Deep Clustering and Visualization for End-to-End High-Dimensional Data Analysis

Identification of potential biomarkers in cardiovascular calcification based on bioinformatics combined with single-cell RNA-seq and multiple machine learning analysis

BACKGROUND

The molecular and genetic mechanisms underlying vascular calcification remain unclear. This study aimed to determine the differences in calcification marker-related gene expression in macrophages.

METHODS

The expression profiling datasets GSE104140 and GSE235995 were analysed to identify differentially expressed genes (DEGs) between fibroatheroma with calcification and diffuse intimal thickening. Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses, Weighted Gene Co-expression Network Analysis (WGCNA), and Gene Set Enrichment Analysis (GSEA) were performed to assess functional characteristics. Hub genes were identified through a protein-protein interaction (PPI) network and machine learning approaches. Single-cell RNA sequencing data (GSE159677) validated the expression of calcification-related genes in macrophages, while Mendelian randomization analysis explored their potential causal relationship with coronary calcification. Further validation was conducted using enzyme-linked immunosorbent assay (ELISA) on coronary calcification samples and immunohistochemistry in ApoE-/- mice. Intravascular ultrasound was performed to assess coronary calcification severity.

RESULTS AND CONCLUSIONS

Two key biomarkers, ITGAX and MYD88, were identified as diagnostic indicators of cardiovascular calcification. Both biomarkers were significantly upregulated in calcified samples and were strongly associated with immune processes. Single-cell RNA sequencing confirmed their high expression in multiple immune cell types. Additionally, molecular docking analysis revealed that retinoic acid interacted with both biomarkers, suggesting potential therapeutic relevance. Immunohistochemical and ELISA analyses further validated their elevated expression in calcified samples. These findings provide novel insights into the molecular mechanisms of vascular calcification and highlight potential diagnostic and therapeutic targets.

Learning clustering-friendly representations via partial information discrimination and cross-level interaction

Despite significant advances in the deep clustering research, there remain three critical limitations to most of the existing approaches. First, they often derive the clustering result by associating some distribution-based loss to specific network layers, neglecting the potential benefits of leveraging the contrastive sample-wise relationships. Second, they frequently focus on representation learning at the full-image scale, overlooking the discriminative information latent in partial image regions. Third, although some prior studies perform the learning process at multiple levels, they mostly lack the ability to exploit the interaction between different learning levels. To overcome these limitations, this paper presents a novel deep image clustering approach via Partial Information discrimination and Cross-level Interaction (PICI). Specifically, we utilize a Transformer encoder as the backbone, coupled with two types of augmentations to formulate two parallel views. The augmented samples, integrated with masked patches, are processed through the Transformer encoder to produce the class tokens. Subsequently, three partial information learning modules are jointly enforced, namely, the partial information self-discrimination (PISD) module for masked image reconstruction, the partial information contrastive discrimination (PICD) module for the simultaneous instance- and cluster-level contrastive learning, and the cross-level interaction (CLI) module to ensure the consistency across different learning levels. Through this unified formulation, our PICI approach for the first time, to our knowledge, bridges the gap between the masked image modeling and the deep contrastive clustering, offering a novel pathway for enhanced representation learning and clustering. Experimental results across six image datasets demonstrate the superiority of our PICI approach over the state-of-the-art. In particular, our approach achieves an ACC of 0.772 (0.634) on the RSOD (UC-Merced) dataset, which shows an improvement of 29.7% (24.8%) over the best baseline. The source code is available at https://github.com/Regan-Zhang/PICI.