Pooled screening for synergistic interactions subject to blocking and noise

Area of Science:

• Systems biology
• Computational biology
• Genomics

Background:

• Cellular molecular networks exhibit complex interactions, including synthetic lethality and synergistic drug effects.
• Discovering these interactions is hindered by the vast number of potential combinations.
• Pooled screening offers efficiency but risks masking genuine interactions due to antagonistic factors.

Purpose of the Study:

• To develop and analyze theoretical models for pooled screening that account for synergy, antagonism, and noise.
• To investigate randomized sequential designs for efficient discovery of synergistic interactions.
• To determine optimal pool sizes and strategies for mitigating testing noise.

Main Methods:

• Exploration of theoretical models for pooled screening with synergy, antagonism, and noise.
• Derivation of formulae for expected tests to discover synergistic interactions.
• Analysis of randomized sequential designs and Bayesian approaches for adaptive pool sizing.

Main Results:

• Randomized pooled designs significantly outperform exhaustive testing, even with antagonism and noise.
• Testing noise does not impact the optimal pool size.
• Selective retesting is more effective than naive replication for noise mitigation.
• A Bayesian approach enables adaptive pool sizing based on parameter uncertainty.

Conclusions:

• Randomized pooled screening is a robust and efficient strategy for discovering cellular interactions.
• The methodology effectively manages challenges like antagonistic interactions and measurement noise.
• Adaptive pool sizing using Bayesian methods optimizes the screening process.