Predicting the impact of sequence motifs on gene regulation using single-cell data

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Gene regulation relies on transcription factor binding at promoters and enhancers.
• Quantifying the impact of regulatory motifs on gene expression is a significant challenge in genomics.

Purpose of the Study:

• To develop a computational model to infer regulatory motifs and their cell type-specific importance from single-cell data.
• To apply this model to identify regulatory elements in complex biological systems.

Main Methods:

• A convolutional neural network (CNN) model named scover was trained on single-cell gene expression data.
• The model was validated on multiple mouse tissues and applied to human brain developmental enhancers identified via scATAC-seq.

Main Results:

• Scover explains 29% of the variance in gene expression across various mouse tissues.
• The model successfully identified cell type-specific motif activities within distal enhancers in the developing human brain.
• Scover provides interpretable parameters and outputs for motif identification and importance.

Conclusions:

• Scover is an effective tool for identifying regulatory motifs and their functional importance from single-cell genomics data.
• The model advances our understanding of gene regulation by pinpointing cell type-specific enhancer activities.
• This approach offers a scalable and interpretable method for dissecting gene regulatory mechanisms.