Incomplete multi-view gene clustering with data regeneration using Shape Boltzmann Machine.

Area of Science:

• Genomics
• Computational Biology
• Machine Learning in Bioinformatics

Background:

• Understanding gene structure and function is crucial for genetic biology and genomics.
• Multi-omics data and machine learning offer powerful tools for biological insights.
• Gene clustering is a key computational technique for analyzing gene behavior.

Purpose of the Study:

• To address the challenge of incomplete data in multi-view clustering for gene analysis.
• To present a novel deep Boltzmann machine-based framework for incomplete multi-view gene clustering.
• To evaluate the framework's performance on NCBI datasets.

Main Methods:

• Utilized a deep Boltzmann machine framework for incomplete multi-view clustering.
• Employed Shape Boltzmann Machines to regenerate incomplete data modalities from three NCBI datasets.
• Assessed performance using Silhouette and Davies-Bouldin indices.

Main Results:

• The proposed framework demonstrated improved performance compared to state-of-the-art methods.
• Comparative analysis indicated enhanced clustering accuracy.
• Welch's t-test confirmed the statistical significance of the performance improvement.

Conclusions:

• The deep Boltzmann machine-based framework effectively handles incomplete multi-view data for gene clustering.
• This approach offers more comprehensive and reliable insights into gene behavior.
• The method shows significant potential for advancing genomic data analysis.