SCIG: Machine learning uncovers cell identity genes in single cells by genetic sequence codes.

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Understanding cell identity is crucial for studying cell differentiation, development, and diseases.
• Cell identity genes (CIGs) are known to have unique epigenetic regulatory patterns.
• Recent findings suggest CIGs also possess distinct genetic sequence signatures.

Purpose of the Study:

• To introduce SCIG, a novel machine-learning method for identifying cell identity genes in single cells.
• To leverage genetic sequence signatures and gene expression data for cell identity gene discovery.
• To provide a tool that does not require comparison with other cells for gene identification.

Main Methods:

• Developed SCIG, a machine-learning approach utilizing genetic sequence signatures and single-cell RNA-seq data.
• Analyzed unique enrichment patterns of cis-regulatory elements as genetic sequence signatures for CIGs.
• Defined a Cell Identity Gene (CIG) score to assess gene identity, outperforming simple expression values in network analysis.

Main Results:

• SCIG effectively uncovers cell identity genes by analyzing genetic sequence signatures and gene expression.
• The SCIG score proved superior to expression values in network analysis for identifying master transcription factors (TFs).
• Application to the human endothelial cell atlas demonstrated the importance of the tissue microenvironment in refining cell identity, complementing master TFs.

Conclusions:

• SCIG is a powerful tool for identifying cell identity genes in single cells, independent of comparative analysis.
• The method highlights the significance of genetic sequence signatures in defining cell identity.
• SCIG advances research in cell differentiation, development, and regenerative medicine by providing a novel analytical approach.