MCPD

MCPD decodes pooled library screening data to infer positive clones using Bayesian inference implemented via Markov chain Monte Carlo (MCMC), modeling priors for positives and errors to produce posterior probabilities for candidate ranking.

Key Features:

• Markov chain Monte Carlo (MCMC): Uses an MCMC algorithm to manage the combinatorial complexity of decoding pooled screening results and to sample posterior distributions.
• Bayesian inference with priors: Employs Bayesian models with explicit prior distributions for positives and error processes to inform inference.
• Posterior probability ranking: Computes posterior probabilities for candidate positives to enable ranked identification of likely positives.
• Error modeling and ambiguity handling: Incorporates error models to address ambiguities introduced by screening errors.
• Scalability demonstrated: Applied to a 1,298-clone library screened with 47 pools and simulated for a 10-fold coverage library of 33,000 clones screened with 253 pools.
• Reduced pooling requirements: Enables robust decoding under conditions with fewer pools while maintaining inference quality.
• Simulation validation: Supports simulation-based evaluation of performance on larger libraries and pooling schemes.

Scientific Applications:

• Decoding pooling experiments: Infers which individual clones are positive from pooled probe screening outcomes.
• Prioritizing confirmatory screens: Ranks candidate positives by posterior probability to guide confirmatory screening efforts.
• High-throughput library screening: Facilitates screening strategies for large clone libraries where combinatorial decoding is impractical.
• Simulation of large-scale screens: Evaluates pooling designs and expected performance for libraries such as a 10-fold coverage 33,000-clone set with 253 pools.

Methodology:

Implements Bayesian inference via Markov chain Monte Carlo (MCMC) with explicit priors for positives and error models to compute posterior probabilities and rank candidate positive clones; includes simulation-based evaluations.