Machine Learning Methods for Small Data Challenges in Molecular Science

Chemical Reviews, Journal Year: 2023, Volume and Issue: 123(13), P. 8736 - 8780

Published: June 29, 2023

Small data are often used in scientific and engineering research due to the presence of various constraints, such as time, cost, ethics, privacy, security, technical limitations acquisition. However, big have been focus for past decade, small their challenges received little attention, even though they technically more severe machine learning (ML) deep (DL) studies. Overall, challenge is compounded by issues, diversity, imputation, noise, imbalance, high-dimensionality. Fortunately, current era characterized technological breakthroughs ML, DL, artificial intelligence (AI), which enable data-driven discovery, many advanced ML DL technologies developed inadvertently provided solutions problems. As a result, significant progress has made decade. In this review, we summarize analyze several emerging potential molecular science, including chemical biological sciences. We review both basic algorithms, linear regression, logistic regression (LR),

Language: Английский

---

Past, Present, and Future Perspectives on Computer-Aided Drug Design Methodologies

Molecules, Journal Year: 2023, Volume and Issue: 28(9), P. 3906 - 3906

Published: May 5, 2023

The application of computational approaches in drug discovery has been consolidated the last decades. These families techniques are usually grouped under common name "computer-aided design" (CADD), and they now constitute one pillars pharmaceutical pipelines many academic industrial environments. Their implementation demonstrated to tremendously improve speed early steps, allowing for proficient rational choice proper compounds a desired therapeutic need among extreme vastness drug-like chemical space. Moreover, CADD allows rationalization biochemical interactive processes interest at molecular level. Because this, tools extensively used also field 3D design optimization entities starting from structural information targets, which can be experimentally resolved or obtained with other computer-based techniques. In this work, we revised state-of-the-art computer-aided methods, focusing on their different scenarios biological interest, not only highlighting great potential benefits, but discussing actual limitations eventual weaknesses. This work considered brief overview methods discovery.

Language: Английский

---

Advances in artificial intelligence for drug delivery and development: A comprehensive review

Computers in Biology and Medicine, Journal Year: 2024, Volume and Issue: 178, P. 108702 - 108702

Published: June 7, 2024

Language: Английский

---

Current Status of Emerging Contaminant Models and Their Applications Concerning the Aquatic Environment: A Review

Water, Journal Year: 2025, Volume and Issue: 17(1), P. 85 - 85

Published: Jan. 1, 2025

Increasing numbers of emerging contaminants (ECs) detected in water environments require a detailed understanding these chemicals’ fate, distribution, transport, and risk aquatic ecosystems. Modeling is useful approach for determining ECs’ characteristics their behaviors environments. This article proposes systematic taxonomy EC models addresses gaps the comprehensive analysis applications. The reviewed include conventional quality models, multimedia fugacity machine learning (ML) models. Conventional have higher prediction accuracy spatial resolution; nevertheless, they are limited functionality can only be used to predict contaminant concentrations Fugacity excellent at depicting how travel between different environmental media, but cannot directly analyze variations parts same media because model assumes that constant within compartment. Compared other ML applied more scenarios, such as identification assessments, rather than being confined concentrations. In recent years, with rapid development artificial intelligence, surpassed becoming one newest hotspots study ECs. primary challenge faced by outcomes difficult interpret understand, this influences practical value an some extent.

Language: Английский

---

First fully-automated AI/ML virtual screening cascade implemented at a drug discovery centre in Africa

Nature Communications, Journal Year: 2023, Volume and Issue: 14(1)

Published: Sept. 15, 2023

Streamlined data-driven drug discovery remains challenging, especially in resource-limited settings. We present ZairaChem, an artificial intelligence (AI)- and machine learning (ML)-based tool for quantitative structure-activity/property relationship (QSAR/QSPR) modelling. ZairaChem is fully automated, requires low computational resources works across a broad spectrum of datasets. describe end-to-end implementation at the H3D Centre, leading integrated unit Africa, which no prior AI/ML capabilities were available. By leveraging in-house data collected over decade, we have developed virtual screening cascade malaria tuberculosis comprising 15 models key decision-making assays ranging from whole-cell phenotypic cytotoxicity to aqueous solubility, permeability, microsomal metabolic stability, cytochrome inhibition, cardiotoxicity. show how profiling compounds, synthesis testing, can inform progression frontrunner compounds H3D. This project first-of-its-kind deployment scale tools research centre operating low-resource setting.

