Inference of RhoGAP/GTPase regulation using single-cell morphological data from a combinatorial RNAi screen

Area of Science:

• Systems Biology
• Computational Biology
• Genomics

Background:

• Biological networks rely on enzymes for signaling, often studied via transcriptional or proteomic data.
• High-content morphological data from RNA interference (RNAi) screens offers a new data source for network inference.
• Existing computational methods are limited in utilizing morphological data for signaling network analysis.

Purpose of the Study:

• To develop a systematic computational framework for identifying genetic interactions using high-dimensional single-cell morphological data.
• To infer signaling relationships within biological networks, specifically Rho GTPase signaling in Drosophila.
• To evaluate the efficacy of the proposed framework in predicting genetic interactions and signaling pathways.

Main Methods:

• Developed a classification model-based computational framework for genetic interaction identification.
• Utilized high-dimensional single-cell morphological data derived from large-scale RNAi screens.
• Applied the framework to Drosophila RhoGAP/GTPase signaling, incorporating single- and double-knockdown data.

Main Results:

• The framework achieved mediocre predictions using only single-knockdown morphological data.
• Accuracy significantly improved when incorporating double-knockdown RhoGAP genetic screen data, highlighting network redundancy.
• The developed model outperformed alternative inference methods.

Conclusions:

• High-throughput morphological data can be systematically and successfully employed to identify genetic interactions.
• Integrating morphological data with basic knowledge of network structure enables effective inference of signaling relationships.
• This approach provides a powerful tool for dissecting complex biological signaling networks.