Methods for generating a circadian clock comprising a dna methylation profile

Abstract

The present invention provides a method for generating a biological clock containing a DNA methylation map applicable to at least two different sample types, the method comprising: (i) providing a first set of DNA methylation maps generated from at least two different sample types from multiple subjects; (ii) generating a composite DNA methylation map from the first set of DNA methylation maps, wherein the composite DNA methylation map contains methylation sites having a matching state in the different sample types; and (iii) generating a biological clock using the composite DNA methylation map with a reference DNA methylation map from one of the at least two sample types.

Description

Technical Field

[0001] This invention relates to a method for determining the biological age and / or health status of a subject using DNA methylation mapping. Specifically, the invention provides a method for generating a biological clock based on DNA methylation mapping, which can be used to determine the biological age and / or health status of a subject from a variety of different sample types. Furthermore, the determined biological age and / or health status can be used to select lifestyle programs, dietary programs, or therapeutic interventions for the subject based on the health status determined from the DNA methylation mapping, or to determine the efficacy of lifestyle programs, dietary programs, or therapeutic interventions. Background Technology

[0002] The ability to determine information about the health status of participants is of great importance for understanding their overall health and well-being.

[0003] It is well known that chronological age is a primary indicator of overall health status, with increasing chronological age associated with declining health. However, depending on genetics, nutrition, and lifestyle, an individual may age at a rate that is slower or faster than their chronological age. Therefore, chronological age may not always reflect an individual's rate of aging or risk of declining health. On the other hand, an individual's biological age (based on, for example, clinical biochemistry and cell biology measures) may differ from that of other individuals of the same chronological age. Methods used to determine biological age may help identify individuals at risk of age-related conditions earlier than expected based on their chronological age (see, for example, WO2019 / 046725).

[0004] Epigenetic clocks used to predict chronological age and infer health status as an indicator of biological age are described in WO2022 / 272120. These epigenetic clocks are primarily based on chronological age as a training parameter.

[0005] Furthermore, existing solutions for predicting a subject's biological age are typically based on the correlation between DNA methylation patterns and age in a single sample type or combination of sample types. However, these methods are not optimal for determining biological age using DNA methylation profiles generated from different sample types. For example, existing methods involve training a biological clock on a first sample type and then transposing the DNA methylation profile to a second sample type by adding an offset or performing a linear transformation. The drawback of this approach is that it is often unreliable and / or inaccurate. Specifically, this can be an oversimplification, as the DNA methylation profile of the test sample type may not be appropriately correlated with the “training” sample type. A second approach is to perform initial training of the biological clock on multiple sample types. However, this approach has drawbacks including the need for a large number of samples to train a sufficiently robust biological clock, and the need to construct a “multi-sample” biological clock that can be applied to different sample types if different sample types are required, making this goal more challenging.

[0006] Therefore, alternative methods are needed to determine the biological age of subjects, especially when it is desirable to be able to use different sample types. Summary of the Invention

[0007] This invention relates to methods for quantifying the health status of subjects based on DNA methylation profiling. These methods enable the determination of a subject's biological age, risk of death, and / or probability of healthy lifespan by assessing the DNA methylation profile of the subject.

[0008] In a first aspect, the present invention provides a method for generating a biological clock comprising a DNA methylation map applicable to at least two different sample types, the method comprising: (i) Provide a first set of DNA methylation profiles generated from at least two different sample types from multiple subjects; (ii) Generate a composite DNA methylation map from the first set of DNA methylation maps, wherein the composite DNA methylation map contains methylation sites that are in a matching state in at least two different sample types; (iii) Using a reference DNA methylation map from at least one of at least two sample types, a composite DNA methylation map is used to generate a biological clock.

[0009] As used herein, a “complex DNA methylation map” can refer to a DNA methylation map containing DNA methylation sites selected as non-changing or stable in at least two different sample types. Appropriately, generating a complex DNA methylation map containing methylation sites with matching states in different sample types means that each sample type has DNA methylation sites with consistent and / or stable methylation states in each sample type used to generate the complex DNA methylation map in step (ii) of this method.

[0010] This "two-stage" process means that a composite DNA methylation map has been screened or rationalized to include DNA methylation sites known to provide stable or matching information in sample types of interest. Training a biological clock on a single sample type using such matched DNA methylation sites means that a biological clock trained on a DNA methylation map of a first sample type from at least two different sample types can be applied to test samples from a second sample type from at least two different sample types.

[0011] Therefore, the present invention provides that a biological clock can be trained on at least one sample type (e.g., blood), but the test sample can be any sample type used to generate the complex DNA methylation map in step (ii) of the method.

[0012] In this regard, it should be understood that large datasets are required to construct accurate and robust biological clocks. Therefore, a potential advantage of the method of this invention can include that the biological clock can be used for multiple sample types (e.g., any sample type used to generate complex DNA methylation profiles), but only one sample type is needed to train the biological clock. For example, a first sample type with sufficient data available (e.g., blood samples from large studies) can be used to train the biological clock; however, a single test sample can be a different second sample type used to generate complex DNA profiles (e.g., saliva or oral swab samples—easier to collect outside of a clinical setting for individuals).

[0013] To avoid being bound by theory, complex DNA methylation maps can be generated from samples from fewer individuals (i.e., biological replicates) compared to the number of corresponding samples required to construct a biological clock.

[0014] Step (ii) of the method of the present invention may include comparing a first set of DNA methylation maps, and: (1) if the methylation site has a matching state in the first set of DNA methylation maps from at least two different sample types, then the methylation site is included in the composite DNA methylation map; and / or (2) if the methylation site does not have a matching state in the first set of DNA methylation maps from at least two different sample types, then the methylation site is excluded from the composite DNA methylation map.

[0015] The method of the present invention may further include: (iv) providing a DNA methylation map from a test sample obtained from a test subject; and (v) using a biological clock generated from the composite DNA methylation map according to steps (i)-(iii) to determine the subject's biological age, risk of death, and / or probability of healthy lifespan.

[0016] In a second aspect, the present invention provides a method for determining a subject's biological age, risk of death, and / or probability of healthy lifespan; the method comprising: a) Provide DNA methylation maps from test samples obtained from subjects; and b) Use the biological clock generated by the composite DNA methylation map produced according to the method of the present invention to determine the subject's biological age, risk of death, and / or probability of healthy lifespan.

[0017] Existing methods for assessing a subject's health status typically determine biological age based on the correlation between DNA methylation and chronological age (see, for example, WO2022 / 272120). Calculating a subject's biological age may involve determining a DNA methylation profile compared to a predicted DNA methylation profile for a given chronological age. Therefore, this approach is based on using chronological age as the primary indicator of overall health.

[0018] In contrast, the present invention can also consider the direct predictive value of DNA methylation profiling for mortality risk and / or the probability of healthy lifespan. For example, a given DNA methylation biomarker may not be directly related to chronological age, but may indicate a specific pathological condition, thereby indicating the probability of increased mortality risk and / or reduced healthy lifespan. Therefore, the method of the present invention can be described as identifying the mortality risk and / or the probability of healthy lifespan of a subject. Thus, the DNA methylation biomarkers and DNA methylation profiling of the present invention are not necessarily related to chronological age, but rather to the difference between the subject's phenotypic age and chronological age.

[0019] A dog's biological age can be expressed in years, months, days, etc.

[0020] Determining mortality risk can refer to determining the likelihood that a subject will survive a longer or shorter period of time compared to comparable subjects of the same age, sex, and breed. Therefore, the method of the present invention can determine the probability of a subject's lifespan, healthy lifespan, and / or longevity compared to comparable subjects of the same age, sex, and breed. Furthermore, methods for improving a subject's mortality risk and / or healthy lifespan probability can improve the subject's likely lifespan, healthy lifespan, and / or longevity.

[0021] As used in this article, "lifespan" can refer to the length of time a subject survives (e.g., years). "Healthy lifespan" can refer to the length of time a subject lives without disease (e.g., years). "Longevity" can refer to the length of time a subject survives beyond their expected lifespan (e.g., years).

[0022] Suitablely, the risk of death can be equated with the probability of a subject's healthy lifespan; wherein a reduced risk of death is equivalent to an increased probability of a subject's longer healthy lifespan, or an increased risk of death is equivalent to a decreased probability of a subject's longer healthy lifespan. The risk of death can be expressed as the difference between a subject's determined age (i.e., biological age) and chronological age. For example, an increase in the difference between the biological age determined by the method of the present invention and chronological age can indicate an increased risk of death for the subject. A decrease in the difference between the biological age determined by the method of the present invention and chronological age can indicate a decreased risk of death for the subject. Suitablely, the risk of death and / or the probability of healthy lifespan can be described as the subject's biological age. Suitablely, the risk of death and / or the probability of healthy lifespan determined using the biomarkers of the present invention can be described as the subject's phenotype age. Suitablely, biological age, risk of death, and / or the probability of healthy lifespan can be described as the subject's epigenetic age. Suitablely, the biological clock of the present invention determined using DNA methylation mapping can be referred to as the epigenetic clock.

[0023] Appropriately, determine that a subject's biological age greater than their chronological age indicates a higher risk of death. Appropriately, determine that a subject's biological age less than their chronological age indicates a reduced risk of death. Appropriately, determine that a subject's biological age greater than their chronological age indicates a lower probability of a longer healthy lifespan. Appropriately, determine that a subject's biological age less than their chronological age indicates an increased probability of a longer healthy lifespan.

[0024] Suitablely, the method of the present invention can be used to determine the biological age of a subject based on the subject's biological age, risk of death, and / or probability of healthy lifespan.

[0025] The present invention also provides a method for selecting a lifestyle program, dietary program or therapeutic intervention for a subject, the method comprising: i) using a method according to the first aspect of the present invention to generate or as further defined herein a complex DNA methylation profile to determine the subject’s biological age, risk of death and / or probability of healthy lifespan; and ii) selecting an appropriate lifestyle program, dietary program or therapeutic intervention for the subject based on the biological age, risk of death and / or probability of healthy lifespan determined in step i).

[0026] As used in this article, “choosing a suitable lifestyle program, dietary program or treatment intervention for the subject” can also include “recommending a lifestyle program, dietary program or treatment intervention for the subject” or “providing the subject with a recommended lifestyle program, dietary program or treatment intervention”.

[0027] In another aspect, the present invention provides a method for determining the efficacy of a lifestyle program, dietary program, or therapeutic intervention in improving a subject's biological age, risk of death, and / or probability of healthy lifespan, the method comprising: a) administering the lifestyle program, dietary program, or therapeutic intervention to the subject, wherein the lifestyle program, dietary program, or therapeutic intervention has been selected according to the present invention; b) after administering the lifestyle program, dietary program, or therapeutic intervention to the subject for a period of time; determining the subject's biological age, risk of death, and / or probability of healthy lifespan using a DNA methylation profile from a test sample obtained from the subject, wherein a composite DNA methylation profile has been generated according to the method of the first aspect of the present invention, or a composite DNA methylation profile as further defined herein; c) determining whether the subject's biological age, risk of death, and / or probability of healthy lifespan has changed after the period of following the lifestyle program, dietary program, or therapeutic intervention.

[0028] In another aspect, the present invention provides a method for determining the efficacy of a lifestyle program, dietary program, or therapeutic intervention in improving a subject's biological age, risk of death, and / or probability of healthy lifespan, the method comprising: a) determining the subject's biological age, risk of death, and / or probability of healthy lifespan using a DNA methylation profile from a test sample obtained from the subject, wherein a composite DNA methylation profile has been generated according to the method of the first aspect of the present invention, or as further defined herein; b) administering to the subject a lifestyle program, dietary program, or therapeutic intervention selected based on the biological age, risk of death, and / or probability of healthy lifespan determined in step a); c) after a period of time following the administration of the lifestyle program, dietary program, or therapeutic intervention to the subject, determining the subject's biological age, risk of death, and / or probability of healthy lifespan using a DNA methylation profile from a second test sample obtained from the subject, wherein a composite DNA methylation profile has been generated according to the method of the first aspect of the present invention, or as further defined herein; d) determining whether the subject's risk of death and / or probability of healthy lifespan has changed between steps a) and c).

[0029] Appropriately, improving a subject's biological age, risk of death, and / or probability of healthy life can refer to a decrease in the difference between the subject's biological age and chronological age, wherein the subject's biological age is greater than their chronological age. Alternatively, improving a subject's biological age, risk of death, and / or probability of healthy life can refer to maintaining or further increasing the difference between the subject's biological age and chronological age, wherein the subject's biological age is less than their chronological age. Alternatively, deterioration of a subject's biological age, risk of death, and / or probability of healthy life can refer to an increase in the difference between the subject's biological age and chronological age, wherein the subject's biological age is greater than their chronological age. Deterioration of a subject's biological age, risk of death, and / or probability of healthy life can also refer to a decrease in the difference between the subject's biological age and chronological age, wherein the subject's biological age is less than their chronological age.

[0030] Appropriately, an improvement in a subject's risk of death and / or probability of healthy life can refer to a reduction in the rate of change between a subject's biological age and chronological age, where the subject's biological age is greater than their chronological age. For example, a subject's biological age may have already increased at a rate of 1.5 years per chronological age. Following a lifestyle and dietary intervention, a reduction in the rate of change, resulting in a subsequent increase in the subject's biological age of 1.25 years per chronological age, could provide an improvement in the subject's risk of death and / or probability of healthy life.