Language: Английский

---

Equivariant Graph Neural Networks for Toxicity Prediction

Chemical Research in Toxicology, Journal Year: 2023, Volume and Issue: unknown

Published: Sept. 10, 2023

Predictive modeling of toxicity is a crucial step in the drug discovery pipeline. It can help filter out molecules with high probability failing early stages de novo design. Thus, several machine learning (ML) models have been developed to predict by combining classical ML techniques or deep neural networks well-known molecular representations such as fingerprints 2D graphs. But more natural, accurate representation expected be defined physical 3D space like ab initio methods. Recent studies successfully used equivariant graph (EGNNs) for based on structures quantum-mechanical properties molecules. Inspired this, we investigated performance EGNNs construct reliable prediction. We transformer (ET) model TorchMD-NET this. Eleven data sets taken from MoleculeNet, TDCommons, and ToxBenchmark considered evaluate capability ET Our results show that adequately learns correlate activity, achieving good accuracies most comparable state-of-the-art models. also test physicochemical property, namely, total energy molecule, inform prediction prior. However, our work suggests these two not related. provide an attention weight analysis helping understand thus increase explainability model. In summary, findings offer promising insights considering geometry information via straightforward way integrate conformers into ML-based pipelines predicting investigating space. expect future, especially larger, diverse sets, will essential tool this domain.

Language: Английский

---

Advancing toxicity studies of per- and poly-fluoroalkyl substances (pfass) through machine learning: Models, mechanisms, and future directions

The Science of The Total Environment, Journal Year: 2024, Volume and Issue: 946, P. 174201 - 174201

Published: June 25, 2024

Language: Английский

---

Guidance for good practice in the application of machine learning in development of toxicological quantitative structure-activity relationships (QSARs)

PLoS ONE, Journal Year: 2023, Volume and Issue: 18(5), P. e0282924 - e0282924

Published: May 10, 2023

Recent years have seen a substantial growth in the adoption of machine learning approaches for purposes quantitative structure-activity relationship (QSAR) development. Such trend has coincided with desire to see shifting focus methodology employed within chemical safety assessment: away from traditional reliance upon animal-intensive vivo protocols, and towards increased application silico (or computational) predictive toxicology. With QSAR central amongst techniques applied this area, emergence algorithms trained through objective toxicity estimation has, quite naturally, arisen. On account pattern-recognition capabilities underlying methods, statistical power ensuing models is potentially considerable–appropriate handling even vast, heterogeneous datasets. However, such potency comes at price: manifesting as general practical deficits observed respect reproducibility, interpretability generalisability resulting tools. Unsurprisingly, these elements served hinder broader uptake (most notably regulatory setting). Areas uncertainty liable accompany (and hence detract applicability of) toxicological previously been highlighted, accompanied by forwarding suggestions “best practice” aimed mitigation their influence. scope exercises remained limited “classical” QSAR–that conducted use linear regression related techniques, comparatively few features or descriptors. Accordingly, intention study extend remit best practice guidance, so address concerns specific employment field. In doing so, impact strategies enhancing transparency (feature importance, feature reduction), (cross-validation) (hyperparameter optimisation) algorithms, real data six common approaches, evaluated.

Language: Английский

---

Decoding Nanomaterial‐Biosystem Interactions through Machine Learning

Angewandte Chemie International Edition, Journal Year: 2024, Volume and Issue: 63(16)

Published: Jan. 25, 2024

Abstract The interactions between biosystems and nanomaterials regulate most of their theranostic nanomedicine applications. These nanomaterial‐biosystem are highly complex influenced by a number entangled factors, including but not limited to the physicochemical features nanomaterials, types characteristics interacting biosystems, properties surrounding microenvironments. Over years, different experimental approaches coupled with computational modeling have revealed important insights into these interactions, although many outstanding questions remain unanswered. emergence machine learning has provided timely unique opportunity revisit further push boundary this field. This minireview highlights development use decode provides our perspectives on current challenges potential opportunities in

Language: Английский

---

Applications of conceptual density functional theory in reference to quantitative structure–activity / property relationship

Molecular Physics, Journal Year: 2024, Volume and Issue: 122(23)

Published: March 22, 2024

To predict the biological effects of chemical compounds based on mathematical and statistical relationships, quantitative structure–activity relationship (QSAR) approach is used. Based molecular characteristics diverse substances, Quantitative Structure–Property Relationship (QSPR) techniques estimate physiochemical attributes whereas Structure Toxicity (QSTR) used as a link between structure species its toxicity. These ligand-based computational screening methods offer cost-effective replacement for laboratory-based procedures. Different QSTR models are established to understand activities related Density Functional Theory (DFT) ab-initio examine external acute toxicity using Quantum Chemical (QC) descriptors electron correlation contribution. Conceptual (CDFT) global local have wide applications in analysing various physical species. The like hardness, electronegativity, electrophilicity index, HOMO–LUMO energy, enthalpy found reliable model terms available experimental data. Various through Multi Linear Regression (MLR) analysis which links calculated with their activities. In this review, CDFT-based descriptors, described detail QSAR / QSPR/ studies.

Language: Английский

---