Method of selection and stratification
By analyzing specific microbiome features and calculating a score based on their correlation with methane emissions, the method addresses the challenge of determining genomic factors for methane emissions in ruminants, enabling effective classification and breeding strategies to reduce methane emissions.
Patent Information
- Application Number
- PCT/AU2025/050517
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2025-05-16
- Publication Date
- 2026-11-19
AI Technical Summary
Current methods for determining the factors influencing methane emissions in ruminants are challenging and costly, and different recording methods show weak correlations, complicating the identification of genomic features for breeding programs aimed at reducing methane emissions.
A method to determine the likelihood of a ruminant being a high or low methane emitter by analyzing the abundance of specific taxonomic or functional features in the microbiome, such as Methanobrevibacter, Entodinium, and KEGG Orthology groups, and calculating a score based on their correlation with methane emissions, enabling stratification and selection for breeding programs.
Enables accurate classification of ruminants into high or low methane emitters, facilitating targeted breeding programs to reduce methane emissions effectively and efficiently.
Smart Images

Figure IMGF000024_0001 
Figure IMGF000029_0001 
Figure IMGF000031_0001
Abstract
Description
Method of selection and stratificationField of the invention
[0001] The present disclosure relates to method for stratification of methane-emitting animals into high- and low- emitting populations and biomarkers associated with low methane emissions.Background of the invention
[0002] Ruminants have evolved in symbiosis with their rumen microbiota for over 50 million years. For this reason, they can transform plant materials that humans cannot digest into vital nutrients and energy. This capability depends on a diverse microbial community that, as a consequence, produces various greenhouse gases.
[0003] Livestock production is responsible for environmental burdens participating in around 15 - 18 % of global emissions of greenhouse gases (GHGs), especially methane (CH4) and nitrous oxide (N2O). Amongst these, methane (CH4) is the second most important greenhouse gas - having a global warming potential of 25 to 28 times to that of carbon dioxide (CO2).
[0004] Enteric methane emissions (EMEs) in ruminant animals, wherein the methane is produced via fermentation of feeds in the rumen and the lower digestive track by methanogenic archaea, represents a loss of 2 % to 12 % of the gross energy of feeds and contributes to GHG-effects. Globally, about 80 million tons of methane is produced annually from enteric fermentation mainly from ruminant animals.
[0005] It has been reported that EME is influenced by a variety of factors, including dietary factors and host factors such as genotype and physiology and rumen microbiome. However, elucidating the specific factors directly influencing EME has been complicated by a variety of factors. For example, directly measuring EME on commercial farms is currently challenging and cost-prohibitive. Additionally, different EME recording methods could show weak correlations, complicating the combination of reference populations.1005811148
[0006] Consequently, there is a need for improved methods and approaches for determining the genomic features or factors that contribute to methane emissions by ruminants. The identification of these factors may be useful for implementation in breeding programs.
[0007] Reference to any prior art in the specification is not an acknowledgment or suggestion that this prior art forms part of the common general knowledge in any jurisdiction or that this prior art could reasonably be expected to be understood, regarded as relevant, and / or combined with other pieces of prior art by a skilled person in the art.Summary of the invention
[0008] The invention seeks to address one or more of the aforementioned problems or limitations.
[0009] In one aspect, there is provided a method of determining the likelihood that a ruminant is low or high methane emitting, comprising:- determining the abundance of taxonomic or functional features in a microbiome sample of the ruminant, wherein the features are selected from:a) Methanobrevibacter, K02416, COG1100b) Methanobrevibacter, Entodinium, K02416, COG1151, COG0417, COG0667, CGG0076;c) Methanobrevibacter, K00937, K01681; ord) Archaea, Bacteria and / or Eukaryota at the superkingdom taxonomic level; or a combination thereof,- calculating a score that ranks the ruminant as being a high or low methane emitter, based on the abundance of the taxonomic or functional features relative to the mean abundance of the features in a reference population and wherein the features Methanobrevibacter, Entodinium, COG0417, COG0667, COG0076, COG1100, Archaea and Eukaryota are positively correlated with EME and the1005811148features K02416, C0G1151, K00937, K01681 and Bacteria, are negatively correlated with EME,thereby determining the likelihood that a ruminant is low or high methane emitting.
[0010] In any embodiment, the abundance may be an absolute abundance, which is the number of microbial sequences detected per sample from the ruminant, assigned to operational taxonomy units (OTUs) at the levels of superkingdom, phylum, class, order, family, genus, and species or to functional annotations such as KEGG Orthology (KO) or Clusters of Orthologous Groups (COG).
[0011] In any embodiment, the abundance may be a relative abundance, which is the absolute abundance of a taxonomic or functional feature expressed as a proportion of the total abundance of features of the same type within the sample of the ruminant. For example, the relative abundance of Methanobrevibacter is calculated relative to the total abundance of all genera in the sample. Similarly, the relative abundance of K02416 is calculated relative to the total abundance of all KEGG Orthology (KO) features, and the relative abundance of COG1151 is calculated relative to the total abundance of all Clusters of Orthologous Groups (COG) features in the sample.
[0012] In a preferred embodiment, the method further comprises estimating the genetic correlation between the features and EME in the reference population prior to obtaining the score.