[0031] Improving biological age, mortality risk, and / or the probability of healthy lifespan can also refer to maintaining or increasing the rate of change between a dog's biological age and its chronological age, where the dog's biological age is less than its chronological age. For example, a dog's biological age may have already increased at a rate of less than one year (e.g., 0.9 years) per chronological age. Following lifestyle, dietary, or therapeutic interventions, the rate of change can be altered so that the dog's biological age subsequently increases by, for example, 0.8 years or less per chronological age, which can provide an improvement in the dog's biological age.

[0032] The method of the present invention for determining the efficacy of lifestyle programs, dietary programs, or therapeutic interventions in improving a subject's biological age, risk of death, and / or probability of healthy lifespan advantageously allows for the continuous monitoring of the effectiveness of lifestyle programs, dietary programs, or therapeutic interventions in improving or maintaining a subject's health. Using such methods advantageously allows for the identification of particularly effective lifestyle programs, dietary programs, or therapeutic interventions. In contrast, if a lifestyle program, dietary program, or therapeutic intervention is determined to be ineffective based on a subject's biological age, risk of death, and / or probability of healthy lifespan, alternative lifestyle programs, dietary programs, or therapeutic interventions can be implemented.

[0033] Therefore, the method of the present invention enables the selection of appropriate lifestyle programs, dietary programs, or therapeutic interventions for subjects based on their biological age, risk of death, and / or probability of healthy lifespan determined from DNA methylation profiles. For example, where the subject is a dog, a highly digestible and high-quality protein diet is typically recommended based on the dog's chronological age. For example, it might be recommended to transition the dog to a senior dog diet around 7 or 8 years of age. However, in the context of the present invention, determining that a dog has an increased biological age and / or an increased risk of death and / or a decreased probability of healthy lifespan (i.e., increased biological age) compared to its chronological age allows for the determination of transitioning the dog to a senior dog diet at an earlier age. In contrast, dogs with a decreased risk of death and / or an increased probability of healthy lifespan (i.e., decreased biological age) compared to their chronological age may be able to maintain an adult dog diet for a longer period.

[0034] Suitablely, the method of the present invention may include selecting and / or administering lifestyle programs, dietary programs or therapeutic interventions to the subject after determining that the subject has an increased biological age and / or an increased risk of death and / or a reduced probability of healthy lifespan compared to their actual age.

[0035] Appropriately, the disease is an age-related disease. For example, age-related diseases include osteoarthritis, dementia, cognitive impairment, prediabetes, diabetes, cancer, heart disease, obesity, gastrointestinal disorders, incontinence, kidney disease, sarcopenia, vision loss, hearing loss, osteoporosis, cataracts, cerebrovascular disease, and / or liver disease.

[0036] This method may optionally also include administering a lifestyle program, dietary program, or therapeutic intervention to the subject. Appropriately, the lifestyle program may be a dietary intervention or a therapeutic approach.

[0037] In another aspect, the present invention provides a method for selecting subjects as suitable subjects for receiving an anti-aging lifestyle program, dietary program, or therapeutic intervention; the method comprising: a) determining the subject's biological age, risk of death, and / or probability of healthy lifespan using a DNA methylation profile from a sample obtained from the subject, wherein a composite DNA methylation profile has been generated according to the method of the first aspect of the present invention, or a composite DNA methylation profile as further defined herein; and b) selecting the subject as suitable for receiving an anti-aging lifestyle program, dietary program, or therapeutic intervention if the subject has an increased biological age and / or an increased risk of death and / or a decreased probability of healthy lifespan compared to their actual age.

[0038] Appropriately, while anti-aging lifestyle programs, dietary programs, or therapeutic interventions may be effective for subjects based on chronological age, they may be particularly effective when applied to subjects who have an increased biological age and / or an increased risk of death and / or a reduced probability of healthy lifespan compared to their chronological age. Therefore, the method of the present invention can advantageously enable the selection of subjects for whom an anti-aging lifestyle program, dietary program, or therapeutic intervention has an increased likelihood of response or an improved degree of response.

[0039] Lifestyle programs, dietary programs, or therapeutic interventions can be selected based on determining that the subject has an increased biological age and / or an increased risk of death and / or a reduced probability of healthy lifespan (i.e., increased biological age) compared to their actual age.

[0040] Lifestyle programs, dietary programs, or therapeutic interventions can be dietary interventions. Dietary interventions can include calorie-restricted diets, senior dog diets, or low-protein diets.

[0041] DNA methylation profiles may be associated with increased biological age in the following areas: (i) tissues; (ii) organs; or (iii) physiological systems such as the immune system, gastrointestinal system, urinary system, muscular system, cardiovascular system, and / or nervous system.

[0042] The present invention also provides a dietary intervention for reducing the biological age and / or mortality risk of a subject and / or increasing the probability of a healthy lifespan of a subject, wherein the dietary intervention is applied to a subject whose biological age, mortality risk and / or probability of healthy lifespan have been determined by the method of the present invention.

[0043] The present invention also relates to the use of dietary interventions in reducing the biological age and / or risk of death and / or increasing the probability of a healthy lifespan of subjects, wherein the dietary intervention is applied to subjects whose biological age, risk of death, and / or probability of healthy lifespan have been determined by the method of the present invention.

[0044] In another aspect, the present invention provides a computer-readable medium comprising instructions that, when executed, cause one or more processors to perform the methods of the present invention.

[0045] In another aspect, the present invention provides a computer system for determining the biological age, risk of death, and / or probability of healthy lifespan of a subject; the computer system is programmed to use complex DNA methylation to determine the biological age, risk of death, and / or probability of healthy lifespan of a subject, wherein the complex DNA methylation is (i) generated by the method of the first aspect of the present invention, or (ii) contains DNA methylation sites as further defined herein. Attached Figure Description

[0046] Figure 1 - Identification of blood biomarkers for predicting mortality risk. The Cox proportional hazards model was adapted for each of the 28 biomarkers assessed, including sex and breed category (small or medium). Adjusted values ​​were used for the p-value for each parameter to account for multiple comparisons (by false discovery rate (fdr)). Parameters shown are those with an adjusted fdr below 0.05.

[0047] Figure 2 – Demonstration of biomarkers that contribute to the predictive power of multiparameter models used to determine phenotypic age.

[0048] Figure 3 - This illustrates the correlation between the blood and oral swab "multi-tissue" phenotypic clock of the present invention and actual age.

[0049] Figure 4 - This shows the correlation between complex DNA methylation profiles between blood and oral swab samples.

[0050] Figure 5 - This illustrates a validation study of a "multi-tissue" phenotypic clock from blood and oral swabs using data from this invention, which utilizes a lifetime calorie restriction study.

[0051] Figure 6 - An illustrative epigenetic clock is shown, containing the first 5, first 10, first 30, and first 50 methylation sites in an illustrative epigenetic clock constructed using a composite DNA methylation map between blood and oral swab samples.

[0052] Figure 7 - This shows the correlation of complex DNA methylation profiles between blood and oral swab samples for the first 5 sites, first 10 sites, first 30 sites, and first 50 sites.

[0053] Figure 8 - This invention illustrates the correlation between the "multi-tissue" phenotypic clock of blood, saliva, and oral swabs and actual age.

[0054] Figure 9 - This shows the correlation of complex DNA methylation profiles between blood and oral swab samples (Figure A) and between blood and saliva samples (Figure B).

[0055] Figure 10 - This study demonstrates a validation study of the "multi-tissue" phenotypic clock from blood, saliva, and oral swabs, using data from a lifetime calorie restriction study.

[0056] Figure 11- An illustrative epigenetic clock is shown, containing the first 5, first 10, first 30, and first 50 methylation sites in an illustrative epigenetic clock constructed using complex DNA methylation maps between blood, saliva, and oral swab samples. Detailed Implementation

[0057] Preferred features and embodiments of the invention will now be described by way of non-limiting examples. Those skilled in the art will understand that they can combine all the features of the invention disclosed herein without departing from the scope of the invention as disclosed.

[0058] It must be noted that, as used herein and in the appended claims, the singular forms “a,” “an,” and “the” include plural references, unless the context clearly specifies otherwise.

[0059] As used herein, the terms “comprising” and “consisting of” are synonymous with “including” or “containing”, and are inclusive or open-ended, and do not exclude additional unlisted members, elements or method steps. The terms “comprising” and “consisting of” also include the term “composed of”.

[0060] The range of numbers includes the numbers that define that range.

[0061] The publications discussed herein are provided only for their disclosure prior to the filing date of this patent application. Nothing herein should be construed as an admission that such publications constitute prior art to the claims appended herein.

[0062] The methods and systems disclosed in this article can be used by veterinarians, healthcare professionals, laboratory technicians, pet care providers, and others.

[0063] Subjects

[0064] The subjects in this invention can be any subject for whom a determination of biological age is desired.

[0065] Appropriately, the subjects can be mammals.

[0066] Appropriately, subjects may be dogs, cats, or humans.

[0067] Preferably, the subject is a dog or a cat; most preferably, a dog.

[0068] Unless otherwise stated, all the contents of this article apply equally to dogs, cats, or humans.

[0069] variety

[0070] In embodiments of the invention where the subject is a dog, the method of the invention can utilize information about the dog breed. For example, dogs can be classified as toy, small, medium, large, or giant breeds. Suitablely, dog breeds can be classified based on the dog's weight. Suitablely, dog breeds can be classified based on the average weight of a given breed of dog.

[0071] Dogs may be classified as small or medium breeds. Classification is determined by the average weight of the adult dogs of that breed. Breeds with an average weight of less than 10 kg are classified as small breeds, and / or breeds with an average weight of more than 10 kg are classified as medium breeds.

[0072] In the alternative aspect where the subject is a cat, the cat can be a domestic cat. Appropriately, the cat can be a domestic shorthaired cat.

[0073] gender

[0074] Appropriately, the sex of the subjects can be classified as male or female.

[0075] Actual age

[0076] Chronological age can be defined as the amount of time elapsed from a subject's birth to a given date. Chronological age can be expressed in years, months, days, etc.

[0077] Appropriately, the method of the present invention can be applied to subjects of any actual age.

[0078] When the subject is a dog, the dog may be at least about 2 years old. Appropriately, the dog may be at least about 2 years old, at least about 3 years old, at least about 4 years old, at least about 5 years old, at least about 6 years old, at least about 7 years old, at least about 8 years old, at least about 9 years old, or at least about 10 years old.

[0079] Appropriately, the dog can be at least about 7 years old.

[0080] sample

[0081] This invention relates to biological clocks and / or methods for determining a subject's biological age, risk of death, and / or probability of healthy lifespan, which can be used for various sample types.

[0082] composite DNA atlas

[0083] The method of the present invention includes providing a first set of DNA methylation maps generated from at least two different sample types from multiple subjects, and generating a composite DNA methylation map from the first set of DNA methylation maps, wherein the composite DNA methylation map contains methylation sites having a matching state in at least two different sample types.

[0084] Complex DNA methylation profiles can be generated from or applied to at least two different sample types. Appropriately, "at least two different sample types" can refer to at least two, at least three, at least four, at least five, or at least ten different sample types. Appropriately, "at least two different sample types" can refer to at least two, at least three, at least four, or at least five different sample types. Appropriately, "at least two different sample types" can refer to two, three, four, or five different sample types.

[0085] Appropriately, "at least two different sample types" can refer to two or three different sample types.

[0086] At least two different sample types can be any sample type containing DNA from which a DNA methylation map can be generated.

[0087] Appropriately, the sample may be a blood, oral swab, saliva, feces, hair (e.g., hair follicles), skin, or organ tissue sample.

[0088] Appropriately, at least two different sample types are independently selected from blood, oral swabs, saliva, feces, hair (e.g., hair follicles), skin, and organ tissue samples.

[0089] Appropriately, at least two different sample types are included, such as blood, oral swabs, and saliva samples.

[0090] Appropriately, at least two different sample types may include blood and oral swab samples.

[0091] Appropriately, at least two different sample types may include blood and saliva samples.

[0092] Suitablely, the sample is derived from blood. The sample may contain blood components or may be whole blood. The sample preferably includes whole blood. The sample may include peripheral blood mononuclear cell (PBMC) or lymphocyte samples. Techniques for collecting samples from subjects and extracting DNA (e.g., genomic DNA) from the samples are well known in the art.

[0093] Suitable, the at least two different sample types used to generate the complex DNA methylation map can come from at least 5, at least 10, at least 20, at least 50, or at least 100 subjects. Advantageously, the number of subjects from the at least two different sample types required to generate the complex DNA methylation map can be less than the number of subjects from the sample types used to generate the biological clock.

[0094] Appropriately, at least two different sample types used to generate a complex DNA methylation map are collected in each subject at the same time (e.g., less than 30 days, less than 14 days, less than 7 days, less than 72 hours, less than 48 hours, less than 24 hours, less than 12 hours, or less than 6 hours apart).

[0095] Using references DNA Methylation profile generation biological clock

[0096] Advantageously, the biological clock according to the invention can be trained on DNA methylation maps from a subset of sample types from at least two different sample types used to generate a complex DNA methylation map.

