Photocatalyzed Azidofunctionalization of Alkenes via Radical‐Polar Crossover

Angewandte Chemie, Год журнала: 2025, Номер unknown

Опубликована: Янв. 2, 2025

Abstract The azidofunctionalization of alkenes under mild conditions using commercially available starting materials and easily accessible reagents is reported based on a radical‐polar crossover strategy. A broad range alkenes, including vinyl arenes, enamides, enol ethers, sulfides, dehydroamino esters, were regioselectively functionalized with an azide nucleophiles such as azoles, carboxylic acids, alcohols, phosphoric oximes, phenols. method led to more efficient synthesis 1,2‐azidofunctionalized pharmaceutical intermediates when compared previous approaches, resulting in both reduction step count increase overall yield. scope limitations these transformations further investigated through standard unbiased selection 15 substrate combinations out 1,175,658 possible clustering technique.

Язык: Английский

---

Exploring BERT for Reaction Yield Prediction: Evaluating the Impact of Tokenization, Molecular Representation, and Pretraining Data Augmentation

Journal of Chemical Information and Modeling, Год журнала: 2025, Номер unknown

Опубликована: Май 1, 2025

Predicting reaction yields in synthetic chemistry remains a significant challenge. This study systematically evaluates the impact of tokenization, molecular representation, pretraining data, and adversarial training on BERT-based model for yield prediction Buchwald-Hartwig Suzuki-Miyaura coupling reactions using publicly available HTE data sets. We demonstrate that representation choice (SMILES, DeepSMILES, SELFIES, Morgan fingerprint-based notation, IUPAC names) has minimal performance, while typically BPE SentencePiece tokenization outperform other methods. WordPiece is strongly discouraged SELFIES notation. Furthermore, with relatively small sets (<100 K reactions) achieves comparable performance to larger containing millions examples. The use artificially generated domain-specific proposed. prove be good surrogate schemes extracted from such as Pistachio or Reaxys. best was observed hybrid combining real domain-specific, artificial data. Finally, we show novel approach, perturbing input embeddings dynamically, improves robustness generalizability success prediction. These findings provide valuable insights developing robust practical machine learning models chemistry. GSK's BERT code base made community this work.

Язык: Английский

---

Substrate-Photocatalyst Reactivity Matching Enables Broad Aryl Halide Scope in Light-Driven, Reductive Cross-Electrophile Coupling Using ¹³C NMR as a Predictor

ACS Catalysis, Год журнала: 2025, Номер unknown, С. 8103 - 8113

Опубликована: Май 1, 2025

Язык: Английский

---

Applying Active Learning toward Building a Generalizable Model for Ni-Photoredox Cross-Electrophile Coupling of Aryl and Alkyl Bromides

Journal of the American Chemical Society, Год журнала: 2025, Номер unknown

Опубликована: Май 22, 2025

When developing machine learning models for yield prediction, the two main challenges are effectively exploring condition space and substrate space. In this article, we disclose an approach mapping Ni/photoredox-catalyzed cross-electrophile coupling of alkyl bromides aryl in a high-throughput experimentation (HTE) context. This model employs active (in particular, uncertainty querying) as strategy to rapidly construct model. Given vastness space, focused on that builds initial then uses minimal data set expand into new chemical spaces. built virtual 22,240 compounds using less than 400 points. We demonstrated can be expanded 33,312 by adding information around 24 building blocks (<100 additional reactions). Comparing learning-based one constructed randomly selected showed was significantly better at predicting which reactions will successful. A combination density function theory (DFT) difference Morgan fingerprints employed random forest Feature importance analysis indicates key DFT features related reaction mechanism (e.g., radical LUMO energy) were crucial performance predictions outside training set. anticipate combining featurization uncertainty-based querying help synthetic organic community build predictive data-efficient manner other feature large diverse scopes.