[0013] It will be appreciated that score and the correlation between the features and EME in the reference population are influenced by the negative or positive correlation of the feature with EME. Accordingly, the skilled person will be able to derive a score based on the understanding as described elsewhere herein, that the features Methanobrevibacter, Entodinium, COG0417, COG0667, COG0076, CGG1100, Archaea and Eukaryota are positively correlated with EME (ie the higher the abundance of these features, the greater the EME is likely to be). In contrast, the features K02416, COG1151, K00937, K01681 and Bacteria, are negatively correlated with EME (ie the lower the abundance of these features, the lower the EME is likely to be).
[0014] Typically, the reference population is a group of ruminants used to estimate the effects of host’s genetic markers on EME and the rumen metagenome feature traits. In1005811148any embodiment, the reference population is the same population of which the ruminant is member. The reference population may be a population of ruminants of the same species, located in a different geographical location. Preferably in those circumstances, the ruminant population has similar environmental conditions as the ruminant for whom the level of EME is being determined. An example of a similar environmental condition includes: similar feed, similar climate conditions, and similar population density.
[0015] In certain embodiments, a combination of two or more groups or subpopulations of ruminants may be used to generate the reference population to estimate the effects of host’s genetic markers on EME and the rumen metagenome feature traits. The combination of two or more group / subpopulations may be from the same, similar or from different geographic or environmental conditions.
[0016] In preferred embodiments, the method further comprises selecting the ruminant for inclusion in a breeding program when the score obtained indicates that the ruminant is a low methane emitter.
[0017] In a further aspect, there is provided a method of stratifying a population of ruminants into low or high methane emitters, comprising:- obtaining or having obtained nucleic acid sequences from microbiome samples of the individuals in the population,- determining the abundance of taxonomic or functional features in the samples, wherein the features are selected from:a) Methanobrevibacter, K02416, COG1100b) Methanobrevibacter, Entodinium, K02416, COG1151, COG0417, COG0667, CGG0076;c) Methanobrevibacter, K00937, K01681; ord) Archaea, Bacteria and / or Eukaryota at the superkingdom taxonomic level; or a combination thereof,1005811148- calculating a score that ranks each ruminant in the population as being a high or low methane emitter, based on the abundance of the taxonomic or functional features relative to the mean abundance of the taxonomic or functional features in the population, and wherein the features Methanobrevibacter, Entodinium, COG0417, COG0667, COG0076, CGG1100, Archaea and Eukaryota are positively correlated with EME and the features K02416, COG1151, K00937, K01681 and Bacteria, are negatively correlated with EME,thereby stratifying a population of ruminants into low or high methane emitters.
[0018] The method may further comprise selecting individuals in the population for inclusion in a breeding program when the score obtained indicates that the individuals are low methane emitters.
[0019] In any embodiment, the abundance may be an absolute abundance, which is the number of microbial sequences detected per sample from the ruminant, assigned to operational taxonomy units (OTUs) at the levels of superkingdom, phylum, class, order, family, genus, and species or to functional annotations such as KEGG Orthology (KO) or Clusters of Orthologous Groups (COG).
[0020] In any embodiment, the abundance may be a relative abundance, which is the absolute abundance of a taxonomic or functional feature expressed as a proportion of the total abundance of features of the same type within the sample of the ruminant. For example, the relative abundance of Methanobrevibacter is calculated relative to the total abundance of all genera in the sample. Similarly, the relative abundance of K02416 is calculated relative to the total abundance of all KEGG Orthology (KO) features, and the relative abundance of COG1151 is calculated relative to the total abundance of all Clusters of Orthologous Groups (COG) features in the sample.
[0021] In a preferred embodiment, the method further comprises estimating the genetic correlation between the features and EME in the population prior to obtaining the score.
[0022] In a further aspect, there is provided a method for identifying features capable of distinguishing a low methane emitting ruminant from a high methane emitting ruminant, the method comprising:1005811148a. identifying an initial set of features in nucleic acid sequences from a microbiome sample of a ruminant, wherein the initial set of features comprise operational taxonomic units (OTUs), KEGG Orthology groups (KO) features and Clusters of Orthologous Groups (COG) features; b. pre-treating the initial set of features, comprising:i. filtering the initial set of features;ii. imputing zero values in the filtered features to non-zero values, thereby generating core features;iii. constructing matrices from the relative abundance of the core features and methane emission traits of the animal; andiv. transforming the matrices of iii.;c. optionally performing principal component analysis of the transformed matrices, thereby identifying principal components; andd. estimating heritability of the core features, genetic and phenotypic correlation between the core features and the methane emission traits, thereby determining the features, wherein the features comprise one or more of OTUs, KEGG features, COG features and PCs.
[0023] In any embodiment, in step a. the nucleic acid sequences are taxonomically assigned at the taxonomic rank and functionally annotated as KOs and COG identifiers. This produces matrices of dimensions animals times rumen features (OTUs, genera, KOs or COGs) populated with the number of reads assigned to each feature. The skilled person will appreciate that these matrices are referred to as absolute abundance matrices.
[0024] As used herein, operational taxonomic units (OTUs) refers to superkingdoms and genera (for example, taxonomic superkingdom features, taxonomic genera features).1005811148
[0025] In any embodiment, the step of imputing zero values (when pre-treating the initial set of features at step b.) refers to assigning a non-zero value to zeros recorded in the filtered absolute abundance matrix
[0026] In any embodiment, the step of filtering the initial set of features comprises 1. removing features of non-microbial clades (e.g., Animalia and Plantae) 2. removing unclassified operational taxonomic units at the family level3. removing features that include genes from the ruminant (such as Bos taurus where the ruminant is a cow)4. removing features that are not present in at least a specified percentage of animals are removed.
