Calculation of immune cell proportion from batch tumor gene expression profile based on support vector regression

Area of Science:

• Immunology
• Bioinformatics
• Oncology

Background:

• The tumor microenvironment (TME) contains numerous immune cells crucial for cancer progression and patient outcomes.
• Accurate quantification of immune cell infiltration is vital for understanding tumor biology.

Purpose of the Study:

• To improve the accuracy and robustness of estimating immune cell proportions within the TME using RNA sequencing (RNA-seq) gene expression data.
• To analyze differences in immune cell fractions between colorectal cancer tissues and normal adjacent tissues.

Main Methods:

• Developed an improved algorithm for relative immune cell proportion estimation based on RNA-seq data.
• Utilized support vector regression (SVR) to solve the multiple linear regression model for enhanced accuracy.
• Applied the pipeline to analyze 41 pairs of colorectal cancer and normal solid tissue samples.

Main Results:

• The SVR-based algorithm demonstrated increased robustness and accuracy in calculating immune cell proportions.
• Colorectal cancer tissues exhibited a higher proportion of CD8 T cells and neutrophils compared to normal tissues.
• Conversely, B cells and monocytes were found in relatively lower proportions in tumor tissues.

Conclusions:

• The developed computational pipeline provides a reliable method for quantifying immune cell infiltration from gene expression profiles.
• Significant shifts in immune cell composition, including increased CD8 T cells and neutrophils, characterize the colorectal cancer TME.
• This tool is publicly available on GitHub for broader research application.