Cancer Segmentation by Entropic Analysis of Ordered Gene Expression Profiles

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Massive gene expression datasets present challenges in data processing and information extraction.
• Accurate classification of tumor versus normal tissues is crucial for cancer diagnosis and treatment.

Purpose of the Study:

• To apply entropic measures for discriminating between tumor and normal tissue samples using comprehensive gene expression data.
• To explore the utility of complexity-entropy diagrams and information distance for cancer classification.

Main Methods:

• Utilized entropic measures on whole gene expression datasets to classify samples.
• Generated complexity-entropy diagrams by ordering gene expression by pathways.
• Applied information distance analysis for further discrimination.

Main Results:

• Achieved high success rates in classifying tumor and normal samples across 13 cancer types.
• Successfully clustered tumor and normal samples using complexity-entropy diagrams for nine of the thirteen cancer types.
• Information distance analysis demonstrated effective discrimination between cancer types, not just tumor vs. normal.

Conclusions:

• Developed a novel procedure using entropic measures for tissue classification without prior gene identification or specific cancer models.
• The proposed methods show significant potential for broad application in various classification problems beyond cancer research.
• This approach offers a powerful tool for analyzing complex biological data and advancing cancer diagnostics.