[0027] In any embodiment, step 4 of the filtering comprises removal of microbial features that did not appear in at least about 95% of animals, more preferably in at least 90% of animals.
[0028] In any embodiment, the step of constructing the relative abundance matrixes comprises obtaining the relative abundance of each feature with respect to other features of the same type within each sample - in other words, transforming the absolute abundance matrices that have those features retained after filters into relative abundance (RA) matrices for each type of feature.
[0029] In any embodiment, the step of transforming the abundance matrixes comprises obtaining a log-ratio transformation (such as centered log-ratio (CLR) transformation) that is applied to the relative abundances of genus-level taxa, KOs, and COGs. The skilled person will appreciate that this transformation produces unitless values that represent the proportion of each feature relative to others of the same type (i.e. , genus to genus, KO to KO, and COG to COG) within the same sample after the mentioned transformations. These quantities are then evaluated as genetic predictors of enteric methane emissions (EME). Exemplary methods for calculating log-ratio transformations (eg CLR) are further provided herein in the examples.1005811148
[0030] In any embodiment, the heritability of the features capable of distinguishing a low methane emitting ruminant from a high methane emitting ruminant is equal to or more than: about 0.05, about 0.1, about 0.15, about 0.2, about 0.25, about 0.3, about 0.35, or about 0.4, or about 0.5 or about 0.6 or about 0.7. Preferably the heritability of the features capable of distinguishing a low methane emitting ruminant from a high methane emitting ruminant is equal to or more than: 0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.35, or 0.4. More prefereably the heritability of the features capable of distinguishing a low methane emitting ruminant from a high methane emitting ruminant is equal to or more than 0.2. Exemplary methods for calculating heritability are further provided herein in the examples.
[0031] In preferred embodiments, once the heritability of the features is determined, the method further comprises obtaining estimated breeding values (EBV) of EME and the features. Preferably this comprises including the features as traits of the ruminant in genomic prediction models (such as univariate models), as known to the skilled person and further described herein in the Examples.
[0032] Once the EBV of EME is obtained, preferably, the EBV of EME and the features are correlated and features with a correlation between their EBV and EBV of EME lower then 0.20 are discarded.
[0033] Estimation of the phenotypic correlation and genetic correlation between the features retained until this point, as well as between the features and EME is conducted using genomic prediction models (such as bivariate prediction models).
[0034] Exemplary methods for conducting Principal Component (PC) analysis are further provided herein in the examples.
[0035] In any embodiment, the absolute value of genetic correlation among the features capable of distinguishing a low methane emitting ruminant from a high methane emitting ruminant is equal to or less than: about 0.99, about 0.95, about 0.9, about 0.85, about 0.8, about 0.75, about 0.7, about 0.65, about 0.6, about 0.65, about 0.6, about 0.55, about 0.5, about 0.45, about 0.4, about 0.35, about 0.3, about 0.25, about 0.2, about 0.15, or about 0.1. Preferably the absolute value of genetic correlation among the features capable of distinguishing a low methane emitting ruminant from a1005811148high methane emitting ruminant is equal to or less than: 0.99, 0.95, 0.9, 0.85, 0.8, 0.75, 0.7, 0.65, 0.6, 0.65, 0.6, 0.55, 0.5, 0.45, 0.4, 0.35, 0.3, 0.25, 0.2, 0.15, or 0.1. More preferably the absolute value of genetic correlation among the features capable of distinguishing a low methane emitting ruminant from a high methane emitting ruminant is equal to or less than 0.9.
[0036] In any embodiment, the absolute value of phenotypic correlation of the features capable of distinguishing a low methane emitting ruminant from a high methane emitting ruminant is equal to or more than: about 0.3, about 0.35, about 0.4, about 0.45, about 0.5, about 0.55, about 0.6, about 0.65, about 0.7, about 0.75, or about 0.8. Preferably the absolute value of phenotypic correlation of the features capable of distinguishing a low methane emitting ruminant from a high methane emitting ruminant is equal to or more than: 0.3, 0.35, 0.4, 0.45, 0.5, 0.55, 0.6, 0.65, 0.7, 0.75, or 0.8. More preferably the absolute value of phenotypic correlation of the features capable of distinguishing a low methane emitting ruminant from a high methane emitting ruminant is equal to or more than 0.5.
[0037] In any aspect or embodiment, a ruminant is classified as a high or low methane emitter based on the deviation of its methane emission from the mean methane emission of the population, for example, by identifying ruminants that emit more or less than a specified number of standard deviations from the mean. The deviation may be calculated using standard normal approximation. In some embodiments, the number of standard deviations is about 0.4, about 0.6, about 0.8, about 1 , about 1.2, about 2, about 3, or about 4. Preferably the number of standard deviations is 0.4, 0.6, 0.8, 1 , 1.2, 2, 3 or 4. More preferably the number of standard deviations is 0.674.
[0038] In any aspect or embodiment, a ruminant is classified as a high methane emitter if it emits methane at a quantity above the population mean, and a ruminant is classified as a low methane emitter if it emits methane at a quantiy below the population mean.