[0097] Suitable, the biological clock according to the invention can be generated using a reference DNA methylation map from at least one of at least two sample types used to generate a complex DNA methylation map.

[0098] Suitablely, the biological clock according to the invention can be generated using reference DNA methylation maps from at least n-1 sample types of at least two sample types used to generate the complex DNA methylation map. For example, if the complex DNA methylation map is generated from two different sample types, the biological clock can be generated using a single sample type from at least two sample types used to generate the complex DNA methylation map. In another instance, if the complex DNA methylation map is generated from three different sample types, the biological clock can be generated using one or two sample types from at least two different sample types used to generate the complex DNA methylation map.

[0099] In a particularly preferred embodiment, the biological clock according to the invention can be generated using a reference DNA methylation map from a single sample type used to generate a complex DNA methylation map.

[0100] Appropriately, the biological clock according to the invention can be generated using reference DNA methylation maps from at least two sample types used to generate a complex DNA methylation map.

[0101] Appropriately, biological clocks can be trained on DNA methylation profiles from blood samples.

[0102] Appropriately, the biological clock can be trained on DNA methylation profiles from samples of at least 100, at least 200, at least 400, at least 600, at least 800, at least 1000, at least 2000, or at least 5000 subjects.

[0103] Test samples

[0104] The invention may also include providing a DNA methylation map from a test sample obtained from a test subject; and using a biological clock generated from the composite DNA methylation map using the method according to the invention to determine the subject's biological age, risk of death, and / or probability of healthy lifespan.

[0105] As used herein, a “test” sample may refer to a sample used to determine a subject’s biological age, risk of death, and / or probability of healthy lifespan using the biological clock according to the present invention.

[0106] The test sample can be any sample type used to generate a complex DNA methylation map. Specifically, the test sample can be any sample type used to generate a complex DNA methylation map before generating the biological clock according to the present invention.

[0107] Suitable test samples may be oral swabs, saliva, or hair follicle samples. Such sample types are particularly suitable if the test samples are to be provided outside a veterinary setting, for example – for example, using a kit according to the invention.

[0108] The method of this invention can be performed on test samples obtained from the subject at different time points. For example, the method can be performed using a first test sample obtained at a given time point and a second test sample obtained after a time interval following the acquisition of the first test sample. The method can be performed more than once on test samples obtained from the same test subject over a period of time. For example, test samples can be obtained repeatedly monthly, annually, or every two years. Suitablely, test samples can be obtained approximately once a year (e.g., during an annual veterinary health check). This can be useful for determining the effects of specific treatments or lifestyle changes, such as dietary interventions or changes in exercise programs.

[0109] In one implementation, the method can be applied to test samples obtained from subjects prior to lifestyle changes (e.g., dietary product interventions or exercise program changes). In another implementation, the method can be applied to test samples obtained from subjects both before and after, for example, dietary product interventions or exercise program changes. The method can also be applied to test samples obtained at predetermined times throughout, for example, the dietary product intervention or exercise program change. These predetermined times can be periodic throughout, for example, the dietary product intervention or exercise program change, such as daily or every three days, or can depend on the subjects being tested.

[0110] DNA methylation

[0111] DNA methylation is the process of covalently adding a methyl group (CH3) to a cytosine base that is part of a DNA molecule. In vivo, this process is catalyzed by the DNA methyltransferase (Dnmt) family, which produces modified cytosine by transferring a methyl group from S-adenosylmethionine (SAM). The cytosine is modified at the 5th carbon atom, and the modified residue is called 5-methylcytosine (5mC). DNA methylation may also include 5-hydroxymethylcytosine (5hmc).

[0112] DNA methylation is an example of an epigenetic mechanism that can modify gene expression without altering the underlying DNA sequence. DNA methylation can suppress gene expression, for example, by acting as a recruitment signal for repressors or by directly blocking the recruitment of transcription factors. DNA methylation primarily occurs in the genome of mammalian somatic cells at sites where dinucleotides (CpGs) are adjacent to cytosine and guanine. While non-CpG methylation is observed in embryonic development, these modifications are significantly reduced in most cell types in adults. CpG islands are DNA segments with high CpG density but are typically unmethylated. These regions are associated with promoter regions, particularly those of housekeeping genes, and are thought to be kept in a permissive state that allows gene expression.

[0113] DNA methylation has been found to change with age in humans and other animals. Aging mammalian tissues show overall DNA hypomethylation, thought to be due to the gradual loss or mistargeting of DMNT1 methyltransferase activity, but with localized hypermethylation of CpG islands. Localized hypermethylation can lead to the repression of certain genes, and this can contribute to age-related diseases. The link between epigenetic changes in DNA methylation and age allows the use of “DNA methylation clocks” to estimate “biological age.” Typically, these clocks have been trained against chronological age using supervised machine learning methods, and the deviation of the “clock age” from an individual’s actual chronological age is considered an indicator of “biological” age. This is related to an individual’s chronological age, but deviations from this correlation can indicate a potential risk of age-related diseases or disorders in the individual.

[0114] Detection of specifically methylated DNA can be accomplished using a variety of methods (see, for example, Zuo et al., 2009; Epigenomics. 1(2): 331-345) and Rauluseviciute et al.; Clinical Epigenetics; 2019; 11(193)). Numerous methods can be used to detect differentially methylated DNA at specific loci in samples such as blood, urine, feces, or saliva. These methods are capable of distinguishing 5-methylcytosine- or methylated DNA from unmethylated DNA and subsequently quantifying the ratio of methylated to unmethylated DNA at specific genomic loci.

[0115] This method may include using any suitable method to determine the subject's DNA methylation profile. Suitable methods include, but are not limited to, those described below.

[0116] Enzymatic methylation sequencing ( EM-seq )

[0117] Suitable for this purpose, enzymatic methods are used to detect 5mC and 5hmC. For example, enzymatic methylation sequencing (EM-seq) can be used.

[0118] In EM-seq, typically in the first enzymatic step, 5mC is oxidized to 5hmC, then to 5fC, and finally to 5caC by the activity of Tet methylcytosine dioxygenase 2 (TET2). Additionally, the use of T4-BGT enzyme glycosylates both the pre-existing 5hmC and the 5hmC generated by TET2 activity. In the second enzymatic step, after double-stranded DNA denaturation, the enzyme apolipoprotein B mRNA editing enzyme catalyzes peptide-like 3A (APOBEC3A) to deaminate cytosine, but not to the oxidized or glycosylated forms of 5mC and 5hmC. Only unmethylated cytosine is deaminated to form uracil bases. Prior to the first enzymatic step, DNA fragments can be generated by mechanical shearing and end repair, A-tailing, and ligation into sequencing adaptors, which can be achieved using, for example, NEBNext. ® DNA Ultra II reagent (NEB) is used. After the second enzymatic step, PCR can be performed using a polymerase that amplifies a template containing uracil (such as NEBNext). ® Q5U ™ EM-seq amplifies deamination-bound single-stranded DNA and can sequence or analyze the resulting library in the same manner as DNA samples generated by bisulfite sequencing. EM-seq output is typically the same as whole-genome bisulfite sequencing, but uses fewer DNA-damaging agents, thus reducing sample loss and offering superior coverage, sensitivity, and accuracy in cytosine methylation recall compared to bisulfite-converted samples. An illustrative EM-seq method is described by Vaisvila et al. (Genome Research; 2021;31:1-10).

[0119] Methods based on bisulfite conversion

[0120] When treated with sodium bisulfite, bisulfite conversion utilizes the selective conversion of unmethylated cytosine to uracil. Denatured DNA is treated with sodium bisulfite, which converts all unmodified cytosine to uracil, and subsequent PCR amplification converts these residues to thymine. Analysis of the resulting DNA sequences can be performed using a variety of methods, examples of which include, but are not limited to: denaturing gel electrophoresis, single-strand conformation polymorphism, melting curves, real-time fluorescence PCR (MethyLight), MALDI mass spectrometry, array hybridization, and sequencing (e.g., whole-genome bisulfite sequencing, WGBS). Recently developed techniques (such as SeqCap Epi) enrich the sequences of interest before sequencing, enabling deeper coverage of more concentrated regions. Comparing the sequence abundance in bisulfite-converted samples with that in untreated controls allows analysis of methylation at target sites, where the proportion of converted sequences indicates the level of methylation at the target site.

[0121] Other variations of the bisulfite conversion method are available that can distinguish 5mC from its oxidized form, 5-hydroxymethylcytosine (5hmC), which behaves identically to 5mC under standard bisulfite conversion, and can detect further modifications such as 5-formylcytosine (5fC). These methods, such as oxBS-Seq and redBS-Seq, utilize the oxidation and reduction of these markers to alter the sensitivity of each substance to bisulfite conversion and quantify the amount of each modification at the target locus through comparative analysis.

[0122] Selective restriction endonuclease digestion method

[0123] Existing methods for analyzing DNA methylation patterns may involve the use of restriction enzymes. These methods include, for example, restriction marker genome scanning (RLGS) (Costello et al., 2000; Nat Genet.; 24(2):132-8), methylation sensitivity representative differential analysis (MS-RDA) (Ushijima et al., Proc Natl Acad Sci US A. 18 March 1997; 94(6):2284-9), and differential methylation hybridization (DMH) (Huang et al., Cancer Res. 15 March 1997; 57(6):1030-4). The digestive activity of restriction endonucleases can be methylation-dependent. This specificity can be used to distinguish between methylated and unmethylated sequences. Some restriction enzymes (e.g.) Bst UI Hpa II and Not I) Sensitive to methylated recognition sequences. Other restriction enzymes (such as...) Mcr BC is specific to methylated sequences.

[0124] For example, differential methylation hybridization (DMH) (Huang et al., ibid.) requires the use of bulk genome restriction enzymes (such as...) Mse I) Initial fragmentation of the genome is performed using an enzyme that breaks the genome into fragments smaller than 200 bp. Following this step, the genome fragments are digested using a methylation-sensitive restriction endonuclease (MRE) or, in some versions of this technique, a mixture of MREs to improve coverage. Depending on the specificity of the one or more enzymes used, methylated or unmethylated sequences will be degraded. The digested sequences are not amplified in subsequent PCR steps. The resulting PCR products are suitable for further processing and analysis by sequencing or microarray hybridization in combination with fluorescent dyes.

[0125] Suitablely, this method utilizes DNA methylation maps generated by methods including the use of one or more MREs.

[0126] Appropriate comparators can be used to study methylation status between conditions. Changes in methylation status can be detected by comparing DNA from healthy subjects with that from older or diseased subjects (Huang et al., Hum Mol Genet. 1999 Mar; 8(3):459-70). Alternatively, methylation-insensitive forms of secondary digestive enzymes (such as...) Hpa II isolytic enzyme Msp I) It can be used to generate control samples, enabling comparisons of DNA methylation within or between genomes (Khulan et al., Genome Res. 2006 Aug; 16(8):1046-55).

[0127] In some embodiments, methods for detecting methylation include randomly shearing or fragmenting genomic DNA, cutting the DNA with a methylation-dependent or methylation-sensitive restriction enzyme, and subsequently selectively identifying and / or analyzing the cut or uncut DNA. Selective identification may include, for example, separating the cut and uncut DNA (e.g., by size) and quantifying the cut or alternatively uncut sequences of interest. Alternatively, the method may encompass amplifying intact DNA after restriction enzyme digestion, thereby amplifying only the DNA that has not been cut by the restriction enzyme in the amplified region. In some embodiments, gene-specific primers may be used for amplification. Alternatively, an adaptor may be added to the end of the randomly fragmented DNA, the DNA may be digested with a methylation-dependent or methylation-sensitive restriction enzyme, and the intact DNA may be amplified using primers that hybridize to the adaptor sequence. In this case, a second step may be performed to determine the presence, absence, or amount of a specific gene in the DNA amplification pool. In some embodiments, real-time quantitative PCR is used to amplify the DNA.

[0128] Suitable methods for detecting nucleic acid digestion include selective hybridization of a probe or primer with undigested nucleic acids. Alternatively, the probe selectively hybridizes with both digested and undigested nucleic acids, but the distinction between the two forms is facilitated, for example, by electrophoresis. Suitable detection methods for achieving selective hybridization with the hybridization probe include, for example, Southern blotting or other nucleic acid hybridization.

[0129] Suitable hybridization conditions can be determined based on the melting temperature (Tm) of the nucleic acid duplex containing the probe. Those skilled in the art will recognize that optimal hybridization reaction conditions for each probe should be determined empirically, but some general principles can be applied. Preferably, hybridization with short oligonucleotide probes is performed with low to moderate stringency. High stringency hybridization and / or washing is preferred in the case of GC-rich probes or primers, or longer probes or primers. High stringency is defined herein as hybridization and / or washing performed in about 0.1×SSC buffer and / or about 0.1% (w / v) SDS or lower salt concentration and / or at a temperature of at least 65°C or equivalent conditions. Specific stringency levels mentioned herein cover equivalent conditions using wash / hybridization solutions other than SSC known to those skilled in the art.

