Junrui Wu

BACKGROUND: The flavor of shrimp paste, a crucial quality indicator, relies on its non‐volatile taste components, which have not been fully characterized. This study aimed to comprehensively evaluate the taste profiles of shrimp paste and identify key taste‐enhancing components. RESULTS: A multifactor analysis model integrating taste indicators such as free amino acids, organic acids, 5′‐nucleotides, and taste values was established to thoroughly evaluate the taste profiles of 34 shrimp paste samples. Fourteen samples with favorable comprehensive scores were selected, among which peptides and their related metabolites were found to be more abundant than other flavor components through untargeted metabolomics, suggesting a potential significant role of peptides in enhancing the flavor profile of shrimp paste. Subsequently, the sample with the highest score underwent peptidomics analysis, leading to the identification of 5969 peptides. Four taste peptides were selected as potential taste‐enhancing candidates via virtual screening. Sensory evaluation demonstrated strong umami properties for all four synthetic peptides. Notably, the taste peptide DVFDL was identified as a taste‐enhancing candidate with pronounced umami and multi‐dimensional taste characteristics and no off‐flavors, which positively influenced the sensory characteristics of complex food systems by notably enhancing umami, saltiness, and richness. CONCLUSION: This study reveals for the first time that taste peptides are primary flavor compounds in shrimp paste and a novel taste peptide, DVFDL, exhibiting distinct taste attributes, has been identified. DVFDL shows potential as a fermentation indicator for quality consistency and as a natural flavor enhancer, although its practical application requires further investigation. © 2026 Society of Chemical Industry.

Taraxacum mongolicum contains various bioactive components. This study investigated the potential of thermostable microRNAs (miRNAs) derived from this plant as cross-kingdom regulators and evaluated their anti-proliferative effects on HepG2 cells. High-throughput sequencing was used to screen thermostable miRNAs and predict their target genes. Enrichment and target gene expression in gavaged mice were analyzed to assess cross-kingdom regulatory potential. Cellular assays evaluated anti-proliferative activity. miR166 was identified as a thermostable miRNA in T. mongolicum, and bioinformatics indicated its possible involvement in cancer-associated mucin-type O-glycan biosynthesis. RT-qPCR revealed that miR166 was negatively correlated with signal transducing adaptor family member 2 (STAP2) mRNA expression. In mice, administration of total RNA extracted from fresh T. mongolicum or its water decoction led to significant changes in miR166 levels in stomach, blood, liver, spleen, and intestine, accompanied by reduced STAP2 mRNA, suggesting cross-kingdom regulation. Overexpression of miR166 induced a delay in G1/S transition in HepG2 cells in a concentration-dependent manner, accompanied by reduced STAP2 expression and decreased cell viability. These findings provide preliminary experimental evidence that thermostable miR166 from T. mongolicum is associated with inhibition of HepG2 proliferation. They also suggest STAP2 as a candidate regulatory target, offer a potential new molecule for further investigation in liver cancer intervention, and provide a preliminary molecular basis for the medicinal and edible value of T. mongolicum.

Cancer has become a major global health threat. Despite advances in modern medicine, current therapeutic strategies still face many limitations. Anticancer peptides (ACPs), due to their high selectivity, low toxicity, and multitarget effects, have gradually become a research focus in the development of novel peptide‐based anticancer drugs. However, traditional screening methods are constrained by their low efficiency, high costs, and technical complexity, limiting their capacity to meet the demands of high‐throughput applications. Artificial intelligence (AI) has provided new methods to address these challenges, significantly improving the efficiency and accuracy of ACP screening through the application of machine learning and deep learning algorithms. To further enhance the application of AI in ACP screening, the advantages and limitations of 68 AI models used for ACP screening are systematically summarized. AI models show considerable potential for discovering ACPs, but most of these models lack interpretability and wet‐laboratory validation, which hinder the credibility and practical effectiveness of AI‐based ACP screening. Therefore, we presented a comprehensive ACP screening framework based on AI models. The presented framework includes data collection and organization, feature extraction, model construction, model interpretability analysis, and experimental validation. Additionally, we integrated this screening framework with multi‐omics and other biotechnologies to promote the translation of AI‐selected ACPs to the clinic. The presented AI‐based ACP screening framework can accelerate the ACP development, increase ACP screening efficiency, and promote clinical ACP application.