[0039] In any aspect or embodiment, a ruminant is classified as as high or low methane emitters based on the percentiles of methane emission as compared to the popualation. For instance, ruminants in the top quartile with respect to methane emission in the population are considered high emitters, while those in the bottom quartile are considered low emitters.1005811148
[0040] In any aspect or embodiment, a ruminant is classified as high methane emitter if it emits methane at a quantity of equal to or more than a reference value. The reference value may be about 380 g / day, about 400 g / day, about 420 g / day, about 440 g / day, about 460 g / day, about 480 g / day, about 500 g / day, about 520 g / day, about 540 g / day, or about 560 g / day. Preferably the reference value is 380 g / day, 400 g / day, 420 g / day, 440 g / day, 460 g / day, 480 g / day, 500 g / day, 520 g / day, 540 g / day, or 560 g / day. More preferably the reference value is 519 g / day. Alternatively the reference value may be about 1300 ppm, about 1350 ppm, about 1400 ppm, about 1450 ppm, about 1500 ppm, about 1550 ppm, about 1600 ppm, about 1650 ppm, or about 1700 pm. Preferably the reference value is 1300 ppm, 1350 ppm, 1400 ppm, 1450 ppm, 1500 ppm, 1550 ppm, 1600 ppm, 1650 ppm, or 1700 pm. More preferably the reference value is 1571 ppm.
[0041] In any aspect or embodiment, a ruminant is classified as low methane emitter if it emits methane at a quantity of equal to or less than a reference value. The reference value may be about about 260 g / day, about 280 g / day, about 300 g / day, about 320 g / day, about 340 g / day, about 360 g / day, about 380 g / day, about 400 g / day, about 420 g / day, about 440 g / day, about 460 g / day or about 480 g / day. Preferably the reference value is 260 g / day, 280 g / day, 300 g / day, 320 g / day, 340 g / day, 360 g / day, 380 g / day, 400 g / day, 420 g / day, 440 g / day, 460 g / day or 480 g / day. More preferably the reference value is 405 g / day. Alternatively the reference value may be about 800 ppm, about 850 ppm, about 900 ppm, about 950 ppm, about 1000 ppm, about 1050 ppm, about 1100 ppm, about 1150 ppm, about 1200 ppm, about 1250 ppm, about 1300 ppm, about 1350 ppm, or about 1400 ppm. Preferably the reference value is 800 ppm, 850 ppm, 900 ppm, 950 ppm, 1000 ppm, 1050 ppm, 1100 ppm, 1150 ppm, 1200 ppm, 1250 ppm, 1300 ppm, 1350 ppm, or 1400 ppm. More preferably the reference value is 1043 ppm.
[0042] The skilled person will appreciate that predicting the cattle’s DNA effect on EME from the microbial features is based on an index that produces a unitless value used to rank potential breeding candidates from low emitters to high EME emitters. (The effect of cattle DNA on E E is captured through the heritability of the traits (E E and microbial features) and the genetic correlations between them. The influence of cattle DNA is estimated using heritability and genetic correlations. Microbial DNA is used to identify the features.) For example,1005811148index output of an animal - (34.93456 * CLR(Methanobrevibacter)) - (46.86615 * CLR(K02416)) + (36.89908 * CLR(COG 1100))where CLR is the centered tog-ratio transformattons described at step b iv. above.
[0043] The index output of all animals is ranked from tow to high and the animals with tower EME values are typically selected for inclusion in breeding programs as parents.
[0044] Given that the methods of the abovementioned aspects enable the identification of a ruminant being useful for inclusion in a breeding program, there is further provided, a method of selecting a ruminant for inclusion in a breeding program, the method comprising:- obtaining or having obtained nucleic acid sequences from a microbiome sample of a ruminant being considered for inclusion in a breeding program,- determining the abundance of taxonomic or functional features in the sample, wherein the features are selected from:a) Methanobrevibacter, K02416, COG1100b) Methanobrevibacter, Entodinium, K02416, COG1151, COG0417, COG0667, CGG0076;c) Methanobrevibacter, K00937, K01681; ord) Archaea, Bacteria and / or Eukaryota at the superkingdom taxonomic level;or a combination thereof,- calculating a score that ranks the ruminant as being a high or low methane emitter, based on the abundance of the features relative to the mean abundance of the features in a reference population and wherein the features Methanobrevibacter, Entodinium, COG0417, COG0667, COG0076, CGG1100, Archaea and Eukaryota are positively correlated with EME and the features K02416, COG1151, K00937, K01681 and Bacteria, are negatively correlated with EME1005811148- selecting the ruminant for inclusion in a breeding program when the score indicates that the ruminant is a low methane emitter or- determining not to include the ruminant in a breeding program when the score indicates that the ruminant is a high methane emitter.
[0045] In any embodiment, the abundance may be an absolute abundance, which is the number of microbial sequences detected per sample from the ruminant, assigned to operational taxonomy units (OTUs) at the levels of superkingdom, phylum, class, order, family, genus, and species or to functional annotations such as KEGG Orthology (KO) or Clusters of Orthologous Groups (COG).
[0046] In any embodiment, the abundance may be a relative abundance, which is the absolute abundance of a taxonomic or functional feature expressed as a proportion of the total abundance of features of the same type within the sample of the ruminant. For example, the relative abundance of Methanobrevibacter is calculated relative to the total abundance of all genera in the sample. Similarly, the relative abundance of K02416 is calculated relative to the total abundance of all KEGG Orthology (KO) features, and the relative abundance of COG1151 is calculated relative to the total abundance of all Clusters of Orthologous Groups (COG) features in the sample.
[0047] In a preferred embodiment, the method further comprises estimating the genetic correlation between the features and EME in the reference population prior to obtaining the score.
[0048] The reference population may be the same population of which the ruminant is member. The reference population may alternatively be a population of ruminants of the same species, located in a different geographical location. Preferably in those circumstances, the ruminant population has similar environmental conditions as the ruminant for whom the level of EME is being determined. An example of a similar environmental condition includes: similar feed, similar climate conditions, and similar population density.
