Sensitive and specific determination of dna methylation profiles

Patent metadata

Field	Value
Publication number	US-2026088128-A1
Application number	US-202319107390-A
Country	US
Kind code	A1
Filing date	Sep 1, 2023
Priority date	Sep 2, 2022
Publication date	Mar 26, 2026
Grant date	—

Abstract

The invention relates to methods for determining the methylation profile of DNA sequences of interest and methods for accurately distinguishing between a healthy methylation profile and a cancerous methylation profile, as well as to kits to implement them.

First claim

1 . A method for determining a CpG methylation profile of at least one DNA sequence of interest or any fragment thereof, wherein the method comprises:
a) clustering a set of sub-sequences obtained from a DNA sequence of interest into clusters of sub-sequences; b) selecting, for and from each cluster, one sub-sequence as a reference sequence among the sub-sequences of the cluster, c) aligning the reference sequences of said clusters by allowing the alignment on positions of CpG dinucleotides, d) aligning the remaining sub-sequences on selected reference sequences; and e) determining the CpG methylation status of each sub-sequence by determining at each CpG site of the sub-sequence if the CpG dinucleotide is methylated or not, thereby determining a CpG methylation profile comprising a CpG methylation level and/or a proportion of CpG methylation haplotype of the sub-sequences,
wherein the DNA sequence of interest is or comprises a repeated sequence, said repeated sequence being distributed throughout the subject's genome, and preferably comprising high density of CpG dinucleotides;
wherein, the method optionally comprises a first step of obtaining or providing a set of sub-sequences of said DNA sequence of interest, and
wherein the method optionally comprises repeating some, or each, of steps a) through e) with other sets of sub-sequences from the DNA sequence of interest.

2 . The method of claim 1 , wherein the repeated sequence is a retrotransposon such as LINE, HERV, SINE, SVA, or a subfamily thereof such as in particular LINE-1, L1PA, HERV-K and Alu, or a satellite repeat such as Sat2 or Sat3 element, preferably a LINE-1 retrotransposon or any fragment or variant thereof, even more preferably a LINE-1 retrotransposon such as described under SEQ ID NO: 2 or 29 or any fragment or variant thereof.

3 . A computer-implemented method of training a classifier for accurately distinguishing between a healthy CpG methylation profile and a cancerous CpG methylation profile, said method comprising:
a) providing a training set of CpG methylation profiles of DNA sequences of interest or sub-sequences thereof, said DNA sequences of interest being repeated and distributed throughout a genome and comprising high density of CpG dinucleotides, or preprocessed information obtained from said training set of CpG methylation profiles of DNA sequences of interest or sub-sequences thereof, as an input to a classifier, said training set of CpG methylation profiles comprising CpG methylation profiles of DNA sequences, or of sub-sequences thereof, from subjects identified as healthy subjects and from subjects identified as cancerous subjects; and, b) generating an output of the classifier for each CpG methylation profile input of DNA sequence of interest or sub-sequences thereof, said output classifying the CpG methylation profile input of DNA sequence of interest or sub-sequences thereof as a healthy CpG methylation profile or as a cancerous CpG methylation profile;
wherein the CpG methylation profile comprises a CpG methylation level and/or proportion of CpG methylation haplotypes of the DNA sequence or sub-sequences thereof.

4 . The method of claim 3 , wherein the CpG methylation profiles of the DNA sequences of interest or sub-sequences thereof are determined by the method of claim 1 .

5 . An in vitro or in silico method of determining the health status of a subject, in particular of determining if the subject is a healthy subject or a subject suffering from cancer or cancer relapse, wherein the method comprises:
a) providing a DNA sequence of interest or sub-sequences thereof from the subject, or preprocessed information obtained from said DNA sequence or sub-sequences, said DNA sequence of interest being a DNA sequence encoding a repeated sequence distributed throughout the subject's genome and comprising high density of CpG dinucleotides as an input to a classifier trained to distinguish between a healthy CpG methylation profile and a cancerous CpG methylation profile, and b) using the classifier to identify the CpG methylation profile of the DNA sequence of interest or sub-sequences thereof of said subject as a healthy CpG methylation profile or as a cancerous CpG methylation profile as an output of the classifier.

6 . An in vitro or in silico method of determining the origin of a tumor from a subject, wherein the method comprises:
a) providing a DNA sequence of interest or sub-sequences thereof from the subject, or preprocessed information obtained from said DNA sequence or sub-sequences, said DNA sequence of interest being a DNA sequence encoding a repeated sequence distributed throughout the subject's genome and comprising high density of CpG dinucleotides as an input to a classifier trained to distinguish between a healthy CpG methylation profile and cancerous CpG methylation profile from different tumors origins, and b) using the classifier to identify the CpG methylation profile of the DNA sequence of interest or sub-sequences thereof of said subject as a healthy CpG methylation profile or as a cancerous CpG methylation profile from a particular tumor origin to determine the origin of the tumor from the subject as an output of the classifier.

7 . An in vitro or in silico method of determining the stage of a tumor from a subject, wherein the method comprises:
a) providing a DNA sequence of interest or sub-sequences thereof from the subject, or preprocessed information obtained from said DNA sequence or sub-sequences, said DNA sequence of interest being a DNA sequence encoding a repeated sequence distributed throughout the subject's genome and comprising high density of CpG dinucleotides as an input to a classifier trained to distinguish between a healthy CpG methylation profile and cancerous CpG methylation profile of different stages, and b) using the classifier to identify the CpG methylation profile of the DNA sequence of interest or sub-sequences thereof of said subject as a healthy CpG methylation profile or as a cancerous CpG methylation profile of a particular stage to determine the stage of the tumor from the subject as an output of the classifier.

8 . An in vitro or in silico method of monitoring the response to an anti-cancer treatment of a subject suffering from cancer, wherein the method comprises:
a) providing at least one DNA sequence of interest or sub-sequences thereof from a first liquid biopsy from a subject suffering from cancer before the administration of the anti-cancer treatment to the subject as a first input, said DNA sequence of interest being repeated through the subject genome and comprising high density of CpG sites or a fragment thereof, or preprocessed information obtained from said first liquid biopsy, and a second liquid biopsy comprising at least one DNA sequence of interest or sub-sequences thereof from said subject after the administration of an anti-cancer treatment as a second input, or preprocessed information obtained from said second liquid biopsy, to a classifier trained to distinguish between DNA sequence having a healthy CpG methylation profile and DNA sequence having a cancerous CpG methylation profile; and b) using the classifier to identify each CpG methylation profile of each DNA sequence of the first liquid biopsy as having a healthy CpG methylation profile or a cancerous CpG methylation profile as a first output of the classifier, and to identify each CpG methylation profile of each DNA sequence of the second liquid biopsy as having a healthy CpG methylation profile or a cancerous CpG methylation profile as a second output of the classifier, and wherein a number of DNA sequence of interest classified as having a healthy CpG methylation profile in the second output of the classifier which is above a number of

Assignees

• Inst Curie
• Inst Nat Sante Rech Med
• Centre Nat Rech Scient
• Ecole Nat Superieure Des Mines De Paris
• Paris Sciences Et Lettres
• Univ Sorbonne

Inventors

• Proudhon Charlotte
• Azencott Chloé-Agathe
• Michel Marc
• Heidary Maryam

Classifications

• C12Q2600/154

  Methylation markers · CPC title

• C12Q1/6886

  for cancer (immunoassay for cancer G01N33/575) · CPC title

• G16B40/20

  Supervised data analysis · CPC title

• G16B30/10 Primary

  Sequence alignment; Homology search · CPC title

• G16B20/00 Primary

  ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations · CPC title