Multiple-cumulative probabilities used to cluster and visualize transcriptomes

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene expression data analysis is crucial for biological discovery.
• Clustering and visualization are key techniques for interpreting high-dimensional gene expression data.
• Existing methods face challenges with high-dimensional data and clear cluster visualization.

Purpose of the Study:

• To develop an improved method for clustering and visualizing gene expression data.
• To address the challenge of high-dimensional Pearson's correlation coefficient of multiple-cumulative probabilities (PCC-MCP) data.
• To enhance the understanding of relationships between gene expression clusters.

Main Methods:

• Utilized Pearson's correlation coefficient of multiple-cumulative probabilities (PCC-MCP) to define gene expression similarity.
• Employed icc-cluster, an iterative clustering algorithm, with PCC-MCP for gene grouping.
• Applied t-statistic stochastic neighbor embedding (t-SNE) of KC-data for optimal cluster mapping (t-SNE-MCP-O maps).

Main Results:

• Demonstrated clear advantages of icc-cluster with PCC-MCP over conventional clustering methods across multiple transcriptome datasets.
• Showcased t-SNE-MCP-O maps providing clear projecting boundaries for PCC-MCP clusters.
• Facilitated easy visualization and understanding of relationships between gene expression clusters.

Conclusions:

• The combination of icc-cluster and PCC-MCP offers a superior approach for gene expression data analysis.
• t-SNE-MCP-O maps effectively visualize complex gene expression patterns and cluster relationships.
• This method enhances biological knowledge discovery from transcriptome data.