Federated retrieval augmentation for ai-based drug discovery and development

A method for performing an AI-based drug discovery and development task, the method comprising:

• a) receiving, by a coordinator communicatively coupled to a plurality of entities, from a respective one of the plurality of entities, a query, one or more of the plurality of entities each maintaining a respective local non-parametric data store comprising respective non-parametric data;
• b) causing, by the coordinator, a similarity determination to be performed between the query and the respective non-parametric data of each of the one or more other entities of the plurality of entities;
• c) transmitting, by the coordinator to the respective one of the plurality of entities, at least one retrieved non-parametric data object selected from the respective non-parametric data based on the respective similarity determinations;
• d) augmenting the query with at least a portion of the at least one retrieved non-parametric data object to form an augmented query;
• e) performing inference on a global neural network using the augmented query to produce an inference output;
• and
• f) using the inference output to support the AI-based drug discovery and development task:
• i) wherein the query is not directly transferred from the respective one of the plurality of entities submitting the query to any of the one or more other entities of the plurality of entities,
• ii) and wherein the respective non-parametric data of any of the one or more other entities of the plurality of entities is not directly transferred to the respective one of the plurality of entities submitting the query.

A method of manufacturing a ligand, the method comprising:

• a) receiving, by a coordinator communicatively coupled to a plurality of entities, from a respective one of the plurality of entities, a query;
• b) causing, by the coordinator, a similarity determination to be performed between the query and respective non-parametric data respectively held by each of one or more other entities of the plurality of entities;
• c) transmitting, by the coordinator to the respective one of the plurality of entities, at least one retrieved non-parametric data object selected from the respective non-parametric data based on the respective similarity determinations;
• d) augmenting the query with at least a portion of the at least one retrieved non-parametric data object to form an augmented query;
• e) generating, by a global neural network, a candidate ligand representation in response to the augmented query;
• and
• f) synthesizing a ligand corresponding to the candidate ligand representation, thereby producing a synthesized ligand:
• i) wherein the query is not directly transferred from the respective one of the plurality of entities submitting the query to any of the one or more other entities of the plurality of entities,
• ii) and wherein the respective non-parametric data of any of the one or more other entities of the plurality of entities is not directly transferred to the respective one of the plurality of entities submitting the query.