Property-guided few-shot learning for molecular property prediction with dual-view encoder and relation graph learning network.

Molecular property prediction is an important task in drug discovery. However, experimental data for many drug molecules are limited, especially for novel molecular structures or rare diseases which affect the accuracy of many deep learning methods that rely on large training datasets. To this end, we propose PG-DERN, a novel few-shot learning model for molecular property prediction. A dual-view encoder is introduced to learn a meaningful molecular representation by integrating information from node and subgraph. Next, a relation graph learning module is proposed to construct a relation graph based on the similarity between molecules, which improves the efficiency of information propagation and the accuracy of property prediction. In addition, we use a MAML-based meta-learning strategy to learn well-initialized meta-parameters. In order to guide the tuning of meta-parameters, a property-guided feature augmentation module is designed to transfer information from similar properties to the novel property to improve the comprehensiveness of the feature representation of molecules with novel property. A series of comparative experiments on four benchmark datasets demonstrate that the proposed PG-DERN outperforms state-of-the-art methods.

Drug-target binding affinity prediction using message passing neural network and self supervised learning.

BACKGROUND: Drug-target binding affinity (DTA) prediction is important for the rapid development of drug discovery. Compared to traditional methods, deep learning methods provide a new way for DTA prediction to achieve good performance without much knowledge of the biochemical background. However, there are still room for improvement in DTA prediction: (1) only focusing on the information of the atom leads to an incomplete representation of the molecular graph; (2) the self-supervised learning method could be introduced for protein representation. RESULTS: In this paper, a DTA prediction model using the deep learning method is proposed, which uses an undirected-CMPNN for molecular embedding and combines CPCProt and MLM models for protein embedding. An attention mechanism is introduced to discover the important part of the protein sequence. The proposed method is evaluated on the datasets Ki and Davis, and the model outperformed other deep learning methods. CONCLUSIONS: The proposed model improves the performance of the DTA prediction, which provides a novel strategy for deep learning-based virtual screening methods.

In situ analysis of surface composition and meteorology at the Zhurong landing site on Mars.

The Zhurong rover of the Tianwen-1 mission landed in southern Utopia Planitia, providing a unique window into the evolutionary history of the Martian lowlands. During its first 110 sols, Zhurong investigated and categorized surface targets into igneous rocks, lithified duricrusts, cemented duricrusts, soils and sands. The lithified duricrusts, analysed by using laser-induced breakdown spectroscopy onboard Zhurong, show elevated water contents and distinct compositions from those of igneous rocks. The cemented duricrusts are likely formed via water vapor-frost cycling at the atmosphere-soil interface, as supported by the local meteorological conditions. Soils and sands contain elevated magnesium and water, attributed to both hydrated magnesium salts and adsorbed water. The compositional and meteorological evidence indicates potential Amazonian brine activities and present-day water vapor cycling at the soil-atmosphere interface. Searching for further clues to water-related activities and determining the water source by Zhurong are critical to constrain the volatile evolution history at the landing site.

Self-Adaptive Electronic Structure of Amphoteric Conjugated Ligand-Modified 3 d Metal-C3 N4 Smart Electrocatalyst by pH Self-Response Realizing Electrocatalytic Self-Adjustment.

Constructing pH-responsive smart material provides a new opportunity to address the problem that traditional electrocatalysts cannot achieve both alkaline oxygen evolution reaction (OER) and acidic hydrogen evolution reaction (HER) activities. In this study, amphoteric conjugated ligand (2-aminoterephthalic acid, BDC-NH2 )-modified 3d metal-anchored graphitic carbon nitride (3d metal-C3 N4 ) smart electrocatalysts are constructed, and self-adaptation of the electronic structure is realized by self-response to pH stimulation, which results in self-adjustment of alkaline OER and acidic HER. Specifically, the amino and carboxyl functional groups in BDC-NH2 undergo protonation and deprotonation respectively under different pH stimulation to adapt to environmental changes. Through DFT calculations, the increase or decrease of electron delocalization range brought by the self-response characteristic is found to lead to redistribution of the Bader charge around the modified active sites. The OER and HER activities are greatly promoted roughly 4.8 and 8.5 times over Co-C3 N4 after BDC-NH2 -induced self-adaptive processes under different environments, arising from the reduced energy barrier of O* to OOH* and ΔGH* . Impressively, the proposed BDC-NH2 -induced smart regulation strategy is applicable to a series of 3d metal anchors for C3 N4 , including Co, Ni and Fe, providing a general structural upgrading method for constructing smart electrocatalytic systems.

"Electron Complementation"-Induced Molybdenum Nitride/Co-Anchored Graphitic Carbon Nitride Porous Nanoparticles for Efficient Overall Water Splitting.

