Methods and systems for designing phage cocktails

A computer-implemented method for designing phage therapy cocktails, method comprising:

• (a) predicting, with a candidate phage selector, a predicted lytic activity of each of a plurality of phages with a first trained machine learning algorithm, and selecting a plurality of candidate phages from the plurality of phages based at least in part on the predicted lytic activity;
• wherein (a) further comprises:
• i) receiving input data comprising phage data of the plurality of phages and host data of a set of hosts;
• ii) extracting attribute data from the input data;
• iii) processing the attribute data using the first trained machine learning algorithm to predict lytic susceptibility of individual hosts from set of hosts to individual phages from the plurality of phages, thereby generating phage-host lytic susceptibility data;
• and
• iv) providing a selection of the plurality of phages from the plurality of phages and a selection of a set of candidate hosts from the set of hosts, based at least in part on the phage-host lytic susceptibility data,
• (b) generating, with a phage cocktail generator based on a second machine learning algorithm, the set of candidate phage cocktails selected in (a) (iv), wherein each phage cocktail of the set of candidate phage cocktails comprises a combination of at least two phages from a plurality of candidate phages, and wherein each phage cocktail of the set of candidate phage cocktails is generated based at least in part of the predicted lytic activity of the at least two phages, wherein the second machine learning algorithm uses as constraint the genetic diversity between phages;
• and
• wherein (b) further comprises:
• a) receiving a phage-host lytic susceptibility matrix and phage data retrieved from phage databases;
• b) selecting a plurality of candidate phage cocktails based at least in part on the phage host lytic susceptibility matrix and the phage data;
• and
• c) predicting lytic susceptibility of hosts to individual candidate phage cocktails from the plurality of candidate phage cocktails;
• and
• (c) visually presenting, by a feedback panel display, the generated set of phage cocktails from the set of candidate phage cocktails of (b) to a user interface to a user for selecting phage cocktails of interest from the generated set of candidate phage cocktails to conduct in vitro formulation development and testing for developing a plurality of developed phage cocktail formulations, and wherein the user receives the results obtained from the in vitro testing of the selected cocktails and inputs the results, by a feedback panel display, into at least one of the first trained machine learning algorithm and one of the second trained machine learning algorithm, and
• (d) outputting, by the feedback panel display, a developed phage cocktail formulation for subsequently in vitro verification;
• (f) verifying the developed phage cocktail formulation by double layer agar;
• (e) selecting the developed phage cocktail formulation of (d) based on a predetermined in vitro effectiveness percentage threshold, wherein the predetermined in vitro effectiveness percentage threshold is greater than 60.00,
• (g) generating the developed phage cocktail by repeating steps (a)-(e) for making developed phage cocktail formulations;
• (h) making the developed phage cocktail formulations by a process that comprises double layer agar.