Iridium-Catalyzed Selective O-H Bond Insertion Of Sulfoxonium Ylides For The Synthesis Of Carboxymethyl Ketones And Yield Prediction

Carboxymethyl ketones are a class of important organic structural skeletons,especially those containing heterocycles,which are widely found in natural products,drug molecules,and bioactive compounds.Currently,the transition metal catalyzed O–H bond insertion reactions between diazo compounds and carboxylic acids has attracted extensive attention as an efficiently method for preparing carboxymethyl ketones.However,due to the instability,toxicity,and potential explosion hazards of diazo compounds,as well as the difficulties in the preparation of diazo compounds,their application is limited.Therefore,the development new methods for the synthesis of carboxymethyl ketones from easily available starting materials is highly desired.With the continuous development of artificial intelligence and chemical informatics,machine learning has gradually developed into an effective tool for predicting the selectivity and yield of organic reactions.Using this tool,it is possible to predict the yield of some unknown reactions.In this paper,2-（1H-indol-3-yl）acetic acid and sulfoxonium ylide were used as model substrates,factors affecting the reaction,such as catalyst,ligand,solvent,and temperature,were screened through a high throughput experimentation（HTE）platform.The selective O–H bond insertion reactions of sulfoxonium ylides with carboxylic acids for the synthesis of carboxymethyl ketones.Through the optimal reaction conditions obtained,heterocyclic carboxylic acids containing different functional group substituents were investigated.In order to deeply explore the substrate range of carboxylic acids in this reaction,a large amount of reaction data was collected using high throughput experiments.Using reaction data as learning samples,the yield prediction of the selective O–H bond insertion reaction between sulfoxonium ylides and carboxylic acids toward carboxymethyl ketones was achieved by machine learning.The main innovations of this study include:1)A highly selective O–H bond insertion reaction of carboxylic acids with sulfoxonium ylides via iridium（I）-catalyzed has been developed.For heterocyclic carboxylic acids containing different functional group substituents（indole,pyrrole,pyridine,pyrimidine,quinoline,benzoxazole,benzofuran,benzothiophene,etc.）,the desired target product can be obtained.Moreover,the unprotected OH-and NH₂-substituted heterocycles can also react smoothly to obtain the corresponding target product in a medium to higher yield.More importantly,phenolic hydroxyl groups,primary alcohols,secondary amines,olefins,alkyne substituted aryl or alkyl carboxylic acids,and various complex biologically active carboxylic acids can also react well.In addition,disubstituted sulfoxonium ylides are also good substrates.Finally,all synthesized compounds were conducted PLpro/3CLpro enzyme activity tests.2)A standardized reaction dataset was quickly obtained using a high-throughput experimental platform,and a yield prediction model based on deep learning（Pan Gu Fine-tuned model）was established.This model can predict the reaction yields of diverse unseen reaction substrates and explore reaction space for this reaction.