JBCS

This work paper the applications of Natural Language Processing (NLP) and Large Language Models (LLMs) in chemistry and materials science, highlighting their role in chemical entity recognition, reaction prediction, materials discovery, and literature analysis.

Machine learning uses algorithms and statistical models for defined tasks and learning patterns from data without explicit instructions. Its basic concepts and some applications are reviewed.

Kinetic of thermal processes can be accurately determined by combining artificial neural network with thermal analysis techniques.

Application of machine learning and computational tools to predict antileishmanial activity, emphasizing the synergy of computational and experimental methods in developing novel therapeutic agents.

Machine learning combined with virtual screening has enabled the discovery of a significant variety of molecules exhibiting high affinity with shikimate kinase. Subsequent evaluation through molecular dynamics and free energy calculations has facilitated the identification of potential inhibitor candidates.

De novo design of a new lead compound with potential inhibitory effect on monkeypox virus F13 protein (VP37) by deep reinforcement learning and structure-based drug design.

This study utilizes Quantitative Structure-Activity Relationship (QSAR)-based machine learning models, validated with bioactivity data median inhibitory concentration (IC₅₀) of compounds against Trypanosoma brucei and Trypanosoma cruzi, for screening Food and Drug Administration (FDA)-approved compounds as candidates for repurposing in the treatment of both trypanosomiases.

The blood composition imbalance following coronavirus disease (COVID-19) causes a systematic change in impedance, which can be modelled by multivariate analysis.

Molecules from the QM-symex database are converted to SMILES (simplified molecular input line entry system) and stored in a new QM-symex-modif dataset with their target properties. The data is processed and used in machine learning models to develop predictive models for photophysical properties.

Full voltammogram analysis through machine learning for enhanced detection in electrochemical immunosensors.

This study employs supervised learning methods, including Partial Least Squares Discriminant Analysis (PLS-DA) with bootstrap resampling and Support Vector Machine (SVM) ensemble, to analyze biogenic volatile organic compounds (BVOCs) emissions in the Atlantic Forest, achieving high classification accuracy.

A quantum active learning method (QAL) for automatic structural determination of doped materials has been developed and implemented in the QMLMaterial software. QAL uses quantum circuits for data encoding to create quantum machine learning models on-the-fly.

The figure illustrates the core components of quantitative structure-activity relationship (QSAR)-Lit, a platform designed to streamline the QSAR modeling process. It encompasses data curation, descriptor calculation, machine learning, and virtual screening, enabling seamless and efficient analysis for drug discovery applications.

Machine learning models were used to predict the biological activity of natural products against Schistosoma mansoni. Virtual screening identified 14 promising compounds, which were further analyzed for absorption, distribution, metabolism, excretion and toxicity (ADMET) properties.

From chemical compounds to predicting the antitrypanosomal activity of new candidates against Trypanosoma cruzi trypomastigotes.

Prediction of glycerol electrooxidation potentials using machine learning, with improved feature treatment.

Ant Colony Optimization method (inspired by the collective behavior of ants in optimizing paths) to optimize one (1D) to five (5D) parameters of an analytical density functional for the ground-state energy of strongly correlated systems. The performance is assessed by the mean relative error (MRE): for 3D and 5D we find MRE ca. 0.8%, an error reduction of 67% compared to the original parametrization (MRE = 2.4%).

Machine learning and multivariate statistical analyses identified molecular features correlated with biological activity of phenylpropanoid against Trypanosoma cruzi.