Information processing system, information processing method, program, and method for producing antigen-binding molecule or protein

Patent metadata

Field	Value
Publication number	US-2022253669-A1
Application number	US-202017616421-A
Country	US
Kind code	A1
Filing date	Jun 8, 2020
Priority date	Jun 7, 2019
Publication date	Aug 11, 2022
Grant date	—

Abstract

A sequence learner performs machine learning on the basis of sequence information representing sequences including some or all of the sequences of a plurality of antigen-binding molecules and proteins and thereby generates a trained model. A virtual sequence generator generates, on the basis of the trained model, virtual sequence information obtained by mutating at least one of constituent elements constituting a sequence represented by the sequence information inputted into the trained model.

First claim

1 . An information processing system comprising:
a sequence learner configured to perform machine learning on the basis of sequence information representing sequences including some or all of the sequences of a plurality of antigen-binding molecules and thereby generate a trained model that has learned a character of the sequences; and a sequence generator configured to generate virtual sequence information obtained by mutating at least one of constituent elements constituting a sequence represented by the sequence information on the basis of the trained model.

2 . An information processing system comprising:
a sequence learner configured to perform machine learning on the basis of sequence information representing sequences including some or all of the sequences of a plurality of proteins and thereby generate a trained model that has learned a character of the sequences; and a sequence generator configured to generate virtual sequence information obtained by mutating at least one of constituent elements constituting a sequence represented by the sequence information on the basis of the trained model.

3 . The information processing system according to claim 1 , wherein
the character of the sequences is a character including positions of the constituent elements in the sequences and anteroposterior relationship of the constituent elements.

4 . The information processing system according to claim 1 , wherein the virtual sequence information is generated by changing at least one of the constituent elements in preset sites including one or more of the constituent elements in a sequence.

5 . The information processing system according to claim 4 , wherein the plurality of sites is included in a sequence of a heavy chain variable region, a light chain variable region, or a constant region of an antibody.

6 . The information processing system according to claim 1 , wherein the sequence information is sequence information selected according to results of characterization of antigen-binding molecules or proteins with the sequences represented by the sequence information.

7 . The information processing system according to claim 6 , wherein the sequence learner performs the machine learning with use of a deep learning model or a probability model.

8 . The information processing system according to claim 7 , wherein the sequence learner performs the machine learning with use of a deep learning model, and
performs the machine learning with use of, as the deep learning model, a Long short-term memory (LSTM), a recursive neural network (RNN), a Gated Recurrent Unit (GRU), a Generative Adversarial Network (GAN), or a Variational Autoencoder (VAE), or a Flow deep generative model.

9 . An information processing system comprising:
a learner configured to perform machine learning on the basis of sequence information representing sequences including some or all of the sequences of a plurality of antigen-binding molecules or proteins and results of characterization of antigen-binding molecules or proteins represented by the sequences and thereby generate a second trained model; and an estimator configured to input virtual sequence information generated on the basis of a first trained model being the trained model according to claim 1 into the second trained model, execute arithmetic processing of the second trained model, and thereby estimate predicted values for characterization of antigen-binding molecules or proteins with sequences represented by the inputted virtual sequence information.

10 . The information processing system according to claim 9 , comprising:
an output configured to, according to the predicted values estimated by the estimator, output on the basis of virtual sequence information and the predicted values.

11 . The information processing system according to claim 10 , wherein
the sequence learner to generate the first trained model performs machine learning on the basis of the virtual sequence information and thereby generates a new version of the first trained model, and/or the learner to generate the second trained model performs machine learning on the basis of the virtual sequence information and results of characterization of antigen-binding molecules or proteins with sequences represented by the virtual sequence information and thereby generates a new version of the second trained model.

12 . The information processing system according to claim 1 , wherein the sequence learner performs the machine learning on the basis of the sequence information represented by character strings, numeric vectors, or physical property values of constituent elements constituting sequences.

13 . An information processing method in an information processing system, the method comprising:
performing machine learning on the basis of sequence information representing sequences including some or all of the sequences of a plurality of antigen-binding molecules or proteins and thereby generating a trained model that has learned a character of the sequence information; and generating virtual sequence information obtained by mutating at least one of constituent elements constituting a sequence represented by the sequence information on the basis of the trained model.

14 . A non-transitory computer readable recording medium storing a program configured to allow a computer in an information processing system to execute:
performing machine learning on the basis of sequence information representing sequences including some or all of the sequences of a plurality of antigen-binding molecules or proteins and thereby generating a trained model; and generating virtual sequence information obtained by mutating at least one of constituent elements constituting a sequence represented by the sequence information on the basis of the trained model.

15 . A method for producing an antigen-binding molecule or protein with use of the information processing system according to claim 9 , wherein the antigen-binding molecule or protein is represented by a virtual sequence, and a predicted value for characterization has been estimated for the virtual sequence.

Assignees

• Chugai Pharmaceutical Co Ltd

Inventors

• Teramoto Reiji
• METSUGI Shouichi
• KAKUZAKI Taro
• Saka Koichiro
• Koga Hikaru
• Sampei Zenjiro

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G06N3/0455

  Auto-encoder networks; Encoder-decoder networks · CPC title

• G06N3/0475

  Generative networks · CPC title

• G06N3/09

  Supervised learning · CPC title