoring tasks. Here, we describe the proximity graph network (PGN) package, an open-source toolkit that constructs ligand-receptor based atom allows users to rapidly apply evaluate MPNN architectures for a broad range We demonstrate utility PGN by introducing benchmarks affinity docking score prediction Graph...
Researchers are developing increasingly robust molecular representations, motivating the need for thorough methods to stress-test and validate them. Here, we use a variational auto-encoder (VAE), an unsupervised deep learning model, generate anomalous examples of SELF-referencIng Embedded Strings (SELFIES), popular string format. These anomalies defy assertion that all SELFIES convert into valid SMILES strings. Interestingly, find specific regions within VAE's internal landscape (latent...