Characterizing protein interactions employing a genome-wide siRNA cellular phenotyping screen.

Area of Science:

• Cellular Biology
• Computational Biology
• Systems Biology

Background:

• Understanding protein-protein interactions (PPIs) is crucial for deciphering cellular signaling networks.
• Existing methods for inferring PPIs lack the ability to characterize their functional effects (activating or inhibiting).

Purpose of the Study:

• To develop and validate a novel computational method for identifying the activating and inhibiting effects of PPIs.
• To apply this method to a genome-wide RNAi knockdown screen in HeLa cells.
• To create a publicly accessible database of predicted PPI effects.

Main Methods:

• Utilized mitotic phenotypes from a genome-wide RNAi knockdown screen.
• Employed a machine learning approach for prediction.
• Validated predictions against a gold standard of 6,870 known activating and inhibiting PPIs.

Main Results:

• Achieved high prediction accuracy (82% AUC) using cross-validation.
• Predicted de novo activating and inhibiting effects for 1,954 PPIs in HeLa cells across ten major Kyoto Encyclopedia of Genes and Genomes signaling pathways.
• Demonstrated that predicted PPI effects can cluster genes with similar biological processes.

Conclusions:

• The novel computational method successfully characterizes activating and inhibiting PPI effects.
• The predictions enhance the interpretability of large-scale PPI datasets.
• This work provides a valuable resource for studying cellular signaling pathway functions.