[0049] In certain embodiments, a combination of two or more groups or subpopulations of ruminants may be used to generate the reference population to estimate the effects of host’s genetic markers on EME and the rumen metagenome feature traits. The1005811148combination of two or more group / subpopulations may be from the same, similar or from different geographic or environmental conditions.
[0050] In any embodiment or aspect herein, selection of a ruminant for inclusion in a breeding program may also involve consideration of one or more additional properties of the ruminant. For example, the one or more properties of the ruminant (such as a cow) may comprise milk yield (MY) on the day of obtaining the milk of the cow, previous lactation (305-day) milk yield, previous lactation (305-day) fat yield, previous lactation (305-day) protein yield, days in milk (DIM) of the cow on the day of obtaining the milk of the cow, days from calving to insemination event (DAI) of the cow, calving age of the cow from a previous insemination, fertility genomic estimated breeding value (GEBV), and genotype of the cow.
[0051] Other properties of the cow are contemplated provided that they are related to, and contribute to, methane outcomes in cows. Some of these properties can be measured or accessed on-farm or on-site provided the facility has the necessary resources and previous conception and milk content information of each cow to do so. Otherwise, the information can be accessed from previously collated information which has been generated and stored off-site.
[0052] In any aspect or embodiment, the microbiome sample comprises one or more of: a sample of rumen fluid, a faecal sample, an oral sample, a sample taken from the rumen or any other location within the gastrointestinal tract, or a combination thereof.
[0053] In any embodiment, the methods described herein may further include determining the enteric methane emissions (EME) of the ruminant or population of ruminants by any suitable method known to the skilled person. Such methods include the use of the “SFe technique”, and the “sniffers” technique, which are known to the skilled person but also further defined herein.
[0054] In any aspect or embodiment, the ruminant or population of ruminants is bovine, ovine, caprine or bubaline. In any aspect or embodiment, where the ruminant or population of ruminants is bovine, the bovine(s) may be dairy cattle, beef cattle, or any other type of cattle, regardless of their zootechnical, socio-cultural or economic use.1005811148Preferably where the ruminant or population of ruminants is bovine, the bovine(s) may be dairy cattle or beef cattle.
[0055] In preferred examples the cattle is dairy cattle or a dairy cattle population. In a particularly preferred example, the dairy cattle is an Australian dairy cattle or an Australian dairy cattle population.
[0056] In a further aspect, the present disclosure provides for a kit for enabling the identification of whether a ruminant is a low or high methane emitter in accordance with the present invention - and, for example, that the ruminant is therefore suitable for inclusion in a breeding program.
[0057] In a preferred embodiment, the kit provides means for obtaining a rumen metagenome genotype for a ruminant. Such means, may comprise one or more probes or reagents for detecting nucleic acids in a rumen sample obtained from the ruminant. In particular, the means (such as oligonucleotide probes) may enable the detection and quantitation of nucleic acids from one or more of, encoding one or more of, or indicative of the abundance of one or more of the features: Methanobrevibacter, Entodinium, K02416, COG1100, COG1151, COG0417, COG0667, CGG0076, K00937, K01681 or nucleic acids indicative of the abundance of nucleic acids from Archaea, Bacteria and / or Eukaryota.
[0058] Optionally, the kit comprises written instructions for use in accordance with any method herein. The kit may also provide information relating to a reference population genome for enabling comparison of the test sample thereto, or a reference genome sample for enabling direct comparison of the quantity of nucleic acids.
[0059] As used herein, except where the context requires otherwise, the term "comprise" and variations of the term, such as "comprising", "comprises" and "comprised", are not intended to exclude further additives, components, integers or steps.
[0060] Further aspects of the present invention and further embodiments of the aspects described in the preceding paragraphs will become apparent from the following description, given by way of example and with reference to the accompanying drawings.1005811148Brief description of the drawings
[0061] Figure 1 - first two principal components (PC) depicting coordinates of the ruminal metagenome of dairy cattle located in Australia (A) and Spain (B) at genus taxonomic rank, coloured by enteric methane emissions and content of Archaea, Bacteria, and Eukaryota. MeP: enteric methane production recorded as grams per day (g / d) in Australia. MeC: Enteric methane concentration (MeC) recorded as parts per million (ppm) in Spain.
[0062] Figure 2 - Principal components (PC) 2 and 3 depicting coordinates of ruminal metagenome of dairy cattle located in Australia and Spain at genus taxonomic rank, coloured by population, content of Archaea, Bacteria, and Eukaryota, and rank of enteric methane emissions (EME Rank) within each population where higher values are animals with higher EME.
[0063] Figure 3 - genetic correlations between taxonomic genera, KOs and COGs considered potential genetic indicators of EME in the Australia (A) or Spain (B) population.
[0064] Figure 4 - estimated reduction of enteric methane emissions (EME) per generation in a dairy cattle population located in Australia. Top line: direct selection on EME. Middle line: indirect selection on ruminal microbial features. Bottom line:Combination of direct selection on EME and indirect selection on ruminal microbial features.
