TimesVector: a vectorized clustering approach to the analysis of time series transcriptome data from multiple...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Time series gene expression data analysis is crucial for understanding biological mechanisms.
• Analyzing three-dimensional gene-time-condition data presents high computational complexity.
• Existing algorithms struggle with the complexity of multi-dimensional time series gene expression data.

Purpose of the Study:

• To develop a triclustering algorithm, TimesVector, for analyzing three-dimensional time series gene expression data.
• To identify gene sets with distinctively similar or different expression patterns across multiple sample conditions.
• To overcome the computational challenges of traditional time series clustering methods.

Main Methods:

• TimesVector employs a three-step approach: dimension reduction and clustering, post-processing for pattern detection, and gene rescue.
• The algorithm processes gene expression data across time and sample conditions.
• It utilizes vector-based clustering for identifying patterns in three-dimensional data.

Main Results:

• TimesVector successfully detected biologically meaningful clusters of high quality from diverse datasets.
• The algorithm demonstrated improved clustering quality compared to existing triclustering tools.
• TimesVector uniquely identified clusters exhibiting differential expression patterns across conditions.

Conclusions:

• TimesVector is an effective tool for analyzing complex, three-dimensional time series gene expression data.
• The algorithm provides a robust method for identifying biologically relevant gene expression patterns.
• TimesVector advances the field of time series transcriptomic data analysis.