QOT: Quantized Optimal Transport for sample-level distance matrix in single-cell omics.

Area of Science:

• Bioinformatics
• Computational Biology
• Data Science

Background:

• Single-cell technologies generate high-dimensional data, posing computational challenges for comparative analyses.
• Generating sample-level distance matrices is crucial for understanding cellular architecture across biological conditions.
• Optimal Transport (OT) is a geometric data analysis tool used in bioinformatics.

Purpose of the Study:

• To introduce Quantized Optimal Transport (QOT), a novel method for efficient sample-level distance matrix computation.
• To address the computational and analytical challenges of large-scale single-cell omics data.
• To improve accuracy and robustness in comparative single-cell analyses.

Main Methods:

• Developed QOT, incorporating a quantization step for efficient computation.
• Applied QOT to large-scale single-cell genomics and pathomics datasets.
• Compared QOT performance against existing Optimal Transport-based algorithms.

Main Results:

• QOT significantly outperforms existing OT-based algorithms in accuracy and robustness.
• The method efficiently computes sample-level distance matrices from high-throughput single-cell data.
• Demonstrated the utility of QOT-derived matrices in downstream analyses, such as disease progression.

Conclusions:

• QOT provides an efficient and accurate solution for analyzing large-scale single-cell omics data.
• The method facilitates the extrapolation of cell-level insights to sample-level categorizations.
• QOT has broad applications in biomedical informatics and data science for comparative studies.