Predicting miRNA targets for head and neck squamous cell carcinoma using an ensemble method

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• Head and neck squamous cell carcinoma (HNSCC) is a significant health concern.
• Identifying specific microRNA (miRNA) targets is crucial for understanding HNSCC pathogenesis.
• Existing methods for miRNA target prediction have limitations.

Purpose of the Study:

• To develop and validate an ensemble method for predicting miRNA targets in HNSCC.
• To identify novel miRNA-mRNA interactions and associated pathways in HNSCC.
• To provide potential biomarkers for HNSCC diagnosis and prognosis.

Main Methods:

• An ensemble approach combining Pearson's correlation coefficient (PCC), Lasso, and intervention calculus without a directed acyclic graph (IDA) was employed.
• The Borda count election method integrated predictions from individual methods.
• Target gene validation utilized multiple databases (TarBase, miRecords, miRWalk, miRTarBase) and pathway enrichment analysis (KEGG).

Main Results:

• The study identified 33 miRNAs and 288 mRNAs with high correlation (absolute PCC >0.7).
• Ten target genes were extracted with a frequency of occurrence ≥3, with MYO5C showing the highest prediction frequency (7 miRNAs).
• Eight significant KEGG pathways were identified, notably cytokine-cytokine receptor interaction and chemokine signaling pathways.

Conclusions:

• The developed ensemble method effectively predicts miRNA targets and pathways in HNSCC.
• The findings offer valuable insights into HNSCC biology and potential diagnostic/prognostic markers.
• MYO5C and identified pathways represent promising areas for further HNSCC research.