Multi-omic assessment using proteins and nucleic acids
Inventors
Ma, Philip • Wilcox, Bruce • Collin, Francois • BELTHANGADY, Chinmay • Yang, Mi • Khadka, Manoj • Liu, Manway • Blume, John • Langer, Jr., Robert S. • Khaledian, Ehdieh
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12334190-B2
Publication Date
2025-06-17
Expiration Date
Abstract
Described herein are methods such as multi-omic methods for assessing a disease such as cancer. The multi-omic methods may integrate proteomic, transcriptomic, genomic, lipidomic, or metabolomic data. The method screening diseases or disease states. Also described herein are methods for screening for diseases or disease states from biological samples. The methods may include assessing whether a nodule, mass, or cyst is cancerous.
Core Innovation
The invention provides a multi-omic method for determining presence or absence of cancer in a subject by analyzing a biofluid sample. Nucleic acids obtained from the biofluid sample are sequenced to obtain nucleic acid sequence information, and the nucleic acids are quantified to obtain nucleic acid quantification information. Proteins from the biofluid sample are processed to obtain proteomic data.
A combined dataset is generated that includes at least a portion of proteomic data, at least a portion of nucleic acid quantification information, and at least a portion of nucleic acid sequence information. The combined dataset is input into a trained machine learning classifier trained on cancer samples and non-cancer samples to determine cancer presence or absence. The classifier has a performance characteristic comprising an AUC of a ROC curve of at least 0.83.
In a further embodiment, the biofluid sample is combined with internal standard proteins, the internal standard proteins are processed, and endogenous proteins of the biofluid sample are processed based on the internal standard proteins to obtain proteomic data. The combined dataset is then formed for classifier input, and performance is measured with an average AUC of a ROC curve of at least 0.83.
Claims Coverage
The consolidated claim coverage includes two independent inventive features. Both focus on combining nucleic-acid sequence information, nucleic-acid quantification information, and proteomic data from a biofluid sample for cancer classification, with specified ROC AUC performance thresholds; one feature further requires internal standard proteins for proteomic processing.
Multi-omic dataset classification for cancer presence or absence with ROC AUC threshold
A multi-omic method that sequences nucleic acids from a biofluid sample to obtain nucleic acid sequence information, quantifies the nucleic acids to obtain nucleic acid quantification information, processes proteins from the biofluid sample to obtain proteomic data, generates a combined dataset comprising at least a portion of the proteomic data, at least a portion of the nucleic acid quantification information, and at least a portion of the nucleic acid sequence information, and determines a presence or an absence of cancer by inputting the combined dataset into a trained machine learning classifier trained on cancer samples and non-cancer samples, wherein the trained machine learning classifier has an AUC of a ROC curve of at least 0.83.
Internal standard proteins combined with multi-omic classifier to measure average ROC AUC threshold
A multi-omic method that sequences nucleic acids obtained from a biofluid sample to obtain nucleic acid sequence information, quantifies the nucleic acids to obtain nucleic acid quantification information, combines the biofluid sample with internal standard proteins, processes the internal standard proteins, processes endogenous proteins of the biofluid sample based on the internal standard proteins to obtain proteomic data, generates a combined dataset comprising at least a portion of the proteomic data, at least a portion of the nucleic acid quantification information, and at least a portion of the nucleic acid sequence information, determines a presence or an absence of cancer by inputting the combined dataset into a classifier, and measures a performance characteristic comprising an average AUC of a ROC curve of at least 0.83.
The claims require generation of a combined dataset from nucleic-acid sequence information, nucleic-acid quantification information, and proteomic data, followed by cancer determination using a trained classifier with ROC/AUC thresholds of at least 0.83. One claim additionally includes internal standard proteins used in processing endogenous proteins for proteomic data generation.
Stated Advantages
Provides a trained machine learning classifier with a performance characteristic comprising an ROC AUC of at least 0.83 for determining cancer presence or absence.
Provides an average ROC AUC of at least 0.83 when measuring a performance characteristic for determining cancer presence or absence.
Documented Applications
Assessing cancer using a non-invasive multi-omic method based on a biofluid sample that includes proteomic data and nucleic-acid sequencing/quantification information.
Determining cancer presence or absence for a cancer selected from lung cancer, pancreatic cancer, colon cancer, liver cancer, breast cancer, or ovarian cancer.
Generating an indicative biological state via a classifier and using that determination to administer chemotherapy, radiation, or surgical cancer treatment.
Interested in licensing this patent?