AI for organic and polymer synthesis

Science China Chemistry, Journal Year: 2024, Volume and Issue: 67(8), P. 2461 - 2496

Published: June 26, 2024

Language: Английский

---

Prediction of 19F NMR chemical shift by machine learning

Artificial Intelligence Chemistry, Journal Year: 2024, Volume and Issue: 2(1), P. 100043 - 100043

Published: Jan. 4, 2024

Fluorine-19 (19F) is a nucleus of great importance in the field Nuclear Magnetic Resonance (NMR) spectroscopy due to its high receptivity and wide chemical shift dispersion. 19F NMR plays crucial roles both organic synthesis biomedicine. Herein, machine learning-based comprehensive prediction model was established based on experimental dataset from book by Dolbier open database nmrshiftdb2. Fluorine radical SMILES (Fr-SMILES) that reflected fluorine equivalence, designed as representation molecule. Model trained with graph convolution network (GCN) algorithm gave low mean absolute error (MAE) 3.636 ppm testing set. This exhibits broad applicability can effectively predict shifts for range molecules. We believe current work will provide powerful tool not only predicting but also aiding analysis identification these diverse compounds. An online platform constructed model, which be found at https://fluobase.cstspace.cn/fnmr.

Language: Английский

---

Rational Design of Conjugated Polymers for Photocatalytic CO₂ Reduction: Towards Localized CO Production and Macrophage Polarization

Journal of the American Chemical Society, Journal Year: 2024, Volume and Issue: 146(36), P. 24832 - 24841

Published: Aug. 15, 2024

Light presents substantial potential in disease treatment, where the development of efficient photocatalysts could enhance utilization photocatalytic systems biomedicine. Here, we devised a novel approach to designing and synthesizing conjugated polymers for CO2 reduction, relying on multiple linear regression model built with theoretically calculated descriptors. We established logarithmic relationship between molecular structure CO yield identified poly(fluorene-co-thiophene) deviant (PFT) as optimal one. PFT excited regeneration ratio 231 nmol h–1 acetonitrile 46 an aqueous solution reaction selectivity 88%. Further advancements were made through liposomes encapsulating targeted macrophage delivery. By distributing liposome membranes, our constructed system efficiently generated situ from surrounding CO2. This localized production served endogenous signaling molecule, promoting desirable polarization macrophages M1 M2 phenotype. Consequently, cells reduced secretion pro-inflammatory cytokines (TNF-α, IL-6, IL-1β). also demonstrated efficacy treating lipopolysaccharide-induced inflammation cardiomyocytes under white light irradiation. Moreover, research provides comprehensive understanding intricate processes involved reduction by combination theoretical calculations experimental techniques including transient absorption, femtosecond ultrafast spectroscopy, infrared spectroscopy. These findings pave way further biomedical investigation.

Language: Английский

---

Atom-Based Machine Learning for Estimating Nucleophilicity and Electrophilicity with Applications to Retrosynthesis and Chemical Stability

Chemical Science, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 1, 2025

Nucleophilicity and electrophilicity are important properties for evaluating the reactivity selectivity of chemical reactions. It allows ranking nucleophiles electrophiles on scales, enabling a better understanding prediction reaction outcomes. Building upon our recent work (N. Ree, A. H. Göller J. Jensen, Automated quantum chemistry estimating nucleophilicity with applications to retrosynthesis covalent inhibitors, Digit. Discov., 2024, 3, 347-354), we introduce an atom-based machine learning (ML) approach predicting methyl cation affinities (MCAs) anion (MAAs) estimate electrophilicity, respectively. The ML models trained validated QM-derived data from around 50 000 neutral drug-like molecules, achieving Pearson correlation coefficients 0.97 MCA 0.95 MAA held-out test sets. In addition, demonstrate two different applications: first, as general tool filtering retrosynthetic routes based predictions, second, assessing stability esters carbamates towards hydrolysis code is freely available GitHub under MIT open source license web application at https://www.esnuel.org.

Language: Английский

---

Predicting Lewis Acidity: Machine Learning the Fluoride Ion Affinity of p‐Block‐Atom‐Based Molecules

Angewandte Chemie International Edition, Journal Year: 2024, Volume and Issue: 63(17)

Published: March 7, 2024

Abstract “How strong is this Lewis acid?” a question researchers often approach by calculating its fluoride ion affinity (FIA) with quantum chemistry. Here, we present FIA49k, an extensive FIA dataset 48,986 data points calculated at the RI‐DSD‐BLYP‐D3(BJ)/def2‐QZVPP//PBEh‐3c level of theory, including 13 different p ‐block atoms as accepting site. The FIA49k was used to train FIA‐GNN, two message‐passing graph neural networks, which predict gas and solution phase values molecules excluded from training mean absolute error 14 kJ mol −1 ( r 2 =0.93) SMILES string acid only input. accuracy notable, given wide energetic range 750 spanned FIA49k. model's value demonstrated four case studies, predictions for extracted Cambridge Structural Database reproducing results catalysis research available in literature. Weaknesses model are evaluated interpreted chemically. FIA‐GNN can be reached via free web app www.grebgroup.de/fia‐gnn ).

