Search

Active learning allows algorithms to steer iterative experimentation to accelerate and de-risk molecular optimizations, but actively trained models might still exhibit poor performance during early project stages where the training data is limited and model exploitation might lead to analog identification with limited scaffold diversity. Here, we present ActiveDelta, an adaptive approach that leverages paired molecular representations to predict improvements from the current best training compound to prioritize further data acquisition. We apply the ActiveDelta concept to both graph-based deep (Chemprop) and tree-based (XGBoost) models during exploitative active learning for 99 Ki benchmarking datasets. We show that both ActiveDelta implementations excel at identifying more potent inhibitors compared to the standard exploitative active learning implementations of Chemprop, XGBoost, and Random Forest. The ActiveDelta approach is also able to identify more chemically diverse inhibitors in terms of their Murcko scaffolds. Finally, deep models such as Chemprop trained on data selected through ActiveDelta approaches can more accurately identify inhibitors in test data created through simulated time-splits. Overall, this study highlights the large potential for molecular pairing approaches to further improve popular active learning strategies in low data regimes by enabling faster and more accurate identification of more diverse molecular hits against critical drug targets.

Abstract Established molecular machine learning models process individual molecules as inputs to predict their biological, chemical, or physical properties. However, such algorithms require large datasets and have not been optimized to predict property differences between molecules, limiting their ability to learn from smaller datasets and to directly compare the anticipated properties of two molecules. Many drug and material development tasks would benefit from an algorithm that can directly compare two molecules to guide molecular optimization and prioritization, especially for tasks with limited available data. Here, we develop DeepDelta, a pairwise deep learning approach that processes two molecules simultaneously and learns to predict property differences between two molecules from small datasets. On 10 ADMET benchmark tasks, our DeepDelta approach significantly outperforms two established molecular machine learning algorithms, the directed message passing neural network (D-MPNN) ChemProp and Random Forest using radial fingerprints, for 70% of benchmarks in terms of Pearson’s r, 60% of benchmarks in terms of mean absolute error (MAE), and all external test sets for both Pearson’s r and MAE. We further analyze our performance and find that DeepDelta is particularly outperforming established approaches at predicting large differences in molecular properties and can perform scaffold hopping. Furthermore, we derive mathematically fundamental computational tests of our models based on mathematical invariants and show that compliance to these tests correlates with overall model performance — providing an innovative, unsupervised, and easily computable measure of expected model performance and applicability. Taken together, DeepDelta provides an accurate approach to predict molecular property differences by directly training on molecular pairs and their property differences to further support fidelity and transparency in molecular optimization for drug development and the chemical sciences.

Active machine learning is an established and increasingly popular experimental design technique where the machine learning model can request additional data to improve the model's predictive performance. It is generally assumed that this data is optimal for the machine learning model since it relies on the model's predictions or model architecture and therefore cannot be transferred to other models. Inspired by research in pedagogy, we here introduce the concept of yoked machine learning where a second machine learning model learns from the data selected by another model. We found that in 48% of the benchmarked combinations, yoked learning performed similar or better than active learning. We analyze distinct cases in which yoked learning can improve active learning performance. In particular, we prototype yoked deep learning (YoDeL) where a classic machine learning model provides data to a deep neural network, thereby mitigating challenges of active deep learning such as slow refitting time per learning iteration and poor performance on small datasets. In summary, we expect the new concept of yoked (deep) learning to provide a competitive option to boost the performance of active learning and benefit from distinct capabilities of multiple machine learning models during data acquisition, training, and deployment.

Multi-factor screenings are commonly used in diverse applications in medicine and bioengineering, including optimizing combination drug treatments and microbiome engineering. Despite the advances in high-throughput technologies, large-scale experiments typically remain prohibitively expensive. Here we introduce a machine learning platform, structure-augmented regression (SAR), that exploits the intrinsic structure of each biological system to learn a high-accuracy model with minimal data requirement. Under different environmental perturbations, each biological system exhibits a unique, structured phenotypic response. This structure can be learned based on limited data and once learned, can constrain subsequent quantitative predictions. We demonstrate that SAR requires significantly fewer data comparing to other existing machine-learning methods to achieve a high prediction accuracy, first on simulated data, then on experimental data of various systems and input dimensions. We then show how a learned structure can guide effective design of new experiments. Our approach has implications for predictive control of biological systems and an integration of machine learning prediction and experimental design.

Abstract Background Patients undergoing cardiac surgery are prone to numerous complications. Increased vascular permeability may be associated with morbidity and mortality due to hemodynamic instability, fluid overload, and edema formation. We hypothesized that markers of endothelial injury and inflammation are associated with capillary leak, ultimately increasing the risk of postoperative complications. Methods In this prospective, observational, multidisciplinary cohort study at our tertiary academic medical center, we recruited 405 cardiac surgery patients. Patients were assessed daily using body impedance electrical analysis, ultrasound, sublingual intravital microscopy, and analysis of serum biomarkers. Multivariable models, as well as machine learning, were used to study the association of angiopoietin-2 with extracellular water as well as common complications after cardiac surgery. Results The majority of patients underwent coronary artery bypass grafting, valvular, or aortic surgeries. Across the groups, extracellular water increased postoperatively (20 ± 6 preoperatively to 29 ± 7L on postoperative day 2; P

