Related Experiment Video
Updated: Feb 26, 2026
05:12
Author Spotlight: AQRNA-seq Role in Mapping Small RNAs and Unraveling Protein Translation Mechanisms
Published on: February 2, 2024
1.4K
Independent Component Analysis (ICA) based-clustering of temporal RNA-seq data
Moysés Nascimento1, Fabyano Fonseca E Silva2, Thelma Sáfadi3
1Department of Statistics, Federal University of Viçosa, Viçosa, Minas Gerais, Brazil.
Plos One
|July 18, 2017
Summary
This study introduces ICAclust, a novel method for clustering gene expression time series data. ICAclust outperforms traditional K-means clustering, effectively grouping genes with similar temporal expression patterns.
Area of Science:
- Bioinformatics
- Computational Biology
- Genomics
Background:
- Gene Expression Time Series (GETS) analysis identifies genes with similar expression patterns over time.
- Traditional clustering methods struggle with GETS data due to non-normality and limited temporal observations.
- Independent Component Analysis (ICA) offers a way to handle distribution-free, small sample data while considering temporal dependencies.
Purpose of the Study:
- To present ICAclust, a two-step clustering methodology combining Independent Component Analysis (ICA) and hierarchical clustering for GETS analysis.
- To evaluate the performance of ICAclust against K-means clustering using simulated and real RNA-seq data.
- To demonstrate ICAclust's effectiveness in grouping genes with similar temporal expression profiles.
Main Methods:
- A novel two-step clustering approach, ICAclust, was developed, integrating ICA with hierarchical clustering.
- Simulated and real RNA-seq data from pig breeds at different gestational ages were used for analysis.
- Performance was compared against K-means clustering, a standard method for temporal gene expression data.
Main Results:
- ICAclust demonstrated superior performance compared to K-means, with an average absolute gain of 5.15% and up to 84.85% improvement in certain scenarios.
- The method successfully grouped genes into distinct clusters with highly differentiated temporal expression patterns in real RNA-seq data.
- ICAclust effectively identified sets of genes exhibiting similar longitudinal expression profiles.
Conclusions:
- The proposed ICAclust method is an effective two-step clustering approach for Gene Expression Time Series analysis.
- ICAclust offers significant advantages over traditional K-means clustering, particularly for data with non-normality and small sample sizes.
- This methodology enhances the ability to summarize biological processes and regulatory mechanisms by accurately clustering genes with similar temporal expression patterns.