[0130] Simplified representation of bisulfite sequencing ( RRBS )

[0131] Simplified representation of bisulfite sequencing (RRBS) uses the MspI restriction enzyme to enrich CpG-rich genomic regions—which cuts DNA at all CpG sites regardless of their DNA methylation status at CG sites—and is able to measure 5% to 10% of DNA methylation levels at all CpG sites in the mammalian genome.

[0132] Therefore, this method involves digesting DNA with a methylation-insensitive MspI prior to bisulfite conversion and sequencing. Digesting genomic DNA with MspI produces fragments that always begin with C (if cytosine is methylated) or T (if cytosine is not methylated and is converted to uracil during the bisulfite conversion reaction). This results in a non-random base pair composition. Furthermore, the basic composition is skewed due to the skewed frequencies of C and T within the sample. Various software programs are available for alignment and analysis, such as Maq, BS Seeker, Bismark, or BSMAP. Alignment with a reference genome allows the procedure to identify methylated base pairs within the genome.

[0133] Affinity-based enrichment methods

[0134] Methylated DNA can be distinguished from unmethylated DNA by using antibodies containing a methyl-CpG-binding domain (MBD), such as anti-5mC and / or methylated CpG-binding proteins. Antibodies with MBD-domain proteins can specifically separate methylated DNA relative to unmethylated DNA. The antibody-based method is commonly referred to as MeDIP, while the method using methylated CpG-binding proteins is commonly referred to as the MBD or MIRA method.

[0135] These methods require initial fragmentation of the genome, which can be achieved using enzymes that frequently cleave the genome (such as...). Mse I) Perform extensive genome digestion, followed by affinity purification of methylated fragments. The input DNA can be compared with the purified methylated DNA via microarray hybridization or sequencing to obtain comparative analysis of methylation levels at specific sites.

[0136] Other variations of affinity-based enrichment methods are available, such as MethylCap-Seq or MBD-Seq. These methods reduce sample complexity by using a salt gradient to elute methylated DNA fragments in a methyl-CpG abundance-dependent manner, separating CpG islands and other highly methylated loci from loci with lower CpG density. These fragments can then be sequenced separately, thereby improving sequence coverage.

[0137] Based on single-molecule sequencing and de novo methylation sequencing methods

[0138] Modern sequencing methods can sequence individual molecules directly. Single-molecule real-time (SMRT) DNA sequencing is available, such as the Sequel system from Pacific Biosciences, and has been shown to identify modified bases (such as methylated cytosine) based on polymerase kinetics. Nanopore sequencing devices, such as the MinION, Gridion, and Promethion nanopore sequencers from Oxford Nanopore Technologies, can sequence long chains of DNA individually and detect de novo base modifications, including methylation.

[0139] DNA methylation sites

[0140] Appropriately, DNA methylation sites may refer to the presence or absence of 5mC at a single cytosine, or appropriately a single CpG dinucleotide.

[0141] Suitablely, a DNA methylation site can refer to the presence or absence of methylation at multiple CpG sites across a DNA region (i.e., the number or percentage of 5mC). Suitablely, a DNA methylation site can refer to the methylation level at multiple CpG sites across a DNA region (i.e., the number or percentage of 5mC). A “DNA region” can refer to a specific portion of genomic DNA. These DNA regions can be designated by reference gene names or a set of chromosomal coordinates. Both gene names and chromosomal coordinates are well known and understood by those skilled in the art.

[0142] Appropriately, gene names and / or coordinates may be based on the “Tasha” canine reference genome (https: / / www.ncbi.nlm.nih.gov / assembly / GCF_000002285.5; Jagannathan et al.; Genes (Bsael); 2021; 12(6); 847) or the “CanFam3.1” canine reference genome (https: / / www.ncbi.nlm.nih.gov / datasets / genome / GCF_000002285.3 / ; Lindblad-Toh et al.; Nature 438, 803–819(2005)).

[0143] For example, a DNA region can define the portion of DNA near the promoter of a gene. Promoter regions are known to be CpG-rich. For instance, a DNA region could refer to approximately 3 kb upstream and 3 kb downstream of the promoter; approximately 2 kb upstream and 2 kb downstream; approximately 2 kb upstream and 1 kb downstream; approximately 2 kb upstream and 0.5 kb downstream; approximately 1 kb upstream and 0.5 kb downstream; or approximately 0.5 kb upstream and 0.5 kb downstream. Suitablely, a DNA region could refer to approximately 1 kb upstream and 0.5 kb downstream of the promoter.

[0144] DNA regions can define other DNA segments that may be located therein – including but not limited to CpG islands, enhancers, open chromatin, transcription factor binding sites, and miRNA promoter regions.

[0145] Suitable, the DNA region may contain or consist of CpG sites spaced less than about 5,000, 4,000, 3,000, 2,000, 1,000, 500, or 200 bases apart.

[0146] Suitable, the DNA region may contain or consist of CpG sites spaced about 200 to about 5,000, about 200 to about 4,000, about 200 to about 3,000, about 200 to about 2,000, or about 200 to about 1,000 bases apart.

[0147] Suitablely, a DNA region may contain one or more CpG islands. Suitablely, a DNA region may consist of CpG islands.

[0148] “CpG islands” can refer to DNA regions containing at least 200 bp, a GC percentage greater than 50%, and an observed CpG ratio greater than 60% of the expected CpG.

[0149] Suitable, the DNA methylation site does not contain X and / or Y chromosome CpG.

[0150] Suitable, the DNA methylation site does not contain CpGs that are known to contain SNPs at the CpG site.

[0151] References to each gene / DNA region detailed above should be understood as references to all forms of these molecules and their fragments or variants. As will be understood by those skilled in the art, some genes are known to exhibit allelic variations or single nucleotide polymorphisms among individuals. Variants include nucleic acid sequences from the same region sharing at least 90%, 95%, 98%, or 99% sequence identity, i.e., having one or more deletions, additions, substitutions, reverse sequences, etc., relative to the DNA region described herein. Therefore, the invention should be understood to extend to such variants, which, for the purposes of this application, can achieve the same results despite minor genetic variations in the actual nucleic acid sequences between individuals. Thus, the invention should be understood to extend to all forms of DNA resulting from any other mutation, polymorphism, or allelic variation.

[0152] Regarding the screening of methylation in these gene regions, it should be understood that the assay can be designed to screen specific DNA. Selecting which strand to analyze and targeting based on chromosome coordinates is entirely within the skill of those skilled in the art. In some cases, assays can be developed to screen two strands.

[0153] "Methylation state" can be understood as the presence, absence, and / or amount of methylation at one or more specific nucleotides within a DNA region. The methylation state of a particular DNA sequence (e.g., a DNA region as described herein) may indicate the methylation state of each base in the sequence, or the methylation state of a subset of base pairs within the sequence (e.g., the methylation state of cytosine or the methylation state of one or more specific restriction enzyme recognition sequences), or it may indicate information about the regional methylation density within the sequence without providing precise information about where methylation occurs within the sequence. Methylation state may optionally be represented or indicated by a "methylation value."

[0154] Suitablely, an EM-Seq strategy can be used to determine DNA methylation. In this method, the methylation level can be determined as the fraction of "C" bases in the total "C"+"U" bases at the target CpG site "i" after enzyme and APOBEC3A transformation treatment. In other embodiments, the methylation level can be determined as the fraction of "C" bases in the total "C"+"T" bases at site "i" after enzyme and APOBEC3A transformation treatment and subsequent nucleotide amplification. The average methylation level at each site can then be evaluated to determine whether one or more thresholds are met.

[0155] In some implementations, particularly when using bisulfite conversion and sequencing methods, the methylation level can be determined as the fraction of "C" bases in the total "C"+"U" bases at the target CpG site "i" after bisulfite treatment. In other implementations, the methylation level can be determined as the fraction of "C" bases in the total "C"+""T" bases at site "i" after bisulfite treatment and subsequent nucleotide amplification. The average methylation level at each site can then be evaluated to determine whether one or more thresholds are met.

[0156] Alternatively, methylation values ​​can be generated, for example, by quantifying the amount of intact DNA present after restriction digestion with a methylation-dependent restriction enzyme. In this example, if a specific sequence in the DNA is quantified using quantitative PCR, an amount of template DNA approximately equal to that of the simulated treatment control indicates that the sequence is not highly methylated, while a amount substantially less than that of the template present in the simulated treatment sample indicates the presence of methylated DNA at that sequence. Therefore, values ​​from the example above (i.e., methylation values) represent the methylation status and can thus be used as a quantitative indicator of methylation status. This is particularly useful when it is necessary to compare the methylation status of a sequence in a sample with a threshold.

[0157] This invention is not limited to the exact number of methylated residues considered to indicate biological age, as some variation will occur between samples. This invention is also not necessarily limited to the location of methylated residues (e.g., specific methylation sites).

[0158] In one implementation, a screening method may be employed that is specifically designed to assess the methylation status of one or more specific cytosine residues or the corresponding cytosine at position n+1 on the DNA strand.

[0159] Enrichment and Detection Methods

[0160] Determining a DNA methylation profile may include the step of enriching selected DNA regions in a DNA sample. For example, the method may include the step of enriching DNA regions in a DNA sample that contain DNA methylation sites, which constitute a DNA methylation profile.

[0161] Suitable enrichment methods are known in the art and include, for example, amplification-based or hybridization-based methods. Amplification enrichment typically refers to PCR-based enrichment, for example, using primers targeted at the DNA region to be enriched. Any suitable form of amplification can be used, such as polymerase chain reaction (PCR), rolling circle amplification (RCA), reverse polymerase chain reaction (iPCR), in situ PCR, strand displacement amplification, or cycling probe techniques.

[0162] Hybridization enrichment, or capture-based enrichment, typically refers to the use of hybridization probes (or capture probes) that hybridize with the DNA region to be enriched.

[0163] Hybridization probes can be directly attached to a solid vector, or they can contain a portion, such as biotin, to allow binding to a solid vector (e.g., beads coated with streptavidin) suitable for capturing the biotin portion. In either case, DNA containing a sequence complementary to the probe can be captured, allowing the separation of DNA containing the region of interest from DNA not containing that region. Therefore, such a capture step allows for the enrichment of the region of interest. For example, the DNA region could be a region adjacent to a gene promoter.

[0164] The arrays used in this article can vary depending on the probe composition and the intended use of the array. For example, the number of nucleic acids (or CpG sites) detected in the array can be at least 10, 100, 1,000, 10,000, 100,000, 1 million, 10 million, 100 million, or more. Alternatively or additionally, the number of nucleic acids (or CpG sites) detected can be selected to be no more than 100 million, 10 million, 1 million, 100,000, 10,000, 1,000, 100, or fewer. Similar ranges can be obtained using nucleic acid sequencing methods, such as those known in the art; for example, next-generation or massively parallel sequencing.

[0165] Appropriately, the enrichment step can be performed before or after the step of separating or differentiating methylated and unmethylated DNA.

[0166] As used herein, the term "enrichment" or "DNA" or "DNA region" refers to the process of increasing the (absolute) amount and / or proportion of DNA containing the desired sequence compared to the amount and / or proportion of DNA containing the desired sequence in the starting material. In this respect, enrichment by amplification increases the amount and proportion of the desired sequence. Enrichment by capture-based enrichment increases the proportion of DNA containing the desired sequence.

[0167] After processing the DNA to distinguish between methylated and unmethylated sites, this method may also include a step of identifying methylated or unmethylated sites (i.e., in the original sample).

[0168] The identification steps may include any suitable method known in the art, such as array detection or sequencing (e.g., next-generation sequencing).

[0169] The sequencing qualification step preferably includes next-generation sequencing (massively parallel or high-throughput sequencing). Next-generation sequencing methods are well known in the art, and in principle, any method can be considered for use in this invention. Next-generation sequencing technology can be performed according to the manufacturer's instructions (e.g., provided by Roche, Illumina, or Applied Biosystems).

[0170] In one implementation, samples are processed by using an enzymatic reaction to convert DNA to methylation, preparing whole-genome libraries, and measuring methylation profiles by sequencing (EM-Seq).

[0171] In one embodiment, the sample is processed by using an enzymatic reaction to convert DNA to methylation, preparing a whole-genome library, hybridizing the converted library with a capture probe (preferably a capture probe capable of capturing DNA regions near gene promoters), and measuring the methylation profile by sequencing (EM-Seq).

[0172] In some implementations (e.g., using the DNA methylation maps provided in Tables 3-6), commercially available DNA methylation arrays can be used to perform the methods of the present invention.

[0173] Appropriately, DNA methylation is converted by bisulfite conversion, the converted DNA is optionally amplified, and then labeled (e.g., with a fluorescent dye) and hybridized with a methylation array (e.g., a mammalian methylation array) to process the sample. Suitable methylation arrays are available, for example, from Illumina and are described in WO20150705 and Arneson et al. (Nature Communications; 13(782); 2022).

[0174] DNA methylation map

[0175] A “DNA methylation profile” or “methylation map” can refer to the presence, absence, amount, or level of 5mC at one or more DNA methylation sites. Preferably, a “methylation profile” refers to the presence, absence, amount, or level of 5mC at multiple DNA methylation sites. Therefore, the presence, absence, amount, or level of 5mC at each individual DNA methylation site within multiple sites can be assessed, and this helps determine the subject’s risk of death and / or probability of healthy lifespan. Therefore, the quality and / or efficacy of this method can be improved by combining values ​​from multiple DNA methylation markers.

