PhageCocktail

PhageCocktail provides an R package that designs optimized phage cocktails from phage-bacteria infection networks to maximize bacterial lysis for phage therapy applications.

Key Features:

• ExhaustiveSearch: Guarantees identification of the optimal phage cocktail by exhaustively evaluating all possible combinations within a given network, with longer computational runtimes.
• ExhaustivePhi: Focuses evaluation on a single cocktail size deemed most effective to reduce runtime relative to ExhaustiveSearch, potentially yielding cocktails with more phages than strictly necessary.
• ClusteringSearch: Uses clustering techniques to accelerate cocktail design, often completing within milliseconds while prioritizing speed over precision.
• ClusteringPhi: Leverages clustering for rapid design targeted to a specific estimated optimal cocktail size, providing quick outcomes with potentially lower accuracy than exhaustive methods.
• Evaluation dataset: Methods were evaluated on 13 empirical phage-bacteria infection networks.
• Performance metrics: Evaluation focused on runtime and expected success rate, where expected success rate is defined as the fraction of bacteria lysed by the designed cocktails.

Scientific Applications:

• Phage therapy (medical): Supports design of phage cocktails for therapeutic applications to lyse pathogenic bacteria.
• Food industry interventions: Facilitates formulation of phage cocktails for bacterial control in food safety contexts.
• Network-scale analysis: Enables exploration and optimization of large phage-bacteria infection networks and selection between speed- or accuracy-prioritized strategies in research and practical applications.

Methodology:

Implements four computational methods (ExhaustiveSearch, ExhaustivePhi, ClusteringSearch, ClusteringPhi); ExhaustiveSearch evaluates all combinations, ExhaustivePhi and ClusteringPhi target a single estimated optimal cocktail size, ClusteringSearch and ClusteringPhi use clustering techniques, and methods were assessed on 13 empirical phage-bacteria infection networks using runtime and expected success rate (fraction of bacteria lysed).