Uncovering hub genes and immunological characteristics for heart failure utilizing RRA, WGCNA and Machine learning.

Background: Heart failure (HF) is a major public health issue with high mortality and morbidity. This study aimed to find potential diagnostic markers for HF by the combination of bioinformatics analysis and machine learning, as well as analyze the role of immune infiltration in the pathological process of HF.

Methods: The gene expression profiles of 124 HF patients and 135 nonfailing donors (NFDs) were obtained from six datasets in the NCBI Gene Expression Omnibus (GEO) public database. We applied robust rank aggregation (RRA) and weighted gene co-expression network analysis (WGCNA) method to identify critical genes in HF. To discover novel diagnostic markers in HF, three machine learning methods were employed, including best subset regression, regularization technique, and support vector machine-recursive feature elimination (SVM-RFE). Besides, immune infiltration was investigated in HF by single-sample gene set enrichment analysis (ssGSEA).

Results: Combining RRA with WGCNA method, we recognized 39 critical genes associated with HF. Through integrating three machine learning methods, FCN3 and SMOC2 were determined as novel diagnostic markers in HF. Differences in immune infiltration signature were also found between HF patients and NFDs. Moreover, we explored the potential associations between two diagnostic markers and immune response in the pathogenesis of HF.

Conclusions: In summary, FCN3 and SMOC2 can be used as diagnostic markers of HF, and immune infiltration plays an important role in the initiation and progression of HF.