[0176] Suitablely, the biological clock of the present invention includes methylation maps from multiple methylation sites.

[0177] Appropriately, the presence or absence of 5mC from at least 3, at least 5, at least 10, at least 20, at least 50, at least 100, at least 200, at least 500, at least 1000, at least 2000, at least 5000, at least 10000, at least 50000, at least 10000, at least 250000, or at least 500000 DNA methylation sites can be used to determine the subject’s risk of death and / or probability of healthy lifespan (i.e., biological age).

[0178] Suitablely, a methylation profile can refer to the presence or absence of 5mC from at least 100, at least 200, at least 500, at least 1000, or at least 2000 DNA methylation sites.

[0179] Appropriately, a methylation profile can refer to the presence or absence of 5mC from approximately 100, 200, 500, 1000, or 2000 DNA methylation sites.

[0180] To generate a biological clock for determining mortality risk and / or the probability of healthy lifespan, an initial methylation map can be processed or simplified to produce a restricted methylation map, which can then be used to generate the biological clock.

[0181] For example, the initial methylation profile can be processed or simplified by, for instance, using DNA regions instead of individual cytosines, selecting a subset of methylation sites associated with a specific physiological or biochemical pathway, performing correlation analysis and retaining one or more representative DNA methylation sites for each cluster, or performing differential analysis to preselect DNA methylation sites or retain DNA methylation sites that vary more between young and old subjects.

[0182] For example, a DNA region can be any DNA region as defined herein.

[0183] Suitablely, a methylation map can point to DNA methylation sites of genes associated with specific physiological or biochemical pathways. Therefore, a methylation map can enable the determination of the biological age of a specific tissue, organ, or physiological system. Determining the biological age of a specific tissue, organ, or physiological system advantageously allows the method to be used in a manner focused on the pathology and disease of that tissue, organ, or physiological system. For example, if a particular breed of dog is known to be associated with muscular or cardiovascular disease, determining the biological age of that physiological system may be advantageous.

[0184] Appropriately, the physiological system can be the inflammatory system, the muscular system, the cardiovascular system, and / or the nervous system.

[0185] The biological age of a specific tissue, organ, or physiological system can be determined using a DNA methylation map, which contains or is composed of methylation sites of genes preferentially or specifically expressed from that tissue, organ, or physiological system. Gene classification by specific tissue, organ, or physiological system is publicly available at, for example, Gene Ontology (http: / / geneontology.org / ), the KEGG pathway database (https: / / www.genome.jp / kegg / ), or MSIgDB (https: / / www.gsea-msigdb.org / gsea / msigdb / index.jsp).

[0186] In some implementations, the threshold is selected for those sites that have the highest average methylation values ​​for epigenetic age predictors. For example, the threshold could be those sites with the highest average methylation levels, which are the top 50%, top 40%, top 30%, top 20%, top 10%, top 5%, top 4%, top 3%, top 2%, or top 1% of the average methylation levels of all sites “i” tested for predictors such as biological clocks.

[0187] Alternatively, the threshold may be those sites where the average methylation level is at or above the percentile of 50, 60, 70, 80, 90, 95, 96, 97, 98, or 99. In other embodiments, the threshold may be based on the absolute value of the average methylation level. For example, the threshold may be those sites where the average methylation level is greater than 99%, greater than 98%, greater than 97%, greater than 96%, greater than 95%, greater than 90%, greater than 80%, greater than 70%, greater than 60%, greater than 50%, greater than 40%, greater than 30%, greater than 20%, greater than 10%, greater than 9%, greater than 8%, greater than 7%, greater than 6%, greater than 5%, greater than 4%, greater than 3%, or greater than 2%. Relative and absolute thresholds may be applied alone or in combination to the average methylation level at each site “i”. As an illustration of the combined threshold application, a subset of sites may be selected from the top 3% of all sites that pass the average methylation level test and also have an absolute average methylation level greater than 6%. The result of this selection process is a DNA methylation profile of a specific hypermethylation site (e.g., a CpG site), which is considered to be the most informative for determining the probability of death risk and / or healthy lifespan.

[0188] complex DNA Methylation map

[0189] As used herein, a “complex DNA methylation map” can refer to a DNA methylation map containing DNA methylation sites selected as non-changing or stable in at least two different sample types. Appropriately, a complex DNA methylation map containing methylation sites with matching states in different sample types means that the DNA methylation sites in the complex DNA methylation map have a consistent and / or stable methylation state in each of at least two different sample types.

[0190] Appropriately, a composite DNA methylation map can be generated by comparing a set of DNA methylation maps from at least two different sample types, and: (1) if the methylation site is matched in the DNA methylation maps from different sample types, the DNA methylation site is included in the composite DNA methylation map; and / or (2) if the DNA methylation site is not matched in the first set of DNA methylation maps from different sample types, the DNA methylation site is excluded from the composite DNA methylation map.

[0191] Appropriately, the matching DNA methylation sites that constitute a complex DNA methylation map can have substantially the same methylation state in at least two different sample types.

[0192] Methods for identifying matching DNA methylation sites across different sample types or biological replicates are known in the art. For example, matching DNA methylation sites can be determined by comparing the methylation status of methylation sites in at least two different sample types using an Epigenome Wide Association Study (EWAS) analysis.

[0193] For example, appropriate EWAS analysis can be performed using methods known in the art; such as mean absolute error (MAE) comparisons, logistic regression, linear models, or generalized linear models. Specifically, it is known in the art how to identify DNA methylation sites that do not differ between different sample types (in other words – matched, unchanging, or stable).

[0194] For example, a matched DNA methylation site can be defined as a DNA methylation site whose methylation state does not differ statistically significantly between at least two sample types. Appropriately, a matched DNA methylation site can be defined as having a mean absolute error of less than 0.05 between the two sample types. Appropriately, a matched DNA methylation site can be defined as having a p-value greater than 5%, 10%, or 20% in a linear or generalized linear model explained by the sample types.

[0195] Appropriately, a methylation site can be defined as a methylation marker present in any one or more of SEQ ID NO: 1-160. SEQ ID NO: 1-160 show sequences near methylation markers in the “CanFam3.1” canine reference genome (https: / / www.ncbi.nlm.nih.gov / datasets / genome / GCF000002285.3 / , Lindblad-Toh et al.; Nature 438, 803–819 (2005)), where the “CG” methylation marker is located at the end of the sequence (at the beginning or end of the sequence, depending on whether the site is on the positive or negative strand in the reference genome). The location of the “CG” methylation marker is provided in Table 3. In addition, the corresponding CGid is provided for each “CG” methylation marker (see Arneson et al.; Nature Communications; 13(783); 2022 and https: / / github.com / shorvath / MammalianMethylationConsortium / tree / v1.0.0).

[0196] The methylation sites defined according to this system are provided in Tables 3-6. Appropriately, the methylation sites can be defined by the CGstart and CGend columns in Table 3. For example, for DNA methylation site number 1 (SEQ ID NO: 1), the provided sequence is chr14: 41536869-41536918, and the methylation marker is chr14: 41536869-41536870.

[0197] Appropriately, a methylation site can be defined as a methylation marker present in any one or more of SEQ ID NO: 161-309. SEQ ID NO: 161-309 shows the sequence on either side of the methylation marker in the “Tasha” canine reference genome (https: / / www.ncbi.nlm.nih.gov / assembly / GCF_000002285.5; Jagannathan et al., Genes (Bsael); 2021; 12(6); 847). The “CG” methylation marker is the 26th and 27th nucleotides in the sequence (i.e., 25 nucleotides before and 25 nucleotides after the methylation marker).

[0198] The methylation sites defined according to this system are provided in Tables 7-10. These methylation sites can be defined as the middle positions in the columns marked "Sites" in Table 7. For example, for sites chr12.63269973-63269975, the methylation marker is chr12:63269974.

[0199] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may include at least one methylation site selected from sites numbered 1-138 as listed in Table 3.

[0200] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may include at least one methylation site as listed in Table 3.

[0201] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may include at least one methylation site as listed in Table 7.

[0202] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may contain at least 3, at least 5, at least 10, at least 20, at least 50, at least 100 or each of the sites numbered 1-138 as listed in Table 3.

[0203] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may contain at least 3, at least 5, at least 10, at least 20, at least 50, at least 100, or each methylation site as listed in Table 3.

[0204] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may contain at least 3, at least 5, at least 10, at least 20, at least 50, at least 100, or each methylation site as listed in Table 7.

[0205] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may include DNA methylation sites as listed in any of Tables 4-6 or Tables 8-10.

[0206] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may include DNA methylation sites as listed in Table 4.

[0207] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may include DNA methylation sites as listed in Table 5.

[0208] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may include DNA methylation sites as listed in Table 6.

[0209] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may include DNA methylation sites as listed in Table 8.

[0210] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may include DNA methylation sites as listed in Table 9.

[0211] Appropriately, DNA methylation maps, especially complex DNA methylation maps, may include DNA methylation sites as listed in Table 10.

[0212] DNA methylation sites / DNA methylation maps indicating biological age, risk of death, and / or probability of healthy lifespan The determination

[0213] This invention includes the use of DNA methylation mapping, particularly complex DNA methylation mapping as defined herein, to determine a subject's biological age, risk of death, and / or probability of healthy lifespan. Therefore, this invention includes the use of DNA methylation mapping to generate a biological clock associated with biological age, risk of death, and / or probability of healthy lifespan. The biological clock of this invention may also be referred to as an "epigenetic clock."

[0214] For example, DNA methylation sites or DNA methylation maps indicating biological age can be provided through training datasets and machine learning methods. Appropriately, the machine learning method can be a supervised machine learning method.

[0215] For example, DNA methylation sites or DNA methylation maps can be trained on a dataset containing subjects of known full age. Appropriately, DNA methylation sites or DNA methylation maps can be trained on a dataset containing subjects of known full age and known breed and / or sex.

[0216] For example, a model of DNA methylation sites or DNA methylation maps that indicate biological age can be provided by using a machine learning framework to train a dataset of methylation states at multiple DNA methylation sites on a training dataset of subjects with known full ages, and then testing the model on a retained cohort to validate the model's accuracy.

[0217] Machine learning frameworks may include fitting penalized regressions to a training dataset of subjects with known full-age (and optional breed and / or sex); for example, using the glmnet R package.

[0218] Machine learning frameworks may include fitting elastic network regressions to a training dataset of subjects with known full-age (and optional breed and / or sex); for example, using the glmnet R package.

[0219] Suitable machine learning frameworks may include penalized regressions, such as resilient network regressions, that fit the actual age as interpreted by DNA methylation profiles (and optionally breed, age, and / or sex).

[0220] When the subjects are dogs, the machine learning framework may include penalized regressions, such as elastic network regressions, that fit the true age as interpreted by DNA methylation profiles, breed, age, and sex.

[0221] Appropriately, machine learning frameworks can be used to determine models that include a set of DNA methylation sites or DNA methylation maps that indicate biological age.

[0222] The model may include the methylation state at multiple DNA methylation sites; where the methylation state at each site is considered by multiplying by a coefficient value in the model.

[0223] The coefficient value of each parameter typically depends on the unit of measurement of all variables in the model. As a technician will understand, the value of each coefficient will therefore depend on, for example, the number and nature of the different parameters used in the model, as well as the nature of the training data provided. Therefore, conventional statistical methods can be applied to the training dataset to obtain the coefficient values.

[0224] For example, DNA methylation sites or DNA methylation maps indicating mortality risk and / or the probability of healthy lifespan can be provided through training datasets and machine learning methods. Suitablely, the machine learning method can be a supervised machine learning method.

[0225] For example, DNA methylation sites or DNA methylation maps can be trained on datasets containing subjects with known mortality outcomes (survival or death) and actual age. Appropriately, DNA methylation sites or DNA methylation maps can be trained on datasets containing subjects with known mortality outcomes and actual age, as well as known breed and / or sex.

[0226] For example, a model of DNA methylation sites or DNA methylation maps that indicate the risk of death and / or the probability of healthy lifespan can be provided by using a machine learning framework to train a dataset of methylation states at multiple DNA methylation sites on a training dataset of subjects with known mortality outcomes (survival or death) and full-term age, and testing the model against a retained cohort to validate the model's accuracy.

[0227] Machine learning frameworks may include fitting a penalized model to a training dataset of subjects with known mortality outcomes (survival or death) and full-time age (and optional breed and / or sex); for example, using the glmnet R package.

[0228] Machine learning frameworks can include fitting a penalized model to a training dataset of dogs with known mortality outcomes (survival or death, age at death) and actual age (and optional breed and / or sex); for example, using the glmnet R package.

[0229] Appropriately, the penalty model can be, for example, a penalty Cox regression, a minimum angle regression path (LARS) Cox regression, or a penalty survival model.

[0230] Machine learning frameworks may include fitting penalized Cox regressions to a training dataset of subjects with known mortality outcomes (survival or death) and full-time age (and optional breed and / or sex); for example, using the glmnet R package.

