HDSCC: A robust clustering approach for Single Cell RNA-seq data using Hyperdimensional Encoding.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables high-throughput gene expression analysis, advancing cellular structure understanding.
• Microfluidics and unique molecular identifiers (UMIs) enhance scRNA-seq data quality and scale.
• Analyzing large scRNA-seq datasets presents computational challenges due to processing time, resource demands, and data sparsity/noise.

Purpose of the Study:

• To introduce a novel computational method for effective clustering of large scRNA-seq datasets.
• To address the challenges of processing time, computational resources, and noisy/sparse data in scRNA-seq analysis.
• To apply Hyper Dimensional Computing (HDC) to scRNA-seq data analysis for noise-robust clustering.

Main Methods:

• Implementation of a Hyper Dimensional Computing (HDC) approach for scRNA-seq data analysis.
• Development of a new clustering method designed for large-scale RNA sequencing datasets.
• Comparative analysis against state-of-the-art single-cell clustering techniques.

Main Results:

• The proposed HDC method demonstrates promising performance and robustness in clustering scRNA-seq data.
• Experimental validation was conducted on a standard laptop (3.2-GHz CPU, 32 GB RAM), indicating efficiency.
• The method effectively handles noisy and highly sparse data inherent in single-cell studies.

Conclusions:

• Hyper Dimensional Computing (HDC) offers a viable and effective solution for clustering large scRNA-seq datasets.
• The novel HDC-based clustering method provides a robust alternative to existing approaches for single-cell data analysis.
• This work represents a significant step in leveraging advanced computational techniques for single-cell genomics research.