Abstract Background The concomitant occurrence of the symptoms intravascular hypovolemia, peripheral edema and hemodynamic instability is typically named Capillary Leak Syndrome (CLS) and often occurs in surgical critical ill patients. However, neither a unitary definition nor standardized diagnostic criteria exist so far. We aimed to investigate common characteristics of this phenomenon with a subsequent scoring system, determining whether CLS contributes to mortality. Methods We conducted this single-center, observational, multidisciplinary, prospective trial in two separately run surgical ICUs of a tertiary academic medical center. 200 surgical patients admitted to the ICU and 30 healthy volunteers were included. Patients were clinically diagnosed as CLS or No-CLS group (each N = 100) according to the grade of edema, intravascular hypovolemia, hemodynamic instability, and positive fluid balance by two independent attending physicians with > 10 years of experience in ICU. We performed daily measurements with non-invasive body impedance electrical analysis, ultrasound and analysis of serum biomarkers to generate objective diagnostic criteria. Receiver operating characteristics were used, while we developed machine learning models to increase diagnostic specifications for our scoring model. Results The 30-day mortility was increased among CLS patients (12 vs. 1%, P = 0.002), while showing higher SOFA-scores. Extracellular water was increased in patients with CLS with higher echogenicity of subcutaneous tissue [29(24–31) vs. 19(16–21), P

Abstract Continuous monitoring in the intensive care setting has transformed the capacity to rapidly respond with interventions for patients in extremis. Noninvasive monitoring has generally been limited to transdermal or intravascular systems coupled to transducers including oxygen saturation or pressure. Here it is hypothesized that gastric fluid (GF) and gases, accessible through nasogastric (NG) tubes, commonly found in intensive care settings, can provide continuous access to a broad range of biomarkers. A broad characterization of biomarkers in swine GF coupled to time‐matched serum is conducted . The relationship and kinetics of GF‐derived analyte level dynamics is established by correlating these to serum levels in an acute renal failure and an inducible stress model performed in swine. The ability to monitor ketone levels and an inhaled anaesthetic agent (isoflurane) in vivo is demonstrated with novel NG‐compatible sensor systems in swine. Gastric access remains a main stay in the care of the critically ill patient, and here the potential is established to harness this establishes route for analyte evaluation for clinical management.

Summary: Biological screens are plagued by false-positive hits resulting from aggregation. Methods to triage small colloidally aggregating molecules (SCAMs) are in high demand. Herein, we disclose a neural network to flag such entities. Our data demonstrate the utility of machine learning for predicting SCAMs, achieving 80% of correct predictions in an out-of-sample evaluation. The tool is competitive with a panel of expert chemists, who correctly predict 61% ± 7% of the same molecules in a Turing-like test. Our computational routine provides insight into features governing aggregation that had remained hidden to expert intuition. Further, we quantify that up to 15%–20% of ligands in publicly available chemogenomic databases have high potential to aggregate at a typical screening concentration (30 μM), imposing caution in systems biology and drug design programs. Our approach provides a means to augment human intuition and mitigate attrition and a pathway to accelerate future molecular medicine.

Summary: Inactive ingredients and generally recognized as safe compounds are regarded by the US Food and Drug Administration (FDA) as benign for human consumption within specified dose ranges, but a growing body of research has revealed that many inactive ingredients might have unknown biological effects at these concentrations and might alter treatment outcomes. To speed up such discoveries, we apply state-of-the-art machine learning to delineate currently unknown biological effects of inactive ingredients—focusing on P-glycoprotein (P-gp) and uridine diphosphate-glucuronosyltransferase-2B7 (UGT2B7), two proteins that impact the pharmacokinetics of approximately 20% of FDA-approved drugs. Our platform identifies vitamin A palmitate and abietic acid as inhibitors of P-gp and UGT2B7, respectively; in silico, in vitro, ex vivo, and in vivo validations support these interactions. Our predictive framework can elucidate biological effects of commonly consumed chemical matter with implications on food- and excipient-drug interactions and functional drug formulation development. : Reker et al. use machine learning to identify biological activities of food and drug additives. Validation confirms vitamin A palmitate as an inhibitor of P-glycoprotein transport and abietic acid as an inhibitor of UGT2b7 metabolism. Such associations have important implications as food- or excipient-drug interactions. Keywords: machine learning, pharmacology, virtual screening, excipient-drug interactions, food-drug interactions, pharmacokinetics, data science, inactive ingredients, drug delivery

Predicting the macromolecular targets of drug-like molecules has become everyday practice in medicinal chemistry. We present an overview of our recent research activities in the area of polypharmacology-guided drug design. A focus is put on the self-organizing map (SOM) as a tool for compound clustering and visualization. We show how the SOM can be efficiently used for target-panel prediction, drug re-purposing, and the design of focused compound libraries. We also present the concept of virtual organic synthesis in combination with quantitative estimates of ligand-receptor binding, which we used for de novo designing target-selective ligands. We expect these and related approaches to enable the future discovery of personalized medicines.