scCompass: An Integrated Multi-Species scRNA-seq Database for AI-Ready

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell sequencing generates vast datasets for cellular dynamics and gene regulation analysis.
• Artificial intelligence (AI) advances life sciences but faces challenges with inconsistent data processing quality and standards.
• A unified, standardized resource is needed for large-scale single-cell data analysis and AI integration.

Purpose of the Study:

• To introduce scCompass, a comprehensive, scalable, and AI-friendly database for single-cell transcriptomic data.
• To standardize data pre-processing and integrate multi-species single-cell data for enhanced research accessibility.
• To facilitate the identification of key gene expression patterns and support AI model training.

Main Methods:

• Curated and integrated transcriptomic data from approximately 105 million single cells across 13 species.
• Applied standardized data pre-processing pipelines to ensure data consistency and quality.
• Developed scalable datasets and provided pre-trained checkpoints for AI model development.

Main Results:

• Successfully integrated and curated a large-scale, multi-species single-cell transcriptomic dataset.
• Identified stable expression genes (SEGs) and organ-specific expression genes (OSGs) in humans and mice.
• Provided AI-ready datasets and pre-trained models to support advanced single-cell analysis.

Conclusions:

• scCompass serves as an efficient and scalable database for AI-ready single-cell data.
• The platform simplifies data access, sharing, and analysis for single-cell biology researchers.
• Facilitates discovery of gene expression patterns and accelerates AI-driven insights in the field.