Yan Liu’s research while affiliated with Peking University School of Stomatology and other places

This study investigates the effects of aluminum (Al) and titanium (Ti) additions on the porosity, microstructure, and corrosion performance of Hastelloy C276-based coatings fabricated via laser cladding on nodular cast iron substrates. Nickel-based alloy powders blended with varying Ti (1–10 wt.%) and Al (0.5–2.5 wt.%) contents were deposited under optimized laser parameters. Microstructural characterization revealed that Ti addition refined the grain structure and promoted the formation of TiC phases, while Al addition dispersed eutectic networks into isolated island-like structures. Both elements effectively suppressed porosity by reducing gas entrapment during solidification. However, excessive Ti (10 wt.%) induced brittle fracture due to TiC agglomeration, and Al addition caused interfacial cracks owing to Al2O3 formation. Electrochemical tests in a 3.5 wt.% NaCl solution indicated that Al/Ti additions enhanced initial passivation but reduced corrosion resistance due to weakened oxide film stability. XPS analysis revealed that Al-enriched coatings formed Al2O3 and Al(OH)3, whereas Ti-modified coatings developed TiO2 and TiC, both influencing the passivation behavior. These findings provide critical insights into tailoring laser-clad coatings for marine applications by balancing porosity suppression and corrosion resistance.

We introduce EvaLearn, a pioneering benchmark designed to evaluate large language models (LLMs) on their learning capability and efficiency in challenging tasks, a critical, yet underexplored aspect of model potential. EvaLearn contains 648 challenging problems across six task types, grouped into 182 sequences, each sequence dedicated to one task type. Diverging from most existing benchmarks that evaluate models in parallel, EvaLearn requires models to solve problems sequentially, allowing them to leverage the experience gained from previous solutions. EvaLearn provides five comprehensive automated metrics to evaluate models and quantify their learning capability and efficiency. We extensively benchmark nine frontier models and observe varied performance profiles: some models, such as Claude-3.7-sonnet, start with moderate initial performance but exhibit strong learning ability, while some models struggle to benefit from experience and may even show negative transfer. Moreover, we investigate model performance under two learning settings and find that instance-level rubrics and teacher-model feedback further facilitate model learning. Importantly, we observe that current LLMs with stronger static abilities do not show a clear advantage in learning capability across all tasks, highlighting that EvaLearn evaluates a new dimension of model performance. We hope EvaLearn provides a novel evaluation perspective for assessing LLM potential and understanding the gap between models and human capabilities, promoting the development of deeper and more dynamic evaluation approaches. All datasets, the automatic evaluation framework, and the results studied in this paper are available at the GitHub repository.

Osmia spp. is a genus of solitary bees that serves as excellent pollinators for various fruit trees and has the potential to enhance pollination services in both agricultural and natural ecosystems. However, the absence of high-quality genomic resources limits our insights into evolutionary biology and ecological adaptations of Osmia. Here, we present a chromosome-level genome of Osmia excavata, using PacBio, Illumina, and Hi-C data. The genome has a total size of 164.35 Mb, with a scaffold N50 of 9.81 Mb, and the majority of contigs (98.50%, 161.88 Mb) are organized into sixteen chromosomes. BUSCO analysis reveals a completeness score of 99.7% (n = 1,367), with 99.6% identified as single-copy BUSCOs and 0.1% as duplicated BUSCOs. The genome contains 13.46% (22.11 Mb) repetitive elements and encodes 11,452 predicted protein-coding genes. This study provides a crucial genomic resource for our understanding of solitary bees’ evolution and ecological roles.

The incidence of stress‐induced gastritis is gradually increasing. Vitamin E (VE) is widely used in inflammatory diseases due to its efficient antioxidant and anti‐inflammatory effects. Here, we investigated the protective role of VE on stress‐induced gastritis and its potential mechanisms. Mice were subjected to high‐intensity stress caused by the forced swim test (FST) and gavaged with VE (300 mg/kg) at different time points. The results showed that VE significantly alleviated stress‐induced gastric mucosal injury and related histopathological changes. Additionally, the expression levels of nuclear factor erythroid 2‐related factor 2 (Nrf2), haem oxygenase‐1 (Hmox1) and NAD(P)H:quinone oxidoreductase 1 (NQO1) were upregulated in the administrated groups, while the nuclear factor kappa B (NF‐κB) signalling pathway was inhibited, manifested as the expression level of p‐NF‐κB p65 protein decreased. Furthermore, VE reduced the infiltration of macrophages in gastric tissue, followed by a synchronous decrease in the expression level of interleukin‐1 beta (IL‐1β) protein. Importantly, the detection of cell death by TUNEL assay and SYTOX green staining demonstrated that VE reduced cell death of gastric tissue and subsequently downregulated the pro‐apoptotic factor BCL2‐associated X Protein (Bax). Hence, our study suggested that VE has an outstanding preventive and therapeutic effect on stress‐induced gastritis via promoting Nrf2 and inhibiting NF‐κB signalling.

Periodontal disease (PD), an inflammatory disease initiated by oral microbiota, may aggravate hypertension (HTN). Few studies were employed to characterize the oral microbiota in hypertensive patients with periodontitis. To investigate the interplay between oral microbiota and hypertension in individuals with periodontitis, we initiated a metagenomic sequencing study on subgingival plaque and saliva samples sourced from HTN patients and those with hypertension and periodontitis (PDHTN). Our primary objective was to characterize species serving as pivotal links (bridge species) in exacerbating hypertension induced by periodontal disease. Within subgingival plaque and saliva specimens, we pinpointed 31 and 28 bridge species, respectively. Furthermore, we noted a decrease in the abundance of nitrate-reducing bacteria, such as Actinomyces spp., Rothia spp., and Veillonella spp., in PDHTN samples. Employing network analysis, we distinguished distinct polymicrobial clusters within the two patient groups. These bridge species coalesced into polymicrobial clusters, revealing intricate symbiotic and competitive relationships. To substantiate our findings, we leveraged an angiotensin II-infused animal model of ligature-induced periodontitis (LIP), confirming the contributory role of Filifactor alocis —a selectively analyzed subgingival bridge species—in exacerbating hypertension and upregulating the frequency of renal CD4 ⁺ IFNγ ⁺ and CD8 ⁺ IFNγ ⁺ T cells. Our study screened a list of species linking PD and HTN. PD may aggravate HTN by decreasing the abundance of nitrate-reducing bacteria and increasing the abundance of pathogens. Using an animal model, we demonstrated that F. alocis aggravates HTN via the accumulation of IFNγ ⁺ T cells in the kidneys. IMPORTANCE Both periodontal disease and hypertension are widely prevalent all over the world. PD may aggravate the development of HTN via oral microbiota. However, few studies were employed to characterize the oral microbiota in hypertensive patients with periodontitis. Here, the present study profiled the oral microbiota in hypertensive participants with periodontitis. We found that the depleted abundance of nitrate-reducing bacteria and the enriched abundance of pathogens. Finally, we validated the role of Filifactor alocis in exacerbating HTN via infiltration of IFNγ ⁺ T cells in mice kidneys. Our study improved the understanding of oral microbiota linking PD and HTN.