[0065] Figure 5 - Distribution of predicted responses to selection per generation against enteric methane production in grams per day (g / d) in Australian dairy cattle for three strategies (Methane, Microbes, and Methane and Microbes), based on 1,000 Monte Carlo simulations using the top 30% of the population and the breeding core subset A7. Methane: direct selection on methane; Microbes: indirect selection via rumen microbiota features; Methane and Microbes: combined direct and indirect selection. Each violin shows variability in response due to uncertainty in heritability and genetic correlations, simulated from normal distributions defined by their reported estimations and standard errors, with values constrained between 0 and 1 for heritabilities and between -1 and 1 for genetic correlations. Boxplots represent the median and1005811148interquartile range; violins indicate the kernel density of simulated outcomes. Predicted responses differed significantly across all strategies (Friedman test: x2= 405.83, df = 2, p < 2.2 x 10’16), with all pairwise comparisons statistically significant after Bonferroni correction (p < 2.0 x 1016).Detailed description of the embodiments
[0066] Direct EME measurements are costly and impractical for thousands of national herd animals. To solve this problem the inventors have developed the present invention to enable prediction of breeding values for EME for individual ruminants, using microbial features of the animal’s microbiome as indicators of EME.
[0067] Here, the inventors propose a new approach and contrasts with methods described in the prior art. In particular, the present inventors propose use of a frequentist framework and applied equations used in animal breeding which can be used to estimate the response to selection on emissions when selecting for the features.
[0068] In addition, the present inventors propose the integration of genomic information classified according to genera, KOs and COGs. The inventors propose, for the first time, the concept of using the rumen microbiome to connect reference populations for genomic selection aimed at reducing emissions.General definitions
[0069] It will be understood that the invention disclosed and defined in this specification extends to all alternative combinations of two or more of the individual features mentioned or evident from the text or drawings. All of these different combinations constitute various alternative aspects of the invention.
[0070] One skilled in the art will recognize many methods and materials similar or equivalent to those described herein, which could be used in the practice of the present invention. The present invention is in no way limited to the methods and materials described.
[0071] The various embodiments of the invention set forth below describe the identification of features in dairy cattle. However, it will be understood that in certain1005811148embodiments the invention is applicable to improving methane emissions of a wider range of livestock animals, especially ruminant animals including beef cattle and ruminant animals farmed for red meat production, including sheep, goat and buffalo.
[0072] As used herein, the term ‘ruminant animal’ refers to a mammal of the order Artiodactyla that digests plant-based food by initially softening it within the animal's first stomach, known as the rumen, then regurgitating the semi -digested mass, now known as cud, and chewing it again. The process of again chewing the cud to further break down plant matter and stimulate digestion is called “ruminating”. Ruminant animals include amongst others cattle including domestic cattle and production cattle including dairy cows and beef cattle, sheeps, lambs, goats, giraffes, American Bison, European bison, yaks, water buffalo, deer, camels, alpacas, llamas, wildebeest, antelope, pronghorn, and nilgai.
[0073] As used herein, the term “feature” is typically used to refer to the presence of a particular microbe, KO, COG or nucleic acid sequence present in a microbiome sample (such as rumen fluid, faecal matter) obtained from a ruminant. The features may therefore be identified through the detection of one or more nucleic acid sequences indicative of the presence of a particular microbe (or superkingdom or genera of microbes) in the sample. In further examples, the feature relates to a nucleic acid sequence indicative of the presence of particular enzymes produced by the microbiome, and which contribute to rumen digestive processes. Such nucleic acid sequences may be classified according to a KO or COG, as further herein defined.
[0074] It will be appreciated that the term “taxonomic or functional features” is intended to capture nucleic acid features from the microbiome being assessed. In practice, the nucleic acid sequences are assessed to determine the taxonomic or functional features of the microbiome. Accordingly, the term “nucleic acid feature” may be used interchangeably with “taxonomic and / or functional feature”.
[0075] In certain examples, the feature comprises the presence of a microbe of the genus Methanobrevibacter, or a nucleic acid indicative of the same. As used herein, Methanobrevibacter refers to a genus of archaeans in the family Methanobacteriaceae. The species within Methanobrevibacter are strictly anaerobic archaea that produce methane, for the most part through the reduction of carbon dioxide via hydrogen.1005811148
[0076] In certain examples, the feature comprises the presence of a microbe of the genus Entodinium, ora nucleic acid indicative of the same. As used herein, Entodinium refers to an anerobic binucleated ciliate, representing the most dominant protozoal species in the rumen of ruminants. Rumen ciliates play important roles in rumen function by digesting and fermenting feed and shaping the rumen microbiome.
[0077] As used herein, “KO” refers to the Kyoto Encyclopedia of Genes and Genomes, Orthology database. The KO (KEGG Orthology) database is a database of molecular functions represented in terms of functional orthologs. A functional ortholog is manually defined in the context of KEGG molecular networks, namely, KEGG pathway maps, BRITE hierarchies and KEGG modules, and is given a KO identifier called K number. Most KOs are defined from experimentally characterized genes and proteins in specific organisms, which are then generalized to other organisms based on sequence similarity. The granularity of "function" is context-dependent, and the resulting KO grouping may correspond to a group of highly similar sequences within a limited organism group or it may be a more divergent group.
[0078] The term KO system is used for a network-based classification of KOs.
[0079] It consists of six top categories (09100 to 09160) for KEGG pathway maps and one top category (09180) for BRITE hierarchies, as well as one top category (09190) for those KOs that are not yet included in either of them. The category numbers for these top categories and the second-level categories under metabolism (09101 to 09112) are used to define color coding of functions (see KEGG Color Codes).
[0080] In certain examples, the feature comprises the presence of nucleic acid sequence representative of nucleic acids found in K02416. K02416 refers to fliM; flagellar motor switch protein FliM.
