Development of a prognostic risk model of uveal melanoma based on N7-methylguanosine-related regulators.

BACKGROUND: Uveal melanoma (UVM) stands as the predominant type of primary intraocular malignancy among adults. The clinical significance of N7-methylguanosine (m7G), a prevalent RNA modifications, in UVM remains unclear. METHODS: Primary information from 80 UVM patients were analyzed as the training set, incorporating clinical information, mutation annotations and mRNA expression obtained from The Cancer Genome Atlas (TCGA) website. The validation set was carried out using Gene Expression Omnibus (GEO) database GSE22138 and GSE84976. Kaplan-Meier and Cox regression of univariate analyses were subjected to identify m7G-related regulators as prognostic genes. RESULT: A prognostic risk model comprising EIF4E2, NUDT16, SNUPN and WDR4 was established through Cox regression of LASSO. Evaluation of the model's predictability for UVM patients' prognosis by Receiver Operating Characteristic (ROC) curves in the training set, demonstrated excellent performance Area Under the Curve (AUC) > 0.75. The high-risk prognosis within the TCGA cohort exhibit a notable worse outcome. Additionally, an independent correlation between the risk score and overall survival (OS) among UVM patients were identified. External validation of this model was carried out using the validation sets (GSE22138 and GSE84976). Immune-related analysis revealed that patients with high score of m7G-related risk model exhibited elevated level of immune infiltration and immune checkpoint gene expression. CONCLUSION: We have developed a risk prediction model based on four m7G-related regulators, facilitating effective estimate UVM patients' survival by clinicians. Our findings shed novel light on essential role of m7G-related regulators in UVM and suggest potential novel targets for the diagnosis, prognosis and therapy of UVM.

Disentangled Generation With Information Bottleneck for Enhanced Few-Shot Learning.

Few-shot learning (FSL) poses a significant challenge in classifying unseen classes with limited samples, primarily stemming from the scarcity of data. Although numerous generative approaches have been investigated for FSL, their generation process often results in entangled outputs, exacerbating the distribution shift inherent in FSL. Consequently, this considerably hampers the overall quality of the generated samples. Addressing this concern, we present a pioneering framework called DisGenIB, which leverages an Information Bottleneck (IB) approach for Disentangled Generation. Our framework ensures both discrimination and diversity in the generated samples, simultaneously. Specifically, we introduce a groundbreaking Information Theoretic objective that unifies disentangled representation learning and sample generation within a novel framework. In contrast to previous IB-based methods that struggle to leverage priors, our proposed DisGenIB effectively incorporates priors as invariant domain knowledge of sub-features, thereby enhancing disentanglement. This innovative approach enables us to exploit priors to their full potential and facilitates the overall disentanglement process. Moreover, we establish the theoretical foundation that reveals certain prior generative and disentanglement methods as special instances of our DisGenIB, underscoring the versatility of our proposed framework. To solidify our claims, we conduct comprehensive experiments on demanding FSL benchmarks, affirming the remarkable efficacy and superiority of DisGenIB. Furthermore, the validity of our theoretical analyses is substantiated by the experimental results. Our code is available at https://github.com/eric-hang/DisGenIB.