Deep Learning algorithms have recently received a growing interest to learn from examples of existing solutions and some accurate approximations of the solution of complex physical problems, in particular relying on Graph Neural Networks applied on a mesh of the domain at hand. On the other hand, state-of-the-art deep approaches of image processing use different resolutions to better handle the different scales of the images, thanks to pooling and up-scaling operations. But no such operators can be easily defined for Graph Convolutional Neural Networks (GCNN). This paper defines such operators based on meshes of different granularities. Multi-resolution GCNNs can then be defined. We propose the MGMI approach, as well as an architecture based on the famed U-Net. These approaches are experimentally validated on a diffusion problem, compared with projected CNN approach and the experiments witness their efficiency, as well as their generalization capabilities.

Copper-based catalysts have been recognized as promising candidates for electrochemical conversion of CO2 to value-added chemicals and synthetic fuels. Yet, the challenges of high overpotential and low product selectivity have motivated the rational electrode engineering. In the present work, we prepared CuS catalysts using different sulfur precursors, and we aimed to elucidate the precursor-dependent effect on their structure–property–activity relationships for electrochemical CO2 reduction. The different sulfur precursors exhibited varied S release rates in hydrothermal synthesis, which had induced distinct surface morphological features and diverse sulfur vacancy concentrations, and the intrinsic catalytic activity and product selectivity would be affected. The desired CuS-TU catalyst synthesized using thiourea as the sulfur precursor featured a flower-like morphology and had the highest sulfur vacancy concentration. The nanoflower morphology offered expanded space and considerable undercoordinated sites for facilitated interfacial mass transfer in electrochemical CO2 reduction. Density functional theory calculations confirmed that the abundant sulfur vacancy played an important role in strengthening the adsorption of the *COOH intermediates on the surface, which promoted CO production via the *COOH pathway. The CuS-TU catalyst therefore exhibited a relatively higher CO selectivity of 72.67% at −0.51 V vs RHE. These findings will provide more insights into improving the electrochemical CO2 reduction performance of copper-based catalysts by structure engineering.

No abstract

Significance Brain metastasis with current limited treatment options is a common complication in advanced cancer patients, and breast-to-brain metastasis (B2BM) is one of the major types. In this work, we report that brain metastasis oncogenic long noncoding RNA (BMOR) is a key brain-enriched long noncoding RNA for the development of B2BM. We demonstrate that BMOR allows B2BM cells to colonize the brain tissue by evading immune-mediated killing in the brain microenvironment. At the molecular level, BMOR binds and inactivates IRF3 in B2BM cells. Finally, BMOR silencer can effectively suppress the development of brain metastasis in vivo. Therefore, our findings reveal a way in which cancer cells evade immune-mediated killing in the brain microenvironment for brain metastasis development and establish therapeutic targets with potential targeted strategies against B2BM.