[0081] In certain examples, the feature comprises the presence of nucleic acid sequence representative of nucleic acids found in K00937. As used herein, K00937 refers to K00937 ppk1; polyphosphate kinase [EC:2.7.4.1],
[0082] In certain examples, the feature comprises the presence of nucleic acid sequence representative of nucleic acids found in K01681. As used herein, K01681 refers to AGO; aconitate hydratase [EC:4.2.1.3],1005811148
[0083] As used herein, “COG” refers to entries in the NIH, National Center for Biotechnology Information database: Clusters of Orthologous Genes. The database was initially created in 1997 (Tatusov et al., (1997), Science, 278:631-7) followed by several updates, most recently in 2024 (Galperin et al., (2025) Nucleic Acids Res. 53(D1):D356-D363). The current update includes genomes from 2,103 bacterial and 193 archaeal species, of which 2,232 represent complete genomes in RefSeq and 64 are at the “Chromosome” level. The new features include updated COG annotations with corresponding references and PDB links, where available, and >100 new COGs for proteins involved in protein secretion pathways and those with unknown functions. The latest COG files with incremental updates can be found in the COG2024 folder on the NCBI's FTP site.
[0084] In certain examples, the feature comprises the presence of nucleic acid sequence representative of nucleic acids found in COG1151. As used herein, COG1151 refers to hydroxylamine reductase (hybrid-cluster protein).
[0085] In certain examples, the feature comprises the presence of nucleic acid sequence representative of nucleic acids found in COG0417. As used herein, COG0417 refers to DNA polymerase B elongation subunit.
[0086] In certain examples, the feature comprises the presence of nucleic acid sequence representative of nucleic acids found in COG0667. As used herein, COG0667 refers to Pyridoxal reductase Pdxl or related oxidoreductase, aldo / keto reductase family.
[0087] In certain examples, the feature comprises the presence of nucleic acid sequence representative of nucleic acids found in COG0076. As used herein, COG0076 refers to Glutamate or tyrosine decarboxylase or a related PLP-dependent protein.
[0088] In certain examples, the feature comprises the presence of nucleic acid sequence representative of nucleic acids found in COG1100. As used herein, COG1100 refers to GTPase SAR1 family domain.
[0089] In the context of the present invention, it will be appreciated that some of the nucleic acid features are negatively correlated with EME, and others are positively correlated with EME. The present inventors have found that:1005811148- Methanobrevibacter = positive correlation- Entodinium = positive correlation- K02416 = negative correlation- C0G1151 = negative correlation- COG0417 = positive correlation- COG0667 = positive correlation- COG0076 = positive correlation- C0G1100 = positive correlation- K00937 = negative correlation- K01681 = negative correlation- Bacteria = negative correlation- Eukaryota = positive correlationGenotyping of ruminants and their microbiome
[0090] The present invention requires obtaining genotypic information reflecting the microbiome of a ruminant or population of ruminants. The genotypic information can be obtained by measuring the nucleic acid components of a microbiome sample, such as a sample of rumen fluid, faeces or saliva, or a combination thereof.
[0091] The metagenome of a ruminant refers to the collection of genomes of all microorganisms, including bacteria, fungi and archaea, within a ruminant. In any aspect or embodiment herein, the metagenome of a ruminant may be determined by any genotyping method known to the skilled person including those described herein in the Examples.
[0092] Genotypic information may be determined by sequencing the whole genome, or a part thereof, of the sample, or by determining variations in the genome DNA sequence1005811148which may impart those characteristics. In some embodiments, the genotype may be determined through the identification of single nucleotide polymorphic (SNP) variants present in the genome of the ruminant. Identification of SNP variants may be determined using known techniques including the use of SNP microarrays including those available from Illumina Inc. (San Diego, California, USA) such as the BovineSNP50 Genotyping BeadChip, or via sequencing and analysis of genomic or exomic DNA.
[0093] In certain examples, microbiome metagenome samples (eg rumen fluid, faecal matter or similar) may be sequenced using whole genome sequencing techniques. In one example, Oxford Nanopore Technology may be used to obtain nucleic acid reads.Prediction modelling
[0094] To incorporate genotype data into a prediction model, a genomic relationship matrix (GRM - a matrix estimating the fraction of total DNA that two individual cows share) can first be derived. For example the GRM will be a matrix of size equivalent to the number of genotyped individuals by number of genotyped individuals that each off-diagonal position of the matrix represents.
[0095] The GRM can be derived using the method of Yang J et at., 2010, Nature Genet., 42(7): 565-569. An example of how genotype data is included in the prediction model is application of a principal component analysis on the GRM, where principal components from the GRM are included as additional predictors. Other methods of incorporation of genotype data are contemplated.
[0096] The genomic estimated breeding value (GEBV) is an estimate of the genetic value of an animal. A GEBV may be determined with reference to the likely EME characteristics of an animal and thus an EME GEBV may be calculated using metagenome and / or genotype information of the cow (e.g. genetic marker data such as SNP data) and a known prediction equation of EME (i.e. the GEBV is the sum of the number of specified alleles present at a locus multiplied by the effect at that locus).1005811148Other properties of the ruminant
[0097] Selection of the ruminant for inclusion in a breeding program may also involve consideration of one or more additional properties of the ruminant. The one or more properties of the ruminant (such as a cow) may be selected from: milk yield (MY) on the day of obtaining the milk of the cow, previous lactation (305-day) milk yield, previous lactation (305-day) fat yield, previous lactation (305-day) protein yield, days in milk (DIM) of the cow on the day of obtaining the milk of the cow, days from calving to insemination event (DAI) of the cow, calving age of the cow from a previous insemination, fertility genomic estimated breeding value (GEBV), and genotype of the cow.
