US 11,742,057 B2
Systems and methods for artificial intelligence-based prediction of amino acid sequences at a binding interface
Joshua Laniado, Los Angeles, CA (US); Julien Jorda, Los Angeles, CA (US); Matthias Maria Alessandro Malago, Santa Monica, CA (US); Thibault Marie Duplay, Los Angeles, CA (US); Mohamed El Hibouri, Los Angeles, CA (US); Lisa Juliette Madeleine Barel, Los Angeles, CA (US); and Ramin Ansari, Los Angeles, CA (US)
Assigned to Pythia Labs, Inc., Los Angeles, CA (US)
Filed by Pythia Labs, Inc., Los Angeles, CA (US)
Filed on Jul. 22, 2022, as Appl. No. 17/871,425.
Application 17/871,425 is a continuation in part of application No. 17/384,104, filed on Jul. 23, 2021, granted, now 11,450,407.
Claims priority of provisional application 63/353,481, filed on Jun. 17, 2022.
Claims priority of provisional application 63/224,801, filed on Jul. 22, 2021.
Claims priority of provisional application 63/224,801, filed on Jul. 22, 2021.
Prior Publication US 2023/0040576 A1, Feb. 9, 2023
Int. Cl. G16B 15/30 (2019.01); G16B 45/00 (2019.01); G16B 40/00 (2019.01)
CPC G16B 15/30 (2019.02) [G16B 40/00 (2019.02); G16B 45/00 (2019.02)] 30 Claims
OG exemplary drawing
 
1. A method for the in-silico design of an amino acid interface of a biologic for binding to a target, the method comprising:
(a) receiving, by a processor of a computing device, an initial scaffold-target complex graph comprising a graph representation of at least a portion of a biologic complex comprising the target and a peptide backbone of the in-progress custom biologic, the initial scaffold-target complex graph comprising:
a target graph representing at least a portion of the target; and
a scaffold graph representing at least a portion of the peptide backbone of the in-progress custom biologic, the scaffold graph comprising a plurality of scaffold nodes, a subset of which are unknown interface nodes, wherein each of said unknown interface nodes:
(i) represents a particular amino acid interface site, along the peptide backbone of the in-progress custom biologic, that is located in proximity to one or more amino acids of the target, and
(ii) has a corresponding node feature vector comprising a side chain type component vector populated with one or more masking values, thereby representing an unknown, to-be determined, amino acid side chain;
(b) generating, by the processor, using a machine learning model, one or more likelihood graphs based on the initial scaffold-target complex graph, each of the one or more likelihood graphs comprising a plurality of nodes, a subset of which are classified interface nodes, each of which:
(i) corresponds to a particular unknown interface node of the scaffold graph and represents a same particular interface site along the peptide backbone of the in-progress custom biologic as the corresponding particular interface node, and
(ii) has a corresponding node feature vector comprising a side chain component vector populated with one or more likelihood values;
(c) using, by the processor, the one or more likelihood graphs to determine a predicted interface comprising, for each interface site, an identification of a particular amino acid side chain type; and,
(d) providing the predicted interface for use in designing the amino acid interface of the in-progress custom biologic and/or using the predicted interface to design the amino acid interface of the in-progress custom biologic.