Abstract: A system and method for performing an AI-based drug discovery and development task across a plurality of entities in a federation. One or more of the plurality of entities each maintains a respective local non-parametric data store. A coordinator receives a query from a respective one of the plurality of entities, causes a similarity determination between the query and respective non-parametric data of one or more other entities, and transmits at least one retrieved non-parametric data object to the querying entity. The query is augmented with at least a portion of the retrieved non-parametric data and used to perform inference on a global neural network, producing an inference output supporting the task. The query is not directly transferred to any of the other entities, and the respective non-parametric data of any of the other entities is not directly transferred to the querying entity.

Abstract: Methods for small molecule drug design using neural networks with two or more output heads, wherein one head, a sequence head, is trained to generate the sequence of a small molecule drug, and another head, a structure head, is trained to generate the structure of the small molecule drug; and wherein the neural network is configured to accept a representation of a specified condition as input, and output a representation of a small molecule drug's sequence and structure. The structure head and sequence head each have their own loss functions, and the weights of the neural network body are shared, and jointly updated during training. Non-limiting examples of specified input conditions include representations of associated proteins and/or sets of properties of the desired output small molecule drug. Some embodiments of the invention include for the design and synthesis of effective small molecule drug ligands.

Abstract: Methods and apparatus for protein and drug design using neural networks with two or more output heads, wherein one head, a sequence head, is trained to generate the sequence of a protein, and another head, a structure head, is trained to generate the structure of the protein; and wherein the neural network is configured to accept a representation of a specified condition as input, and output a representation of a protein's sequence and structure. The structure head and sequence head each have their own loss functions, and the weights of the neural network body are shared, and jointly updated during training. Non-limiting examples of specified input conditions include representations of associated proteins and/or sets of properties of the desired output protein. Some embodiments of the invention include for the design and synthesis of effective peptide drug ligands, synthetic biologic antibody drugs, antibody drug conjugates, and monoclonal antibody (mAb) drugs.

Abstract: Methods and apparatus for obtaining representations of proteins and peptide drugs for synthesis; wherein input queries into trained mixed modality protein and natural language models are augmented with relevant query-related documents. In one embodiment, the relevant query-related documents are obtained by maximum inner product search of an embedding latent vector space into which the query and the documents are projected. The top-k most relevant documents to the query are then combined with the query as input into the trained mixed modality language model. In one embodiment, the mixed modality model is an autoregressive multicapitate transformer whose decoder output heads correspond to the represented modalities. The method returns mixed modality output representations of proteins or peptide drugs which are then synthesized.

Abstract: Methods and apparatus using a mixture of representation modalities including natural language, protein sequence, protein structure, property-vector, and small molecule drug representations to jointly train a neural network which accepts mixed modality queries as input and produces mixed modality output responses including representations of proteins for synthesis and of small molecule drugs for manufacture. In one embodiment of the invention, multicapitate transformers wherein each decoder head has a distinct loss function and represents a distinct modality, are used. Modality-specific embeddings are implemented for the mixed modality input query, and an autoregressive process yields the output protein for synthesis or small molecule drug for manufacture.

Abstract: Methods and apparatus for obtaining representations of proteins and small molecule drugs for synthesis; wherein input queries into trained mixed modality protein and natural language models are augmented with relevant query-related documents. In one embodiment, the relevant query-related documents are obtained by maximum inner product search of an embedding latent vector space into which the query and the documents are projected. The top-k most relevant documents to the query are then combined with the query as input into the trained mixed modality language model. In one embodiment, the mixed modality model is an autoregressive multicapitate transformer whose decoder output heads correspond to the represented modalities. The method returns mixed modality output representations of proteins or small molecule drugs for synthesis or manufacture.