GraphEGFR: Multi-task and transfer learning based on molecular graph attention mechanism and fingerprints improving inhibitor bioactivity prediction for EGFR family proteins on data scarcity.

The proteins within the human epidermal growth factor receptor (EGFR) family, members of the tyrosine kinase receptor family, play a pivotal role in the molecular mechanisms driving the development of various tumors. Tyrosine kinase inhibitors, key compounds in targeted therapy, encounter challenges in cancer treatment due to emerging drug resistance mutations. Consequently, machine learning has undergone significant evolution to address the challenges of cancer drug discovery related to EGFR family proteins. However, the application of deep learning in this area is hindered by inherent difficulties associated with small-scale data, particularly the risk of overfitting. Moreover, the design of a model architecture that facilitates learning through multi-task and transfer learning, coupled with appropriate molecular representation, poses substantial challenges. In this study, we introduce GraphEGFR, a deep learning regression model designed to enhance molecular representation and model architecture for predicting the bioactivity of inhibitors against both wild-type and mutant EGFR family proteins. GraphEGFR integrates a graph attention mechanism for molecular graphs with deep and convolutional neural networks for molecular fingerprints. We observed that GraphEGFR models employing multi-task and transfer learning strategies generally achieve predictive performance comparable to existing competitive methods. The integration of molecular graphs and fingerprints adeptly captures relationships between atoms and enables both global and local pattern recognition. We further validated potential multi-targeted inhibitors for wild-type and mutant HER1 kinases, exploring key amino acid residues through molecular dynamics simulations to understand molecular interactions. This predictive model offers a robust strategy that could significantly contribute to overcoming the challenges of developing deep learning models for drug discovery with limited data and exploring new frontiers in multi-targeted kinase drug discovery for EGFR family proteins.

Hydroxy and surface oxygen effects on 5-hydroxymethylfurfural oxidation to 2,5-furandicarboxylic acid on ß-MnO2: DFT, microkinetic and experiment studies.

Manganese dioxide, ß-MnO2, has shown potential in catalyzing the oxidation of 5-hydroxymethylfurfural (HMF) to 2,5-furandicarboxylic acid (FDCA), a monomer of bioplastic polyethylene furanoate (PEF). Herein, the insight into the hydroxy (OH) and surface oxygen effects on the HMF-to-FDCA reaction over ß-MnO2 is clarified through a comprehensive investigation using density functional theory (DFT) calculations, microkinetic modeling, and experiment. Theoretical analyses revealed that both active surface oxygen and OH species (from either base or solvent) facilitate C-H bond breaking and OH insertion, promoting the catalytic activity of ß-MnO2. Microkinetic modeling demonstrated that the FFCA-to-FDCA and DFF-to-FFCA steps are the rate-limiting steps of the hydroxylated and non-hydroxylated surfaces, respectively. These theoretical results agree well with the experiment when water and dimethyl sulfoxide (DMSO) were used as solvents. In addition, the synthesized ß-MnO2 catalyst showed high stability and activity, maintaining stable HMF conversion (≥99 mol%) and high FDCA yield (85-92 mol%) during continuous flow oxidation for 72 hours at pO2 of 1 MPa, 393 K and LHSV of 1 h-1. Thus, considering both hydroxy and surface oxygen species is a new strategy for enhancing the catalytic activity of Mn oxides and other metal oxide catalysts for the HMF-to-FDCA reaction.

Efficient exploration of transition-metal decorated MXene for carbon monoxide sensing using integrated active learning and density functional theory.

The urgent demand for chemical safety necessitates the real-time detection of carbon monoxide (CO), a highly toxic gas. MXene, a 2D material, has shown potential for gas sensing applications (e.g., NH3, NO, SO2, CO2) due to its high surface accessibility, electrical conductivity, stability, and flexibility in surface functionalization. However, the pristine MXene generally exhibits poor interaction with CO; still, transition metal decoration can strengthen the interaction between CO and MXene. This study presents a high-throughput screening of 450 combinations of transition-metal (TM) decorated MXene (TM@MXene) for CO sensing applications using an integrated active learning (AL) and density functional theory (DFT) screening pipeline. Our AL pipeline, adopting a crystal graph convolutional neural network (CGCNN) as a surrogate model, successfully accelerates the screening of CO sensor candidates with minimal computational resources. This study identifies Sc@Zr3C2O2 and Y@Zr3C2O2 as the optimal TM@MXene candidates with promising CO sensing performance regarding the screening criteria of recovery time, surface stability, charge transfer, and sensitivity to CO. The proposed AL framework can be extended for property finetuning in the combinatorial chemical space.

