Identifying and analyzing different cancer subtypes using RNA-seq data of blood platelets

Area of Science:

• Biotechnology
• Genomics
• Machine Learning

Background:

• Traditional cancer detection is slow and costly.
• Liquid biopsy offers a noninvasive, accurate, and cost-effective alternative.
• Analyzing liquid biopsy gene expression data is crucial for improving cancer subtype detection.

Purpose of the Study:

• To apply machine learning algorithms to gene expression data from liquid biopsies.
• To identify optimal gene subsets for discriminating cancer subtypes and healthy controls.
• To evaluate the performance of a machine learning model for cancer detection.

Main Methods:

• Quantitative gene expression profiles were analyzed using machine learning.
• Maximum Relevance Minimum Redundancy (mRMR) and incremental feature selection were employed.
• Support Vector Machine (SVM) algorithm was used for classification with a ten-fold cross-validation.

Main Results:

• An optimal feature subset was extracted, leading to a classification model with 75.1% overall accuracy.
• Eighteen key genes, including TTN, RHOH, RPS20, and TRBC2, were identified as potential biomarkers.
• These genes demonstrated effectiveness in discriminating between different cancer subtypes and healthy individuals.

Conclusions:

• Machine learning analysis of liquid biopsy gene expression data can effectively detect cancer subtypes.
• Identified genes like TTN, RHOH, RPS20, and TRBC2 show promise as diagnostic biomarkers.
• This approach facilitates early cancer prevention and personalized treatment strategies.