Язык: Английский

---

Room-Temperature Copper Cross-Coupling Reactions of Anilines with Aryl Bromides

Organic Letters, Год журнала: 2025, Номер unknown

Опубликована: Май 28, 2025

We present a novel copper-catalyzed method for aniline cross-couplings promoted by 6-hydroxy picolinhydrazide ligand. The achieves room-temperature reactivity with aryl bromides, enabled methanol/ethanol solvent mixture and mild, functional group-compatible base, catalyst loadings as low 0.5 mol %. use of industrially preferred solvents well the high catalytic activity, offers significant advancement in practicality scalability industrial processes. Furthermore, approach extends to cross-coupling chlorides under elevated temperatures demonstrates compatibility additional nucleophile classes.

Язык: Английский

---

Production and evaluation of high-throughput reaction data from an automated chemical synthesis platform

Science China Chemistry, Год журнала: 2025, Номер unknown

Опубликована: Май 26, 2025

Язык: Английский

---

Intermediate knowledge enhanced the performance of the amide coupling yield prediction model

Chemical Science, Год журнала: 2025, Номер unknown

Опубликована: Янв. 1, 2025

Amide coupling, a key medicinal chemistry reaction, benefits from AI to minimize trial-and-error.

Язык: Английский

---

Rapid Prediction of Conformationally-Dependent DFT-Level Descriptors using Graph Neural Networks for Carboxylic Acids and Alkyl Amines

Опубликована: Фев. 23, 2024

Data-driven reaction discovery and development is a growing field that relies on the use of molecular descriptors to capture key information about substrates, ligands, targets. Broad adaptation this strategy hindered by associated computational cost descriptor calculation, especially when considering conformational flexibility. Descriptor libraries can be pre-computed agnostic application reduce burden data-driven development. However, as one often applies these models evaluate novel hypothetical structures, it would ideal predict compounds on-the-fly. Herein, we report DFT-level for ensembles 8528 carboxylic acids 8172 alkyl amines towards goal. Employing 2D 3D graph neural network architectures trained culminated in predictive molecule-level descriptors, well bond- atom-level conserved reactive site (carboxylic acid or amine). The predictions were confirmed robust an external validation set medicinally-relevant amines. Additionally, retrospective study correlating rate amide coupling reactions demonstrated suitability predicted downstream applications. Ultimately, enable high-fidelity vast number potential greatly increasing accessibility

Язык: Английский

---

HTE OS: A High-Throughput Experimentation Workflow Built from the Ground Up

Organic Process Research & Development, Год журнала: 2024, Номер 28(7), С. 2875 - 2884

Опубликована: Июнь 27, 2024

HTE OS is a free, open-source high-throughput experimentation workflow that supports practitioners from experiment submission all the way to results presentation. A core Google Sheet responsible for reaction planning and execution as well communication with users robots. All generated data are funneled into Spotfire where analyze it. Tools parsing of LCMS translation chemical identifiers provide data-wrangling capabilities complete workflow.

Язык: Английский

---

Harnessing Electro-Descriptors for Mechanistic and Machine Learning Analysis of Photocatalytic Organic Reactions

Journal of the American Chemical Society, Год журнала: 2024, Номер 146(28), С. 19019 - 19029

Опубликована: Июль 4, 2024

Photocatalysis has emerged as an effective tool for addressing the contemporary challenges in organic synthesis. However, trial-and-error-based screening of feasible substrates and optimal reaction conditions remains time-consuming potentially expensive industrial practice. Here, we demonstrate electrochemical-based data-acquisition approach that derives a simple set redox-relevant electro-descriptors mechanistic analysis performance evaluation through machine learning (ML) photocatalytic These correlate to quantification shifted charge transfer processes response photoirradiation enabled construction reactivity diagram where high-yield reactive "hot zones" can reflect subtle changes system. For model deoxygenation reaction, influence varying carboxylic acids (substrate A, oxidation-intended) alkenes B, reduction-intended) on yield be visualized, while mathematical electro-descriptor patterns further revealed distinct mechanistic/kinetic impacts from different conditions. Additionally, application ML algorithms, experimentally derived overall redox kinetic outcome contributed vast parameters, serving capable means reduce dimensionality case complex multiparameter chemical space. As result, utilization efficient robust quantitative reactivity, demonstrating promising potential introducing operando-relevant experimental insights data-driven chemistry.

Язык: Английский

---