Analysis of Expression Pattern of snoRNAs in Different Cancer Types with Machine Learning Algorithms

Area of Science:

• Molecular Biology
• Genomics
• Bioinformatics

Background:

• Small nucleolar RNAs (snoRNAs) are functional small RNAs involved in RNA modifications.
• Aberrant snoRNA expression is implicated in tumorigenesis through various regulatory mechanisms.
• snoRNAs influence methylation, pseudouridylation, and host gene expression patterns.

Purpose of the Study:

• To investigate the expression patterns of snoRNAs across eight major cancer types using The Cancer Genome Atlas (TCGA) data.
• To identify specific snoRNAs that can discriminate between different cancer types.
• To uncover quantitative recognition rules for snoRNA expression patterns in various cancers.

Main Methods:

• Utilized machine learning algorithms, including Monte Carlo Feature Selection (MCFS) for feature selection and Incremental Feature Selection (IFS) for optimal feature extraction.
• Employed Support Vector Machine (SVM) for classification and Matthew's Correlation Coefficient (MCC) for performance evaluation.
• Applied Johnson reducer and Repeated Incremental Pruning to Produce Error Reduction (RIPPER) for generating classification rules.

Main Results:

• Identified a set of discriminative snoRNAs (e.g., HBII-52-14, SNORD123, U3) capable of predicting cancer types with high accuracy (MCC of 0.881).
• Developed classification rules revealing distinct snoRNA expression patterns across different cancer types.
• Demonstrated the potential of specific snoRNAs as biomarkers for cancer identification.

Conclusions:

• Specific snoRNAs and their expression patterns are crucial for distinguishing between major cancer types.
• Machine learning approaches effectively identify discriminative snoRNAs and elucidate their roles in cancer.
• Quantitative recognition rules derived from snoRNA expression can aid in cancer diagnostics.