Meta-analysis of cell- specific transcriptomic data using fuzzy c-means clustering discovers versatile viral...

Area of Science:

• Bioinformatics
• Systems Biology
• Computational Biology

Background:

• Gene-set enrichment analysis (GSEA) is limited by generalized gene-set definitions from pathway databases.
• Data-driven gene-set identification offers context-specific insights but existing methods use hard clustering, failing to capture pathway overlaps.
• Biological pathways often exhibit overlapping characteristics, necessitating methods that accommodate this complexity.

Purpose of the Study:

• To develop a novel pipeline for identifying context-specific gene sets from transcriptomic data using a soft clustering approach.
• To overcome limitations of hard clustering in gene-set discovery by allowing overlap between identified gene sets.
• To improve the interpretation of transcriptomic data for discovering novel biological processes.

Main Methods:

• Utilized a fuzzy C-means (FCM) soft clustering algorithm to define gene sets, mimicking biological pathway topology.
• Applied Ward's method for optimizing initial conditions and tuned FCM parameters for human cell-specific transcriptomic data.
• Integrated transcriptomic data from monocyte-derived dendritic cells and A549 epithelial cells in response to influenza infection.

Main Results:

• Successfully derived robust gene sets that recapitulate topological features of biological pathways.
• Identified versatile viral-responsive genes by leveraging the soft clustering approach.
• Demonstrated that genes associated with multiple gene sets enhance transcriptomic data interpretation.

Conclusions:

• The FCM clustering algorithm significantly improves the interpretation of transcriptomic data by identifying overlapping gene sets.
• Facilitates the investigation of novel biological processes using publicly available transcriptomic datasets.
• Introduced an interactive package, Fuzzy Inference of Gene-sets (FIGS), to enable pipeline utilization and future extensions for diverse cell types.