US 11,514,289 B1Granted

Generating machine learning models using genetic data

AssigneeFreenome Holdings, Inc.

InventorsGabriel Otte, Charles Roberts, Adam Drake, Riley Charles Ennis

Life of this patent

Estimating the term…

Loading patent…

Abstract

Systems, methods, and apparatuses for generating and using machine learning models using genetic data. A set of input features for training the machine learning model can be identified and used to train the model based on training samples, e.g., for which one or more labels are known. As examples, the input features can include aligned variables (e.g., derived from sequences aligned to a population level or individual references) and/or non-aligned variables (e.g., sequence content). The features can be classified into different groups based on the underlying genetic data or intermediate values resulting from a processing of the underlying genetic data. Features can be selected from a feature space for creating a feature vector for training a model. The selection and creation of feature vectors can be performed iteratively to train many models as part of a search for optimal features and an optimal model.

Claims23

Dependent claims

A method of creating a machine learning model for performing cancer classifications of biological samples, the method comprising:

receiving a plurality of training samples, wherein each of the plurality of training samples is a biological sample obtained from a corresponding human subject and includes nucleic acids of the corresponding human subject;
for each of the plurality of training samples:
sequencing nucleic acids of the corresponding human subject to obtain a set of sequence reads, thereby obtaining a plurality of sets of sequence reads corresponding to a plurality of chromosomes;
and
obtaining a known label for a classification of a cancer for the corresponding human subject;
receiving, at a computer system, the plurality of sets of sequence reads of nucleic acids of the corresponding human subjects and the known labels for the classification of the cancer;
identifying a set of features to be input to the machine learning model for each of the plurality of training samples, the set of features including non-aligned variables, the non-aligned variables including statistical measures of occurrence of Kmers of a Kmer database in the set of sequence reads;
for each set of the sets of sequence reads:
analyzing the set of sequence reads to obtain a training vector of a corresponding training sample, wherein the training vector comprises values of the set of features for the set of sequence reads, each element of the training vector corresponding to a feature that includes one or more variables;
operating on the training vectors using parameters of the machine learning model to obtain output labels for the classification of the cancer for the plurality of training samples;
comparing the output labels to the known labels of the training samples;
iteratively searching for optimal values of the parameters as part of training the machine learning model based on the comparing the output labels to the known labels of the training samples;
and
providing the parameters of the machine learning model and the set of features for the machine learning model.

11 more elements

A method of implementing a machine learning model for performing cancer classifications of biologic al samples, the method comprising:

sequencing nucleic acids of a human subject to obtain a set of sequence reads generated from a sample from the human subject, the set of sequence reads corresponding to a plurality of chromosomes;
storing definitions of a set of features to be input to the machine learning model, the set of features including non-aligned variables, the non-aligned variables including statistical measures of occurrence of Kmers of a Kmer database in the set of sequence reads;
analyzing the set of sequence reads to obtain a feature vector, wherein the feature vector comprises values of the set of features for the set of sequence reads, each element of the feature vector corresponding to a feature that includes one or more variables;
operating on the feature vector using parameters of the machine learning model to obtain an output label of the sample, wherein the output label comprises a classification of a cancer for the sample;
and
providing the output label.

3 more elements

Dependencies are inferred from the claim text.