Using single-cell perturbation screens to decode the regulatory architecture of splicing factor programs

Area of Science:

• Molecular Biology
• Genomics
• Computational Biology

Background:

• Splicing factors regulate gene isoforms, crucial for cell function.
• Dysregulated splicing is linked to diseases like cancer.
• Current methods for mapping splicing factor interactions are limited.

Purpose of the Study:

• To develop a novel computational approach for assessing splicing factor activity.
• To map the regulatory network of splicing factors in cancer.
• To link cancer driver mutations to splicing dysregulation.

Main Methods:

• Utilized shallow artificial neural networks to estimate splicing factor activity from gene expression data.
• Applied Perturb-seq (CRISPR perturbations + single-cell RNA sequencing) for high-throughput screening.
• Analyzed splicing program activity shifts as a reporter for oncogenic splicing factor activity.

Main Results:

• Developed an AI model that bypasses the need for exon-level data to measure splicing factor activity.
• Mapped a cross-regulatory network of splicing factors in carcinogenesis.
• Identified MYC and other pathways connecting cancer mutations to splicing regulation.

Conclusions:

• Established a versatile computational framework for studying splicing factor regulation.
• Demonstrated the utility of the framework in uncovering cancer-related disease mechanisms.
• Highlighted the recapitulation of developmental splicing dynamics in cancer regulation.