Robust reconstruction of single-cell RNA-seq data with iterative gene weight updates

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) reveals cellular heterogeneity but loses spatial and temporal information during dissociation.
• Reconstructing tissue structure from scRNA-seq data is crucial for understanding biological processes but is computationally challenging.
• Existing methods often require prior knowledge of informative genes, limiting their applicability.

Purpose of the Study:

• To develop a novel algorithm for improved tissue reconstruction from scRNA-seq data.
• To address the challenge of identifying informative genes without prior biological knowledge.
• To enhance the accuracy of reconstructing cellular relationships and biological processes.

Main Methods:

• An iterative algorithm was developed to identify manifold-informative genes.
• Existing scRNA-seq reconstruction algorithms were used as subroutines within the iterative process.
• The algorithm was benchmarked on diverse synthetic and real scRNA-seq datasets.

Main Results:

• The proposed algorithm significantly improved the quality of tissue reconstruction.
• Demonstrated enhanced performance on both synthetic and real-world scRNA-seq data.
• Successfully reconstructed tissue structures from mammalian intestinal epithelium and liver lobule datasets.

Conclusions:

• The iterative gene identification approach enhances the accuracy of scRNA-seq based tissue reconstruction.
• This method offers a robust solution for inferring cellular spatial and temporal relationships.
• The developed algorithm provides valuable tools for computational biology and genomics research.