[0231] Suitable, the machine learning framework may include a penalized model, preferably penalized Cox regression, that fits known mortality outcomes (survival or death) / survival rates as interpreted by DNA methylation profiles and full age (and optionally breed and / or sex).

[0232] When appropriate, when the subjects are dogs, the machine learning framework may include a penalized model that fits known mortality outcomes (survival or death) / survival rates as interpreted by DNA methylation profiles, full-time age, breed, and sex, preferably a penalized Cox regression.

[0233] As used in this article, "known mortality outcome (survival or death)" can also be referred to as "survival rate".

[0234] Appropriately, machine learning frameworks can be used to determine models that include a set of DNA methylation sites or DNA methylation maps that indicate the risk of death and / or the probability of healthy lifespan.

[0235] Appropriately, machine learning frameworks can generate predicted risks (e.g., predicted risk ratios); for example, through penalized Cox regression. This can be converted to biological / epigenetic age using methods known in the art; for example, by fitting a linear model to interpret actual age using predicted risks.

[0236] The model may include the methylation state at multiple DNA methylation sites; where the methylation state at each site is considered by multiplying by a coefficient value in the model.

[0237] Appropriately, gender can be encoded as a numerical value, where 0 represents female and 1 represents male.

[0238] Suitable varieties can be coded as numerical values, where 0 represents a small variety and 1 represents a medium variety.

[0239] The biological age of the subjects can be expressed in years, months, days, etc.

[0240] The coefficient values ​​for each parameter typically depend on the units of measurement of all variables in the model. As those skilled in the art will understand, the value of each coefficient will therefore depend on, for example, the number and nature of the different parameters used in the model and the nature of the training data provided. Therefore, conventional statistical methods can be applied to the training dataset to obtain the coefficient values. Such methods include, for example, computing two gompertz or weibull functions on the training set (where, for example, the subject's state (survival or death) is known), one function modeling survival as a function of methylation profile, age at birth, breed class (small or medium-sized subject), and sex (Model 1), and a second function considering only age at birth, breed class, and sex (Model 2). These models can be fitted in the R software environment using the flexsurv package (v 2.1).

[0241] Biological age can be defined as a time variable (“actual age”), under which the survival probability of an animal given by Model 2 is equal to the survival probability given by Model 1 at its actual age.

[0242] A model for DNA methylation sites or DNA methylation maps that indicates mortality risk and / or the probability of healthy lifespan can be provided by training a dataset of methylation states at multiple DNA methylation sites against a dataset of phenotypic ages predicted at the age of DNA sample collection, and testing the model against a retained cohort to validate the model's accuracy.

[0243] Methods for determining the phenotypic age of dogs or cats are described in PCT / EP2023 / 061058 and PCT / EP2023 / 061059, respectively. The calculation of phenotypic age takes into account direct predictive values ​​of blood biomarkers for mortality risk and / or the probability of healthy lifespan. For example, a given biomarker may not be directly related to chronological age but may indicate a specific pathological condition, thereby indicating the probability of an increased risk of death and / or a reduced healthy lifespan.

[0244] Determining a dog's phenotypic age may include determining the levels of one or more biomarkers in one or more samples obtained from the dog, wherein the one or more biomarkers are selected from white blood cell count, serum albumin, serum alkaline phosphatase, serum creatine kinase, hemoglobin, hematocrit, mean corpuscular hemoglobin, serum glucose, mean corpuscular volume, serum globulin, serum calcium, platelet count, and / or red blood cell count.

[0245] Appropriately, a dog's phenotypic age can be provided in the following ways:

[0246] a. Determine the levels of the following biomarkers: white blood cell count, serum albumin, serum alkaline phosphatase, serum creatine kinase, hemoglobin, hematocrit, mean corpuscular hemoglobin, serum glucose, mean corpuscular volume, and serum globulin in one or more samples obtained from dogs; and

[0247] b. Determine the dog's phenotypic age using formula (1):

[0248] Where xb is the sum of the values ​​of each biomarker, sex, and species multiplied by their respective coefficients according to formula (2):

[0249] Gender is encoded as a numerical value, where 0 represents female and 1 represents male. Varieties are coded as numerical values, where 0 represents a small variety and 1 represents a medium-sized variety. Furthermore, phenotypic age is used to determine the dog's risk of death and / or probability of a healthy lifespan.

[0250] The coefficient values ​​for each parameter typically depend on the units of measurement of all variables in the model. As those skilled in the art will understand, the value of each coefficient will therefore depend on, for example, the number and nature of the different parameters used in the model and the nature of the training data provided. Therefore, conventional statistical methods can be applied to the training dataset to obtain the coefficient values ​​used in the above formulas. Such methods include, for example, computing two gompertz or weibull functions on the training set (e.g., where the dog's state (survival or death) is known), one function modeling survival as a function of selected biomarkers, lifespan, breed type (small or medium-sized dog), and sex (Model 1), and a second function considering only lifespan, breed type, and sex (Model 2). These models can be fitted in the R software environment using the flexsurv package (v 2.1).

[0251] Appropriately, a negative coefficient for a given biomarker implies that a higher level of the biomarker has a positive effect on reducing the risk of death, and / or a lower level of the biomarker has a negative effect on reducing the risk of death. Similarly, a positive coefficient for a given biomarker implies that a higher level of the biomarker has a negative effect on reducing the risk of death, and / or a lower level of the biomarker has a positive effect on reducing the risk of death.

[0252] Indicative coefficients and and The values ​​are provided in the table below.

[0253]

[0254] Phenotypic age can be defined as a time variable (“actual age”), under which the survival probability of animals given by Model 2 is equal to the survival probability given by Model 1 at their actual age.

[0255] A dog's phenotype age can be expressed in years, months, days, etc.

[0256] Biomarkers used to determine phenotypic age can be determined using standard methods in the art and are typically measured as part of standard blood tests to determine disease status in animals. For example, biomarkers are often identified as part of standard clinical complete blood counts (CBCs) and standard clinical blood chemistry analyses.

[0257] Appropriately, models indicating DNA methylation sites or DNA methylation maps that indicate mortality risk and / or healthy lifespan probability trained against phenotypic age can be provided in a two-step process.

[0258] In the first step, the machine learning framework may include a penalized model that fits a phenotypic age (PhenoAge) explained by one or more blood biomarkers as described herein and full-time age (and optionally sex and / or breed); for example, using the glmnet R package. Preferably, the machine learning framework may include a penalized model that fits a phenotypic age (PhenoAge) explained by one or more blood biomarkers as described herein, full-time age, sex, and breed.

[0259] Appropriately, the penalty model can be, for example, a penalty Cox regression, a minimum angle regression path (LARS) Cox regression, or a penalty survival model.

[0260] Machine learning frameworks may include penalized Cox regressions that fit phenotypic age (PhenoAge) as explained by one or more blood biomarkers, full-time age, sex, and breed as described herein.

[0261] In the second step, the machine learning framework may include a penalized regression that fits the phenotypic age interpreted from DNA methylation. Suitablely, the machine learning framework may include a penalized regression that fits the phenotypic age interpreted from DNA methylation patterns.

[0262] Penalized regression can be elastic network regression.

[0263] As used herein, the term "one or more biomarkers" may include at least one, at least two, at least three, at least four, at least five, at least six, at least seven, at least eight, at least nine, at least ten, at least eleven, at least twelve, or at least thirteen biomarkers.

[0264] As used herein, the term "one or more biomarkers" may include one, two, three, four, five, six, seven, eight, nine, ten, eleven, twelve, or thirteen biomarkers.

[0265] Suitablely, DNA methylation sites or DNA methylation profiles can be combined with the levels of one or more blood biomarkers described herein to generate models that indicate mortality risk and / or the probability of healthy lifespan. For example, a model incorporating a combination of DNA methylation profiles and the levels of one or more blood biomarkers described herein can be provided by training a dataset of dogs with known mortality outcomes (survival or death) and full-term age on a training dataset of methylation status at multiple DNA methylation sites and levels of one or more blood biomarkers described herein, and then testing the model against a retained cohort to validate its accuracy.

[0266] Machine learning frameworks may include fitting penalized regressions to a training dataset of dogs with known mortality outcomes (survival or death) and actual age (and optional breed and / or sex); for example, using the glmnet R package.

[0267] Machine learning frameworks can include fitting a penalized model to a training dataset of dogs with known mortality outcomes (survival or death, age at death) and actual age (and optional breed and / or sex); for example, using the glmnet R package.

[0268] Machine learning frameworks may include fitting a penalized Cox regression to a training dataset of dogs with known mortality outcomes (survival or death) and actual age (and optional breed and / or sex); for example, using the glmnet R package.

[0269] Machine learning frameworks may include fitting penalized Cox regressions to a training dataset of dogs with known mortality outcomes (survival or death, age at death) and actual age (and optional breed and / or sex); for example, using the glmnet R package.

[0270] Appropriately, machine learning frameworks can generate predicted risks (e.g., predicted risk ratios); for example, through penalized Cox regression. This can be converted to biological / epigenetic age using methods known in the art; for example, by fitting a linear model to interpret actual age using predicted risks.

[0271] Appropriately, machine learning platforms may include one or more deep neural networks. A neural network is a collection of neurons (also called units) connected in a non-cyclic graph. Neural network models are typically organized into different layers of neurons. For most neural networks, the most common layer type is the fully connected layer, where neurons in two adjacent layers are fully paired, but neurons within a single layer do not share connections. One of the main characteristics of deep neural networks is that neurons are controlled by non-linear activation functions. This non-linearity, combined with the deep architecture, enables more complex combinations of input features, ultimately leading to a broader understanding of the relationships between them, and thus a more reliable final output. Deep neural networks have been applied to many types of data, ranging from structural data to chemical descriptors or transcriptomics data.

[0272] Suitablely, the machine learning platform includes one or more generative adversarial networks. Suitablely, the machine learning platform includes an adversarial autoencoder architecture. Suitablely, the machine learning platform includes feature importance analysis for ranking DNA methylation sites by their importance in biological age determination.

[0273] The biological age of the subjects can be expressed in years, months, days, etc.

[0274] Preferably, the risk of death and / or the probability of healthy lifespan are expressed as the difference between the subject's biological age and chronological age.

[0275] Compare with reference or control

[0276] This method may also include the step of comparing differences in DNA methylation at one or more sites in the test sample with one or more references or controls. The presence or absence of DNA methylation at one or more sites in the reference or control may be associated with a predefined probability of mortality risk and / or healthy lifespan (i.e., biological age). In some embodiments, the reference value is a value previously obtained for subjects or subject groups with a known probability of mortality risk and / or healthy lifespan (i.e., biological age). The reference value may be based on the known DNA methylation status at one or more sites from subject groups with a known mortality status (survival or death), full-time age, breed, and / or sex, such as mean or median levels.

[0277] Combining DNA methylation mapping with other measurements and / or features

[0278] Suitable of the invention, the method further includes combining a DNA methylation profile with one or more of the subject's actual age, breed, and / or sex. By combining this information, a biological age associated with biological age, risk of death, and / or probability of healthy lifespan can be determined.

[0279] Subject Stratification

[0280] The biological age determined by the method of this invention can also be compared with one or more predetermined thresholds (i.e., the difference from chronological age). Using such thresholds, subjects can be stratified into categories indicating a defined risk (e.g., low, medium, or high defined risk). The degree of divergence from the thresholds can be used to determine which individuals will benefit most from certain interventions. In this way, dietary interventions and lifestyle changes can be optimized.

[0281] Methods for selecting / monitoring lifestyle programs, dietary programs, or therapeutic interventions for subjects.

[0282] In another aspect, the present invention provides a method for selecting a lifestyle program, dietary program, or therapeutic intervention for a subject. Lifestyle changes can be any changes described herein, such as changes in dietary interventions and / or exercise programs. Lifestyle changes can also include the application of therapeutic methods.

[0283] Lifestyle programs, dietary protocols, or therapeutic interventions can be administered to subjects for any suitable period of time. After the stated period, the subject's biological age, risk of death, and / or probability of healthy lifespan can be reassessed using the methods of this invention to determine the efficacy of the lifestyle program, dietary protocol, or therapeutic intervention in reducing the subject's biological age and / or risk of death and / or increasing the probability of healthy lifespan. For example, the lifestyle program, dietary protocol, or therapeutic intervention can be administered for at least 2 weeks, at least 4 weeks, at least 8 weeks, at least 16 weeks, at least 32 weeks, or at least 64 weeks. The lifestyle program, dietary protocol, or therapeutic intervention can be administered for at least 3 months, at least 6 months, at least 12 months, at least 24 months, at least 36 months, at least 48 months, or at least 60 months.

[0284] Lifestyle programs, dietary programs, or therapeutic interventions can be referred to as anti-aging lifestyle programs, dietary programs, or therapeutic interventions.

[0285] Preferably, the change is a dietary intervention as described herein. The term "dietary intervention" refers to an external factor administered to a subject and causing a change in the subject's diet. More preferably, a dietary intervention includes the administration of at least a dietary product, dietary program, or nutritional supplement.

[0286] Dietary interventions can be meals, meal plans, supplements or supplement plans, or a combination of meals and supplements, or a combination of meals and multiple supplements.

