Deep learning-based clustering approaches for bioinformatics.

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning

Background:

• Clustering is crucial for analyzing high-dimensional biological data like gene expressions, sequences, and images.
• Traditional clustering methods exist, but deep learning (DL) for representation learning in clustering is underutilized.
• Effective clustering relies on both data distribution and learned representations.

Purpose of the Study:

• To review state-of-the-art deep learning-based clustering approaches for bioinformatics.
• To explore training procedures, quality metrics, and applications of DL clustering.
• To provide insights for researchers applying DL unsupervised methods in bioinformatics.

Main Methods:

• Review of deep learning-based representation learning for cluster analysis.
• Detailed exploration of training procedures for DL clustering algorithms.
• Evaluation of DL clustering on bioimaging, cancer genomics, and biomedical text mining datasets.

Main Results:

• Deep neural networks effectively transform high-dimensional data into lower-dimensional spaces for improved clustering.
• The review covers various DL-based clustering techniques and their suitability for bioinformatics.
• Evaluations demonstrate the utility of DL clustering in specific bioinformatics use cases.

Conclusions:

• Deep learning offers powerful new avenues for cluster analysis in bioinformatics.
• This review provides a comprehensive overview and practical insights for applying DL clustering.
• The findings serve as a foundation for future research in DL-based unsupervised bioinformatics methods.