An isomiR expression panel based novel breast cancer classification approach using improved mutual information

Area of Science:

• Biochemistry
• Genomics
• Cancer Biology

Background:

• Gene expression profiling is a common method for identifying breast cancer biomarkers but has limitations.
• IsomiRs, miRNA isoforms, play crucial roles in biological processes and cancer.
• The potential of isomiRs as biomarkers for breast cancer subtypes remains underexplored.

Purpose of the Study:

• To investigate isomiRs as superior biomarkers for classifying breast cancer subtypes.
• To explain molecular differences between breast cancer subtypes using isomiRs.
• To develop a novel method for identifying isomiR biomarkers.

Main Methods:

• A null hypothesis method was used to remove lowly expressed isomiRs from sequencing data.
• An improved mutual information-based feature selection method was developed to weigh isomiR expression importance.
• Support Vector Machine (SVM) classifier was employed to identify isomiR biomarkers for subtyping.

Main Results:

• A novel method successfully identified specific isomiRs for classifying breast cancer subtypes.
• The improved mutual information method accommodates continuous feature data and discrete labels, unlike traditional methods.
• IsomiR-based classification demonstrated higher effectiveness than established gene expression profiling.

Conclusions:

• IsomiRs serve as effective biomarkers for identifying breast cancer subtypes, offering insights into molecular mechanisms.
• IsomiR expression profiling provides a more effective classification of breast cancer subtypes than gene expression profiling.
• The developed method outperforms existing biomarker discovery techniques and has potential applications in other diseases.