[0287] Appropriately, the subjects may be dogs. In such implementations, the dietary interventions or dietary products described herein may be any suitable dietary regimen, such as calorie-restricted diets, senior dog diets, low-protein diets, phosphorus diets, low-protein diets, potassium-supplemented diets, polyunsaturated fatty acid (PUFA) supplemented diets, antioxidant supplemented diets, vitamin B supplemented diets, liquid diets, selenium-supplemented diets, omega 3-6 ratio diets, or diets supplemented with carnitine, branched-chain amino acids or derivatives, nucleotides, nicotinamide precursors such as nicotinamide mononucleotide (MNM) or nicotinamide riboside (NR), or any combination thereof.

[0288] Appropriately, dietary interventions or products may be calorie-restricted diets, senior dog diets, or low-protein diets. Appropriately, dietary interventions or products may be calorie-restricted diets. Appropriately, dietary interventions or products may be low-protein diets.

[0289] Dietary interventions can be determined based on the subject's baseline maintenance energy requirement (MER). Appropriately, MER can be the amount of food required to maintain a stable dog's body weight (a change of less than 5% over three weeks).

[0290] For example, it's generally understood that young, growing dogs benefit from a high-energy / high-protein diet; however, older dogs may have lower energy requirements and therefore their diets can be adjusted accordingly. Specifically, many manufacturers produce "senior dog" lines of dog food that are lower in calories and higher in fiber, but have protein and fat levels suitable for older dogs.

[0291] Appropriately, a calorie-restricted diet may contain approximately 50%, 55%, 60%, 65%, 75%, 80%, 85%, or 90% of a dog's MER. Appropriately, a calorie-restricted diet may contain approximately 60% or 75% of a dog's MER.

[0292] Appropriately, a low-protein diet can contain less than 20% protein (% dry matter). For example, a low-protein diet can contain less than 19% protein (% dry matter).

[0293] These diets are typically recommended based on the dog's actual age. For example, it might be recommended to switch a dog to a senior dog diet around 7 or 8 years of age. However, in the context of this invention, determining that a dog has an increased risk of death compared to what is expected based on its actual age can allow for the determination of a switch to a senior dog diet at an earlier age. In contrast, dogs with a reduced risk of death compared to their actual age may be able to maintain an adult dog diet for a longer period.

[0294] Dietary interventions may include foods, supplements, and / or beverages that contain nutrients and / or bioactive substances that mimic calorie restriction (CR) benefits without restricting daily calorie intake. For example, foods, supplements, and / or beverages may contain functional ingredients with CR-like benefits. Appropriately, foods, supplements, and / or beverages may contain autophagy inducers. Appropriately, foods, supplements, and / or beverages may contain fruits and / or nuts (or extracts thereof). Suitable examples include, but are not limited to, pomegranate, strawberry, blackberry, camu camu, walnut, chestnut, pistachio, and pecan. Appropriately, foods, supplements, and / or beverages may contain probiotics, with or without fruit or nut extracts.

[0295] Modifying a subject's lifestyle also includes instructing them to make lifestyle changes, such as prescribing more physical activity. Similar to dietary interventions, identifying an increased risk of death in a dog compared to what is expected based on its age can allow for determining an appropriate exercise program to be implemented.

[0296] Modifying a subject's lifestyle also includes selecting or recommending treatment methods or protocols. These treatment methods or protocols can be used to treat and / or prevent conditions such as arthritis, dental diseases, endocrine disorders, heart disease, diabetes, liver disease, kidney disease, prostate disorders, cancer, and behavioral or cognitive impairments. Where appropriate, preventative therapies may be administered to subjects identified as being at risk for such conditions due to an increased risk of death and / or based on specific biomarkers known to be associated with disease-related pathways. In other implementations, subjects identified as being at risk for certain conditions (due to an increased risk of death and / or based on specific biomarkers known to be associated with disease-related pathways) may be monitored more regularly so that diagnosis and treatment can be initiated earlier.

[0297] This invention also relates to monitoring and / or determining the efficacy of anti-aging therapies or developing anti-aging therapies. Anti-aging therapies may include, for example, "rejuvenation" interventions. Rejuvenation interventions aim to induce a reduction in the epigenetic or biological age of a subject. Appropriately, a rejuvenation intervention may reprogram the epigenetic age to the age of a very young subject. Examples of such rejuvenation interventions include, but are not limited to, gene therapies that reprogram the epigenetic age to the age of a very young subject. The methods of this invention for monitoring and / or determining the efficacy of lifestyle programs, dietary programs, or therapeutic interventions, or for developing lifestyle programs, dietary programs, or therapeutic interventions for reducing biological age, are particularly suited to this aspect.

[0298] Therefore, the present invention can advantageously enable the identification of subjects who are expected to respond particularly well to a given intervention (e.g., a lifestyle program, dietary program, or therapeutic intervention). Thus, the intervention can be applied in a more targeted manner to subjects who are expected to respond.

[0299] In one aspect, the present invention provides a method for determining the efficacy of a lifestyle program, dietary program, or therapeutic intervention in reducing the biological age and / or risk of death and / or increasing the probability of a healthy lifespan of a subject, the method comprising: a) administering the lifestyle program, dietary program, or therapeutic intervention to the subject, wherein the lifestyle program, dietary program, or therapeutic intervention has been selected according to the method of the present invention; b) after administering the lifestyle program, dietary program, or therapeutic intervention to the subject for a period of time; determining the subject's biological age, risk of death, and / or probability of healthy lifespan using a composite DNA methylation profile from a sample obtained from the subject, wherein the composite DNA methylation profile has been generated according to the method of the present invention, or is a composite DNA methylation profile as further defined herein; c) determining whether the subject's risk of death has changed after the period of adherence to the lifestyle program, dietary program, or therapeutic intervention.

[0300] In another aspect, the present invention provides a method for determining the efficacy of a lifestyle program, dietary program, or therapeutic intervention in reducing the biological age and / or risk of death and / or increasing the probability of a healthy lifespan of a subject, the method comprising: a) determining the subject's biological age, risk of death, and / or probability of healthy lifespan using a DNA methylation profile from a sample obtained from the subject, wherein a composite DNA methylation profile has been generated according to the present invention, or is a composite DNA methylation profile as further defined herein; b) administering to the subject a lifestyle program, dietary program, or therapeutic intervention selected based on the biological age, risk of death, and / or probability of healthy lifespan determined in step a); c) after a period of time following the administration of the lifestyle program, dietary program, or therapeutic intervention to the subject; determining the subject's biological age, risk of death, and / or probability of healthy lifespan using a DNA methylation profile from a sample obtained from the subject, wherein a composite DNA methylation profile has been generated according to the present invention, or is a composite DNA methylation profile as further defined herein; d) determining whether the subject's risk of death and / or probability of healthy lifespan has changed between steps a) and c).

[0301] Appropriately, a lifestyle program, dietary program, or therapeutic intervention may have been administered to the subject for some time prior to the determination of the first biological age, risk of death, and / or probability of healthy life; however, the effectiveness of the lifestyle program, dietary program, or therapeutic intervention in improving the subject's biological age, risk of death, and / or probability of healthy life (i.e., reducing the risk of death and / or increasing the probability of healthy life) can still be monitored by determining the biological age, risk of death, and / or probability of healthy life at two or more time points during the administration of the lifestyle program, dietary program, or therapeutic intervention.

[0302] Suitablely, the method of the present invention may include an "ecosystem"; particularly a digital ecosystem. Suitablely, the method of the present invention may include providing a sample obtained from a subject, optionally using a kit according to the present invention; and (b) providing the sample (e.g., by mail) for subsequent DNA extraction to measure DNA methylation in the extracted DNA from the sample, thereby obtaining a DNA methylation profile.

[0303] DNA methylation maps can then be used by any method according to the invention; preferably, a computer system or computer program product according to the invention is used.

[0304] The computer system or computer program can then prepare and share a report detailing the results of the analysis / method, which may include, for example, selecting or recommending suitable lifestyle programs, dietary programs or therapeutic interventions for the subjects, or any other result of the method of the present invention.

[0305] Appropriately, the sample can be one that can be obtained at home (e.g., by the dog owner, or without the need for a veterinarian or healthcare professional). Appropriately, the sample can be a hair follicle, oral swab, or saliva sample.

[0306] Uses of dietary intervention

[0307] In one aspect, the present invention provides a dietary intervention for reducing the biological age and / or risk of death of a subject and / or increasing the probability of a healthy lifespan of a subject, wherein the dietary intervention is applied to a subject whose biological age, risk of death, and / or probability of a healthy lifespan have been determined by the method of the present invention.

[0308] In another aspect, the present invention provides the use of dietary interventions in reducing the biological age and / or risk of death of subjects and / or increasing the probability of healthy lifespan of subjects, wherein the dietary intervention is applied to subjects whose biological age, risk of death, and / or probability of healthy lifespan have been determined by the method of the present invention.

[0309] As described in this article, dietary interventions can be dietary products, dietary plans, or nutritional supplements.

[0310] Computer program products

[0311] The method of the present invention can be executed using a computer. Therefore, the method of the present invention can be executed in a computer.

[0312] Appropriately, the computer can prepare and share a report detailing the results of the method of the present invention.

[0313] The methods described herein can be implemented as computer programs running on general-purpose hardware, such as one or more computer processors. In some implementations, the functionality described herein can be implemented by devices such as smartphones, tablet terminals, or personal computers.

[0314] In one aspect, the present invention provides a computer program product comprising computer-executable instructions for causing a programmable computer to determine the biological age, risk of death, and / or probability of healthy lifespan of a subject as described herein.

[0315] In one implementation, the user optionally inputs the levels of one or more DNA methylation markers as defined herein, along with actual age, breed, and sex, into the device. The device then processes this information and provides a determination of the subject's biological age. Alternatively, the device then processes this information and, based on the biological age, determines a suitable lifestyle plan, dietary plan, or therapeutic intervention for the subject.

[0316] The device can typically be a server on a network. However, any device can be used as long as it can process biomarker data and / or additional parameter or characteristic data using a processor, central processing unit (CPU), etc. The device can be, for example, a smartphone, tablet terminal, or personal computer, and outputs information indicating the subject's determined biological age or determining a suitable lifestyle plan, dietary plan, or treatment intervention based on the subject's biological age.

[0317] Those skilled in the art will understand that they are free to combine all the features of the invention described herein without departing from the scope of the invention disclosed herein.

[0318] Example

[0319] The present invention will now be further described by way of examples, which are intended to help those skilled in the art to implement the invention, without limiting the scope of the invention in any way.

[0320] Example 1 - Multi-tissue biological clock using DNA methylation array

[0321] Whole blood samples from a canine cohort (26 dogs), including data from blood and oral swab samples, were analyzed as follows: DNA extraction, DNA methylation conversion using bisulfite conversion, and amplification of the converted DNA. The DNA was then hybridized to a mammalian methylation array (Illumina) and labeled with a fluorescent dye. After hybridization, the array was washed and scanned using the iScan microarray scanner. The raw data were read and normalized using the sesame R package (Zhou W, Triche TJ, Laird PW, Shen H (2018). “SeSAMe: ​​reducing artifactual detection of DNA methylation by Infinium BeadChips in genomic deletions.” Nucleic AcidsResearch, gky691. doi:10.1093 / nar / gky691).

[0322] Several steps were taken to process the array data: Outliers in inter-array correlations were removed. Samples that incorrectly predicted species were excluded from the dataset.

[0323] Misclassified samples and technically duplicated samples were also removed to maintain data accuracy.

[0324] The selection of non-changing probes between blood and oral swabs is as follows: Probes with a detection p-value greater than 0.05 in 10% of the samples were removed. This filtering process aims to eliminate probes with low reliability.

[0325] Probes with a mean absolute error (MAE) of <0.05 (swab, blood) were selected as stable probes between different tissues.

[0326] Using the probes selected above, a resilient network regression model was trained on phenotypic age (see Examples 3 and 4) based on methylation profiles from blood samples (from a larger dataset of over 850 dogs).

[0327] A total of 160 probes were selected in the final model (see Table 3).

[0328] Figure 3 The correlation between the "multi-tissue" phenotypic age of blood and oral swabs and actual age was shown.

[0329] Figure 4 The correlation between complex DNA methylation profiles between blood and oral swab samples was shown.

[0330] Figure 5 A validation study of the "multi-tissue" phenotypic clock from blood and oral swabs is shown, using data from a lifetime calorie restriction study. Figure 4 The results showed that the biological age of the calorie restriction group (R) was lower than that of the control group (C).

[0331] Additional biological clocks were generated using only the first 5, 10, 30, and 50 sites from the complete site list shown in Table 3; and each showed a correlation with biological age (see Table 3). Figure 6 These clocks were generated by selecting the first n sites based on the absolute values ​​of the coefficients from the full clock (in descending order, taking the largest coefficient first). The first n sites were used as predictors to fit linear models explaining the actual age. Details of the clocks for the first 5, 10, 30, and 50 sites are shown in Tables 4–7. Phenotypic age (phenoDNA mAge) was calculated using a linear combination of coefficients (phenoDNA mAge = intercept + coefficient). methylation level).

[0332] Figure 7 The correlation of complex DNA methylation profiles between blood and oral swab samples is shown for the first 5, first 10, first 30, and first 50 sites.

