A meta-learning approach for selectivity prediction in asymmetric catalysis

Nature Communications, Journal Year: 2025, Volume and Issue: 16(1)

Published: April 15, 2025

Abstract Transition metal-catalyzed asymmetric reactions are of high contemporary importance in organic synthesis. Recently, machine learning (ML) has shown promise accelerating the development newer catalytic protocols. However, need for large amount experimental data can present a bottleneck implementing ML models. Here, we propose meta-learning workflow that harness literature-derived to extract shared reaction features and requires only few examples predict outcome new reactions. Prototypical networks used as method enantioselectivity hydrogenation olefins. This model consistently provides significant performance improvement over other popular methods such random forests graph neural networks. The our meta-model is analyzed with varying sizes training demonstrate its utility even limited data. A good on an out-of-sample test set further indicates general applicability approach. We believe this work will provide leap forward identifying promising early phases when minimal available.

Language: Английский

---

From Single-Atom to Dual-Atom: A Universal Principle for the Rational Design of Heterogeneous Fenton-like Catalysts

Environmental Science & Technology, Journal Year: 2025, Volume and Issue: unknown

Published: April 22, 2025

Developing efficient heterogeneous Fenton-like catalysts is the key point to accelerating removal of organic micropollutants in advanced oxidation process. However, a general principle guiding reasonable design highly has not been constructed up now. In this work, total 16 single-atom and 272 dual-atom transition metal/nitrogen/carbon (TM/N/C) for H2O2 dissociation were explored systematically based on high-throughput density functional theory machine learning. It was found that TM/N/C exhibited distinct volcano-type relationship between catalytic activity •OH adsorption energy. The favorable energies range -3.11 ∼ -2.20 eV. Three different descriptors, namely, energetic, electronic, structural found, which can correlate intrinsic properties their activity. Using energy, stability, activation energy as evaluation criteria, two CoCu/N/C CoRu/N/C screened out from candidates, higher than best catalyst due synergistic effect. This work could present conceptually novel understanding inspire structure-oriented viewpoint volcano relationship.

Language: Английский

---

Selective C–H Borylation of Polyaromatic Compounds Enabled by Metal-Arene π-Complexation

Journal of the American Chemical Society, Journal Year: 2025, Volume and Issue: unknown

Published: April 23, 2025

The undirected Ir-catalyzed C-H borylation usually occurs preferentially at the least hindered and more acidic bond of aromatic ring. In case polyaromatic compounds possessing multiple unbiased sterically accessible bonds, site selectivity for nondirected is low. Here, we report dramatic effect exerted by π-complexation a chromium tricarbonyl unit on ring in context borylation. Competition experiments demonstrate that bonds an bound to react average two orders magnitude faster toward than unbound arenes. This enables unprecedented with high π-complexed tripod series organic compounds. Besides, drastic enhancement reactivity induced allows occur room temperature substrate as limiting reagent. DFT studies indicate oxidative addition has lower activation barriers when arenes are complexed unit, explaining observed exceptional selectivity. study will further spearhead development bimetallic system harness noncovalent metal-arene π-type interactions functionalization.

Language: Английский

---

The Importance of Atomic Charges for Predicting Site-Selective Ir-, Ru-, and Rh-Catalyzed C–H Borylations

The Journal of Organic Chemistry, Journal Year: 2025, Volume and Issue: unknown

Published: April 23, 2025

A supervised machine learning model has been developed that allows for the prediction of site selectivity in late-stage C-H borylations. Model development was accomplished using literature data site-selective (≥95%) borylation 189 unique arene, heteroarene, and aliphatic substrates feature a total 971 possible sp2 or sp3 sites. The reported experimental supplemented with additional chemoinformatic descriptors, computed atomic charges at sites, from parameterization catalytically active tris-boryl complexes resulting combination seven different Ir-, Ru-, Rh-based precatalysts eight ligands. Of over 1600 parameters investigated, (e.g., Hirshfeld, ChelpG, Mulliken charges) on hydrogen carbon atoms were identified as most important features allow successful whether particular bond will undergo borylation. overall accuracy 88.9% ± 2.5% precision, recall, F1 scores 92-95% nonborylating sites 65-75% demonstrated to be generalizable molecules outside training/test sets an validation set 12 electronically structurally diverse systems.

Language: Английский

---

Feed-Forward Neural Network for Predicting Enantioselectivity of the Asymmetric Negishi Reaction

ACS Central Science, Journal Year: 2023, Volume and Issue: 9(9), P. 1768 - 1774

Published: Aug. 24, 2023

Density functional theory (DFT) is a powerful tool to model transition state (TS) energies predict selectivity in chemical synthesis. However, successful multistep synthesis campaign must navigate energetically narrow differences pathways that create some limits rapid and unambiguous application of DFT these problems. While data science techniques may provide complementary approach overcome this problem, doing so with the relatively small sets are widespread organic presents significant challenge. Herein, we show set can be labeled features from TS calculations train feed-forward neural network for predicting enantioselectivity Negishi cross-coupling reaction P-chiral hindered phosphines. This modeling compared conventional approaches, including exclusive use using ligands or ground states architectures.

Language: Английский

---