Cisformer: a scalable cross-modality generation framework for decoding transcriptional regulation at single-cell...

Area of Science:

• Genomics
• Computational Biology
• Systems Biology

Background:

• Single-cell multiomic technologies offer powerful insights into cellular heterogeneity but present significant experimental and computational challenges.
• Existing computational methods for cross-modality translation in single-cell data often lack biological interpretability, limiting their utility for discovery.
• Analyzing gene expression and chromatin accessibility jointly is crucial for understanding gene regulation.

Purpose of the Study:

• To develop a novel computational model, Cisformer, for accurate cross-modality generation between gene expression and chromatin accessibility at single-cell resolution.
• To enhance the biological interpretability of single-cell multiomic data analysis.
• To facilitate the identification of regulatory elements and transcription factors involved in biological processes such as tumorigenesis and aging.

Main Methods:

• Development of Cisformer, a cross-attention-based generative model specifically designed for single-cell cross-modality translation.
• Application of Cisformer to jointly analyze gene expression and chromatin accessibility data from single cells.
• Systematic benchmarking of Cisformer against existing computational methods using multiple datasets.

Main Results:

• Cisformer demonstrated superior accuracy and generalization capabilities compared to current state-of-the-art methods in cross-modality generation.
• The model's inherent interpretability allowed for precise linking of cis-regulatory elements to their target genes.
• Cisformer facilitated the identification of functional transcription factors implicated in tumorigenesis and aging.

Conclusions:

• Cisformer represents a significant advancement in single-cell multiomic data analysis, offering enhanced accuracy and biological interpretability.
• The model provides a powerful framework for dissecting gene regulatory mechanisms and identifying disease-associated factors.
• Cisformer is a valuable tool for researchers studying complex biological processes at single-cell resolution.