[0333] Example 2 – Multi-tissue biological clocks using EMseq data

[0334] A dataset of 26 dogs was processed, with data from three different tissues (blood, saliva, and oral swabs). The process involved DNA extraction, DNA methylation via enzymatic conversion, whole-genome library preparation, hybridization of the transformed library with capture probes targeting gene promoters, and measurement of methylation patterns via sequencing (EMSeq).

[0335] The capture probe targets approximately 40,000 targets (promoter regions – approximately 1 kb upstream and 0.5 kb downstream of the promoter). These target regions contain potential methylation sites of interest (single cytosine residues that can be methylated).

[0336] The following bioinformatics steps were performed after sequencing and before further analysis: Using FastQC for FastQ quality checks - https: / / www.bioinformatics.babraham.ac.uk / projects / fastqc / Using trimGalore for weaver trimming - https: / / www.bioinformatics.babraham.ac.uk / projects / trim_galore / Alignment to the canine genome using Bismark - https: / / www.bioinformatics.babraham.ac.uk / projects / bismark / Use Picard to mark duplicates - https: / / gatk.broadinstitute.org / hc / en-us / articles / 360037052812 MarkDuplicates-Picard- Using Methyldackel to invoke methylation - https: / / github.com / dpryan79 / MethylDackel First use Boostme The algorithm imputes low coverage (<15) and missing values ​​(Zou, LS, Erdos, MR, Taylor, D. et al. BoostMe accurately predicts DNA methylation values ​​in whole-genome bisulfite sequencing of multiple human tissues). BMC Genomics19, 390 (2018). https: / / doi.org / 10.1186 / s12864-018-4766-y ) This is a tree-based machine learning algorithm, tailored to each sample type.

[0337] The X chromosome was removed.

[0338] Then, methylation sites that did not change across tissues were selected. To achieve this, methylation values ​​were converted from proportions to M values ​​(M = log2(β / (1-β))) and limma analysis was performed, where for each site, the M value was interpreted by conditions (blood, saliva, swab), breed category, sex, and pairing information. Sites with adjusted p-values ​​greater than 5% were selected using a modified F-test. The p-values ​​were adjusted using Benjamini-Hochberg multiple test correction.

[0339] A total of 128,512 sites were selected for further analysis (complex DNA methylation maps).

[0340] Another, larger dataset containing over 750 dogs was processed using the above methods (EMseq, bioinformatics, and boostMe).

[0341] A resilient network regression model was trained on phenotypic age values ​​(see Examples 3 and 4) using a composite DNA methylation profile. The Lambda parameter was selected using 10-fold cross-validation. The model selected 149 DNA methylation sites (see Table 8).

[0342] Figure 8 The correlation between "multi-tissue" phenotypic age and chronological age was shown in blood, saliva, and oral swabs.

[0343] Figure 9 The correlation of complex DNA methylation profiles between blood and oral swab samples (Figure A) and between blood and saliva samples (Figure B) is shown.

[0344] Figure 10 A validation study of the "multi-tissue" phenotypic clock from blood, saliva, and oral swabs was presented, using data from a lifetime calorie restriction study. Figure 10 The results showed that the biological age of the calorie restriction group (R) was lower than that of the control group (C).

[0345] Additional biological clocks were generated using only the first 5, 10, 30, and 50 sites from the complete site list shown in Table 3; and each showed a correlation with biological age (see Table 3). Figure 11These clocks were generated by selecting the first n sites based on the absolute values ​​of the coefficients from the full clock (in descending order, taking the largest coefficient first). The first n sites were used as predictors to fit linear models that account for the actual age. Details of the clocks for the first 5, 10, 30, and 50 sites are shown in Tables 9-11. Phenotypic age (phenoDNA mAge) was calculated using a linear combination of coefficients (phenoDNA mAge = intercept + coefficient). methylation level).

[0346] Reference Example 3 – Determination of Blood Biomarkers Associated with Canine Mortality Risk

[0347] This embodiment is provided for reference on how to generate phenotypic ages, which are used to generate the biological clocks in Examples 1 and 2.

[0348] Predictive blood biomarkers were identified from a panel of biomarkers consisting of standard clinical complete blood count (CBC) and standard clinical blood chemistry analysis. Serum samples were collected after overnight fasting and measured using standard veterinary clinical practice.

[0349] Table 1 - Clinical complete blood count (CBC) and clinical blood chemistry analysis The numerical values ​​are logarithmically transformed using the natural logarithm.

[0350] We used a longitudinal study of dogs in which we repeatedly measured these parameters, along with information about the dogs' status (survival or death), sex, and breed. We first categorized the breed as small or medium based on the mean weight of adult dogs of that breed (less than 10 kg or more than 10 kg, respectively). We then organized the data using the R programming language. For each dog, we recorded the biomarkers as time-dependent covariates using left-open, right-closed time intervals (i.e., (tstart, tstop]), where the biomarker information corresponds to the start of the interval and the event (survival or death) is recorded as the last tstop value. For this, we used the tmerge function of the survival package in R (v. 3.2-13). We then fitted a Cox proportional hazards model to the data for each of the 28 biomarkers, including sex and breed category (small or medium). We then adjusted the p-value for each parameter to account for multiple comparisons (by false discovery rate (fdr)) and selected features with an adjusted fdr below 0.05. Figure 1 ).

[0351] Using this method, we identified 13 biomarkers that can individually predict the survival probability of dogs: White blood cell count (10^3 / ul) Serum albumin (g / dL) Serum alkaline phosphatase (U / L, ln conversion) Serum creatine kinase (IU / L, ln conversion) Hemoglobin (g / dL) Hematocrit (%) Mean corpuscular hemoglobin (pg) Serum sodium (mmol / L) Mean corpuscular volume (fL) Serum globulin (g / dL) Serum calcium (mg / dL) Serum platelet count (10^3 / uL) Red blood cell count (10^3 / uL)

[0352] Refer to Example 4 – A Multi-parameter Model for Predicting Mortality Risk

[0353] This embodiment is provided for reference on how to generate phenotypic ages, which are used to generate the biological clocks in Examples 1 and 2.

[0354] We constructed an optimal model that considers multiple parameters simultaneously, as this is more likely to cover a wide range of organ dysfunctions that occur with age. However, selecting only a few potentially correlated features can lead to bias. To avoid this problem, we used a penalized regression method using the glmnet package (v4.1-3). We fitted a LASSO penalized Cox proportional hazards model to the data and used 20-fold cross-validation to compare different values ​​of the penalized parameter λ. This method selected the 10 blood biomarkers that best predict survival, ranked by importance as follows: White blood cell count (10^3 / ul) Serum albumin (g / dL) Serum alkaline phosphatase (U / L, ln conversion) Serum creatine kinase (IU / L, ln conversion) Hemoglobin (g / dL) Hematocrit (%) Mean corpuscular hemoglobin (pg) Serum glucose (mg / dL) Mean corpuscular volume (fL) Serum globulin (g / dL) We also found that the top 3 biomarkers on the list are the most predictive, and performance can be improved by combining each of the next 7 biomarkers.

[0355] To extract the phenotypic age of animals, we computed two distinct gompertz functions on the training set: one modeling survival as a function of selected biomarkers, age, breed type (small or medium-sized dog), and sex (Model 1), and the second considering only age, breed type, and sex (Model 2). These models were fitted using the flexsurv package (v 2.1). Phenotypic age was defined as a time variable (“age”), at which Model 2 gives the animal’s survival probability equal to Model 1’s survival probability at its chronological age. This results in a mathematical function relating blood biomarkers to phenotypic age, and is given by the following formula:

[0356] Where xb is the sum of the values ​​for each biomarker, sex, and breed multiplied by their corresponding coefficients. Sex and breed are encoded as numerical values, where 0 represents female and 1 represents male for sex, and 0 represents small breed and 1 represents medium breed for breed. The coefficients are given by two gompertz functions trained on our training set.

[0357] As an example, coefficients and and The values ​​have been measured against a complete list of biomarkers from our training set and are given in Table 2.

[0358]

[0359] Table 2 – Coefficients and Sum and The values ​​have been measured from the training set.

[0360] Furthermore, by systematically removing one biomarker from the top of the list to reduce the set of 10 biomarkers, we observed a decrease in the strength of survival predictions (p-value). The decrease in the first parameter was the most pronounced, confirming their largest contribution, but we observed changes in prediction quality with each set reduction, indicating that each parameter contributes to the overall prediction. Figure 2 ).

[0361] All publications mentioned in the foregoing description are incorporated herein by reference. Various modifications and variations of the methods, compositions, and uses disclosed herein will be apparent to those skilled in the art without departing from the scope and spirit of the invention. While the invention has been disclosed in conjunction with specific preferred embodiments, it should be understood that the invention protected by the claims should not be unduly limited to such specific embodiments. In fact, various modifications to the modes disclosed for practicing the invention that are apparent to those skilled in the art are intended to fall within the scope of the following claims.

[0362]

Claims

1. A method for generating a biological clock containing a DNA methylation map, the DNA methylation map being applicable to at least two different sample types, the method comprising: (i) Provide a first set of DNA methylation profiles generated from the at least two different sample types from multiple subjects; (ii) Generate a composite DNA methylation map from the first set of DNA methylation maps, wherein the composite DNA methylation map contains methylation sites that are in a matching state in the at least two different sample types; (iii) Using a reference DNA methylation map from at least one of the at least two sample types, a biological clock is generated using the composite DNA methylation map.

2. The method of claim 1, wherein step (ii) comprises comparing the first set of DNA methylation maps, and: (1) If the methylation site has a matching state in the DNA methylation profiles from the at least two different sample types, then the methylation site is included in the composite DNA methylation profile; and / or (2) If the methylation site does not have a matching state in the DNA methylation map from the at least two different sample types, the methylation site is excluded from the composite DNA methylation map.

3. The method according to claim 1 or 2, wherein the matched DNA methylation sites have substantially the same methylation state in the at least two different sample types.

4. The method according to any of the preceding claims, wherein step (ii) is performed using an Epigenome Widely Associated Study (EWAS) analysis, appropriately by means of mean absolute error (MAE) comparison, logistic regression, linear model or generalized linear model.

5. The method according to any of the preceding claims, wherein the subject is a mammal.

6. The method of claim 5, wherein the subject is a dog, cat, or human; preferably, wherein the subject is a dog.

7. The method according to any of the preceding claims, wherein the first set of DNA methylation patterns is derived from at least three, at least five, or at least ten different sample types.

8. The method according to any of the preceding claims, wherein the at least two different sample types are independently selected from blood, oral swabs, saliva, feces, hair, skin, and organ tissue samples.

9. The method of claim 8, wherein the at least two different sample types include (A) blood, oral swab, saliva, or (B) a blood and oral swab sample.

10. The method according to any of the preceding claims, wherein: (A) Step (iii) is performed using a DNA methylation profile from a single sample type from the at least two different sample types; and / or (B) the sample type used in step (iii) is a blood sample.

11. The method according to any of the preceding claims, wherein the biological clock is adapted to determine the subject's biological age, risk of death, and / or probability of healthy lifespan.

12. The method according to any of the preceding claims, wherein step (iii) comprises generating the biological clock using supervised machine learning; and appropriately using a penalty model.

13. The method according to any of the preceding claims, wherein the biological clock is adapted to determine the subject’s risk of death and / or probability of healthy lifespan; optionally, wherein step (iii) further comprises combining the DNA methylation profile with one or more of the dog’s actual age, breed and / or sex.

14. The method according to any preceding claim, further comprising: (iv) Provide DNA methylation maps from test samples obtained from test subjects; and v) using the biological clock generated from the complex DNA methylation map according to steps (i)-(iii) to determine the subject’s biological age, risk of death, and / or probability of healthy lifespan.

15. A method for determining a subject's biological age, risk of death, and / or probability of healthy life expectancy, said method comprising: a) Provide DNA methylation profiles from test samples obtained from the subject; as well as b) Using a biological clock generated from a composite DNA methylation map produced by any one of claims 1 to 13, the subject's biological age, risk of death, and / or probability of healthy lifespan is determined.

16. A method for selecting a lifestyle program, dietary program, or therapeutic intervention for a subject, said method comprising: a) Provide DNA methylation profiles from test samples obtained from the subject; b) Use the composite DNA methylation profile generated by the method according to any one of claims 1 to 13 to determine the subject's biological age, risk of death, and / or probability of healthy lifespan; as well as c) Based on the biological age, risk of death, and / or probability of healthy life determined in step b), select appropriate lifestyle programs, dietary programs, or therapeutic interventions for the subject.

17. The method of claim 15 or claim 16, wherein the subject is a mammal.

18. The method of claim 17, wherein the mammal is a dog, a cat, or a human; preferably, wherein the mammal is a dog.

19. The method according to any one of claims 16 to 18, wherein the lifestyle program, dietary program, or therapeutic intervention is selected based on determining that the dog has an increased biological age, an increased risk of death, and / or a reduced probability of healthy lifespan compared to its actual age.

20. The method according to any one of claims 16 to 19, wherein the lifestyle program, dietary program or therapeutic intervention is a dietary intervention, preferably a calorie-restricted diet, an old dog diet or a low-protein diet.

Images