Language: Английский

---

Intramolecular chaperone-assisted dual-anchoring activation (ICDA): a suitable preorganization for electrophilic halocyclization

Chemical Science, Journal Year: 2024, Volume and Issue: 15(16), P. 6130 - 6140

Published: Jan. 1, 2024

The halocyclization reaction represents one of the most common methodologies for synthesis heterocyclic molecules. Many efforts have been made to balance relationship between structure, reactivity and selectivity, including design new electrophilic halogenation reagents utilization activating strategies. However, discovering universal or strategies remains challenging due case-by-case practice different substrates cyclization models. Here we report an intramolecular chaperone-assisted dual-anchoring activation (ICDA) model halocyclization, taking advantage non-covalent orientation as driving force. This protocol allows a practical, catalyst-free rapid approach access seven types small-sized, medium-sized, large-sized units realize polyene-like domino halocyclizations, exemplified by nearly 90 examples, risk-reducing flow gram-scale synthesis. DFT studies verify crucial role ICDA in affording suitable preorganization transition state stabilization X

Language: Английский

---

Prediction of Bond Dissociation Energy for Organic Molecules Based on a Machine‐Learning Approach

Chinese Journal of Chemistry, Journal Year: 2024, Volume and Issue: 42(17), P. 1967 - 1974

Published: April 20, 2024

Comprehensive Summary Bond dissociation energy (BDE), which refers to the enthalpy change for homolysis of a specific covalent bond, is one basic thermodynamic properties molecules. It very important understanding chemical reactivities, and transformations. Here, machine learning‐based comprehensive BDE prediction model was established based on i BonD experimental dataset calculated by St. John et al . D ifferential S tructural P hysic OC hemical (D‐SPOC) descriptors that reflected changes in molecules’ structural physicochemical features process bond were designed as input features. The trained with LightGBM algorithm gave low mean absolute error (MAE) 1.03 kcal/mol test set. D‐SPOC could apply accurate phenol O—H bonds, uncommon N‐SCF 3 O‐SCF reagents, β ‐C—H bonds enamine intermediates. A fast online platform constructed model, be found at http://isyn.luoszgroup.com/bde_prediction

Language: Английский

---

Unveiling CO₂ reactivity with data-driven methods

Digital Discovery, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 1, 2025

We present a data-driven approach that integrates supervised learning, quantum chemistry, and uncertainty quantification to determine CO 2 reactivity, enabling advances in carbon capture the design of value-added chemicals.

Language: Английский

---

Direct Amination of Anilines Utilizing Dearomatized Phenolate Species

Organic Letters, Journal Year: 2025, Volume and Issue: unknown

Published: May 15, 2025

Activation of the aryl C-N bond underpins critical challenges in modern organic synthesis. Herein, direct amination anilines is presented via hypervalent iodine-mediated transient dearomatized phenolate intermediates, enabling selective C(aryl)-NH2 cleavage under mild conditions. A library bioactive p-alkylaminophenols synthesized up to 85% yields within 3 h. Being used late-stage drug diversification and mechanistic studies, this protocol offers a modular platform for complex amine construction.

Language: Английский

---

Chemically Active Sulfonate Additive with Transition Metal and Oxygen Dual-Site Deactivation for High-Voltage LiCoO₂

ACS Energy Letters, Journal Year: 2024, Volume and Issue: 9(9), P. 4475 - 4484

Published: Aug. 21, 2024

Modulating a solid electrolyte interphase (SEI) through additives represents crucial approach for stable operation of cathode materials. Traditional perspectives focus on the electrochemical decomposition additive itself while underestimating reactivity and catalytic ability from surfaces. Inspired by sulfur-containing species causing catalyst poisoning, effects LiCoO2 (LCO) sulfur-based molecule (−SOx) have been investigated. We elucidate mechanism assessing multiple interactions with LCO surface, which are primarily characterized chemical ring-opening reactions. Furthermore, decreasing saturated carbon in molecules can enhance its reactivity, −SOx groups deactivate both lattice oxygen transition metal center dual-site bond formation. Benefiting chemically passivated sulfonate enables cycling at high temperature voltage 4.65 V. This provides new insights into mechanisms stabilizing interface molecular design additives.

Language: Английский

---