Maximizing the usable space of electrocatalysts and fine-tuning the interface geometry as well as the electronic structure to facilitate hydrogen and oxygen evolution reactions (HER and OER) have always been the focus of research. Herein, a homogeneous porous nanoparticle construction strategy was proposed, in which molybdenum nitride (Mo2N) particles were prepared by controlled heat treatment of the precursor nanoparticle induced by polyethylene glycol, and the Mo2N/Co-C3N4 heterostructure with a pore size of about 1.13 nm was obtained by compounding Co-anchored graphitic carbon nitride. In particular, exploring the change of charge distribution at the interface based on the principle of "electron complementation" shows that under the regulation of nitrogen with high electronegativity, the affinity of active site Co to oxygenated species in the OER process and the adsorption as well as cleavage ability of HER reactants in the active site were effectively optimized. Thus, Mo2N/Co-C3N4 not only inherits the functions of each component, but also provides an ideal heterogeneous interface for exhibiting impressive bifunctional activity, which only needs 100 and 210 mV to deliver 10 mA cm-2 for the HER and OER, respectively. In addition, the Mo2N/Co-C3N4 catalyst also demonstrates high overall water splitting stability with a slight current decrease after 95 h. Manipulating the electronic structure of multiple sites by constructing electronically complementary interfaces may provide another avenue to develop highly active catalysts for overall water splitting and other applications.

Controlling the D-band for improved oxygen evolution performance in Ni modulated ultrafine Co nanoparticles embedded in Nitrogen-doped carbon microspheres.

Despite the challenges on tuning the d-band structure of transition metals, the d-band is of great importance for promoting the interaction between catalytic and intermediates during the oxygen evolution reaction (OER) process. Herein, ultrafine Co nanoparticles embedded in the surface layer of nitrogen-doped carbon microspheres are prepared through an in-situ co-coordination strategy, and its d-band is modulated by introducing different Ni amounts. The introduction of Ni in the Co crystal lattice can tune the d-band center and unpaired electrons, which collectively result in an enhancement of OER activity and kinetics. By investigating the catalysts with Ni content from 0% to 75%, it is concluded that the catalyst with 25% Ni shows optimal OER activity, lower overpotential (285 mV at 10 mA cm-2) and higher current densities (73.75 mA cm-2 at 1.63 V). Moreover, the good stability is also demonstrated with the negligible decrease on current densities after 3000 CV cycles or 100 h of continuous test in alkaline media. This concept of modulating the d-band structure by introducing a transition metal with different contents in another transition metal crystal lattice could present an alternative pathway to the development of highly active catalytic materials for OER and beyond.

Transferable Active Centers of Strongly Coupled MoS2@Sulfur and Molybdenum Co-doped g-C3N4 Heterostructure Electrocatalysts for Boosting Hydrogen Evolution Reaction in Both Acidic and Alkaline Media.

Designing an excellent acidic and alkaline general-purpose hydrogen evolution electrocatalyst plays an important role in promoting the development of the energy field. Here, a feasible strategy is reported to use the strongly coupled MoS2@sulfur and molybdenum co-doped g-C3N4 (MoS2@Mo-S-C3N4) heterostructure with transferable active centers for catalytic reactions in acidic and alkaline media. Research studies have shown that the unsaturated S site at the edge of MoS2 and the active N atom on the Mo-S-C3N4 substrate are, respectively, the active centers of acidic and alkaline hydrogen evolution reaction. Specifically, Mo-S-C3N4 is regarded as a synergistic catalyst for the active species MoS2 in acidic hydrogen evolution, while MoS2 acts as a co-catalyst when the alkaline active species are transferred to Mo-S-C3N4. The coordination of the electrons between the interfaces achieves a synergistic balance, which provides the optimal sites for the adsorption of the reactants. Such an electrocatalyst exhibits overpotentials of 193 and 290 mV at 10 mA cm-2 in 0.5 M H2SO4 and 1 M KOH, respectively, which was better than numerous previous reports. This research provides an outstanding avenue to realize multifunctional electrocatalysts.

Hyper-Rayleigh scattering in a strong coupling microcavity waveguide.

Second-harmonic generation (SHG) from hyper-Rayleigh scattering (HRS) in a hybrid strong coupling microcavity waveguide (HSCMW) was demonstrated, which indicates a possible method using continuous-wave (cw) incident light. The cw light was coupled into the waveguide with high coupling efficiency by free space coupling technology, and then the electric field intensity of the fundamental wave was enhanced due to local oscillation. HRS occurred by lithium niobite (LN) powder inside the waveguide, resulting in the direct observation of SHG in the transverse direction, with relatively high conversion efficiency measured to be 0.032%/W. This work suggests progress on frequency conversion and is also applicable to other nonlinear processes in a waveguide.