Data-Driven Discovery of Graphene-Based Dual-Atom Catalysts for Hydrogen Evolution Reaction with Graph Neural Network and DFT Calculations.

The flexible tuning ability of dual-atom catalysts (DACs) makes them an ideal system for a wide range of electrochemical applications. However, the large design space of DACs and the complexity in the binding motif of electrochemical intermediates hinder the efficient determination of DAC combinations for desirable catalytic properties. A crystal graph convolutional neural network (CGCNN) was adopted for DACs to accelerate the high-throughput screening of hydrogen evolution reaction (HER) catalysts. From a pool of 435 dual-atom combinations in N-doped graphene (N6Gr), we screened out two high-performance HER catalysts (AuCo@N6Gr and NiNi@N6Gr) with excellent HER, electronic conductivity, and stability using the combination of CGCNN and density functional theory (DFT). Furthermore, comprehensive DFT studies were conducted on these two catalysts to confirm their outstanding reaction kinetics and to understand the cooperative effect between the metal pair for HER. To obtain ideal hydrogen binding in AuCo, the inert Au weakens the strong hydrogen binding of Co, while for NiNi, the two weakly binding Ni cooperate. The present protocol was able to select the two catalysts with different physical origins for HER and can be applied to other DAC catalysts, which should hasten catalyst discovery.

Synthesis and characterization of novel chiral derivatizing agents containing ß-keto-anthracene adducts (KAAs) by 1H-NMR: aromatic influence and chiral alcohol absolute configuration determination.

New alternative chiral derivatizing agents, ß-keto-anthracene adducts (KAAs), were accomplished and the influence of aromatic moieties at the α-carbon position for elucidation of the absolute configuration of chiral secondary alcohols via NMR was studied. The α-benzoyl substituted KAAs strongly enhance the anisotropic effect which produced greater ΔδRS values than other conventional reagents. We propose a simplified model to describe the conformations and to assign the absolute configuration in several chiral alcohol samples.

Theoretical Insights on Solvent Control of Intramolecular and Intermolecular Proton Transfer of 2-(2'-Hydroxyphenyl)benzimidazole.

Excited-state proton transfer (ESPT) processes of 2-(2'-hydroxyphenyl)benzimidazole (HBI) and its complexation with protic solvents (H2O, CH3OH, and NH3) have been investigated by both static calculations and dynamics simulations using density functional theory (DFT) at B3LYP/TZVP theoretical level for ground state (S0) and time-dependent (TD)-DFT at TD-B3LYP/TZVP for excited state (S1). For static calculations, absorption and emission spectra, infrared (IR) vibrational spectra of O-H mode, frontier molecular orbitals (MOs), and potential energy curves (PECs) of proton transfer coordinate were analyzed. Simulated absorption and emission spectra show an agreement with available experimental data. The hydrogen bond strengthening in the S1 state has been proved by the changes of IR vibrational spectra and bond parameters of the hydrogen moiety with those of the S0 state. The MOs provide the visual electron density redistribution confirming the hydrogen bond strengthening mechanism. The PECs show that the proton transfer (PT) process is easier to occur in the S1 state than the S0 state. Moreover, on-the-fly dynamics simulations of all systems were carried out to provide the detailed information on time revolution. The results revealed that the excited-state intermolecular proton transfer for HBI is fast, whereas the excited-state intermolecular proton transfer for HBI with protic solvents are slower than that of HBI because the competition between intra- and intermolecular hydrogen-bonds between HBI and protic solvent. These intermolecular hydrogen-bonds hinder the formation of tautomer, hence explaining the low quantum yield found in the protic solvent experiment. Especially for HBI complexing with methanol, only ESIntraPT occurs with small probability compared to HBI with water and ammonia.