[0098] Other properties of the cow are contemplated provided that they are related to, and contribute to, methane outcomes in cows. Some of these properties can be measured or accessed on-farm or on-site provided the facility has the necessary resources and previous conception and milk content information of each cow to do so. Otherwise, the information can be accessed from previously collated information which has been generated and stored off-site.
[0099] In some embodiments, the milk yield represents the amount of milk (in kilograms) produced by a cow from a current lactation on the day of herd or individual cow testing. In accordance with standard commercial practices of herd-testing in Australia, this represents milk obtained from a cow at an am and pm milking.
[0100] Previous lactation information is commonly determined over a period of 305 days from day 1 to day 305 of the previous lactation period. Milk yield, fat yield and protein yield over the 305 day period can be determined using the methods described above. Yields are typically expressed in kilograms for the 305 day period.
[0101] Days in milk (DIM) refers to the number of days the cow has been producing milk in the current lactation period on the day milk samples of the cow or herd were taken for analysis.
[0102] Days from calving to insemination event (DAI) refers to the number of days from the current calving to an insemination. The calving age of a cow is the age at which the1005811148cow calved from the last insemination event. The calving age is typically measured in months.
[0103] The genotype of a cow refers to the genetic constitution of the cow which is ultimately responsible for determining the characteristics of the cow. The genotype of the cow may be determined by methods described elsewhere herein, including in one or more of the Examples.
[0104] As used herein, the SFe (sulfur hexafluoride) tracer technique is used to measure methane (CH4) emissions from ruminants, especially cattle. It involves releasing a known amount of SFe, a stable and inert gas, and then measuring the ratio of SFe to methane in the animal's breath or exhaled gas. This allows researchers to estimate the total amount of methane produced by the animal.
[0105] As used herein, the “sniffers” technique refers to the measurement of methane using a non-dispersive infrared methane detector or 'sniffer' (The Guardian® NG) from Edinburgh Sensors (Livingston, Scotland, UK).ExamplesExample 1 - materials and methods
[0106] Data was collated from experiments completed in Australia and Spain, where ruminal fluid, enteric methane emissions (EME) and host genotypes were available for 829 Holstein dairy cows. The cited documents are incorporated herein by reference. Animals, Environment and Enteric Methane Emissions
[0107] The Australian population included 403 Holstein lactating cows located at the Ellinbank SmartFarm (Ellinbank, Victoria, Australia). Each cow had dry matter intake (DMI), and EME was measured in 11 cohorts between 2013 and 2017. At the beginning of the study, the cows averaged 110 ± 19.4 (mean ± standard deviation) days in milk (DIM), 2.5 ± 1.25 lactations, and weighed 539 ± 69.8 kg over 32 days in an experimental facility. The cows had continuous access to feed, water, and a bare paddock (loafing area) for rest. The cows were outside except for twice-daily milking at approximately 07:00 h and 15:00 h. All cows were fed a diet of compressed cubes of1005811148alfalfa hay and wheat grain. Individual cow DM I was measured using feed bins equipped with load cells. Electronic monitoring was linked to individual cow identification. Each cow's daily DM I was recorded over 32 days. Daily enteric methane production in the Australian population was measured as grams per day (MePAustraiia; g / d) and obtained using the sulphur hexafluoride (SFe) tracer method described by Deighton et al. (2014) Anim Feed Sci Technol. Methane production was measured over five days (on days 27 to 31) as described in Moate et al. (2021) Animals, emissions per cow per day calculated using the equation:1000wherein RCH4 is the calculated emission rate of ruminal CF (g / animal / d), RSF6 is the measured release rate of SFe from the permeation tube (mg / d), MWCH4 is the molecular mass of methane (16), MWSF6 is the molecular mass of SFe (146), the M subscript indicates a measured sample, and the BG subscript indicates a background concentration. The average over these days was used as MePAustraiia. The average ± standard deviation (SD) of MePAustraiia was 462 ± 84 g / d in the population.
[0108] The Spanish population included 426 Holstein cows, either in their first or second lactation, from 14 commercial farms across four Northern Spanish regions (Cantabria, Pais Vasco, Navarra, and Gerona). In Spain, the cows at the beginning of the measurement period averaged 165 ± 78 DIM, weighed 603 ± 27 kg, 364 were in the first lactation and 62 in the second lactation, 53 were in the first trimester of lactation, 144 in the second trimester and 229 in the last trimester of lactation. These animals were under commercial milk recording schemes consistent with ICAR-accredited recording protocols. Individual methane concentration of each cow in the Spanish population was recorded as parts per million (MeCspam; ppm) using a non-dispersive infrared methane detector or 'sniffer' (The Guardian® NG) from Edinburgh Sensors (Livingston, Scotland, UK) installed in the feed bin of an automatic milking system (Hammond et al. (2016) Anim Feed Sci Technol) following the methodology of Rey et al. (2019) Animals. All cows were measured every time they came into the automatic milking system (between 2 and 8 daily visits) for two to three weeks. The recorded eructation events were averaged per visit to the automatic milking system and minute,1005811148and then per 2-3 weeks period to obtain a single record per cow. The average ± SD of MeCspain was 1,307 ± 391 ppm in the population.
[0109] Based on the values of EME and assuming a roughly normal distribution of EME in the population, the inventors estimate that animals emitting less than approximately 405 g / day could be considered low emitters, and those emitting more than approximately 519 g / day could be considered high emitters. These thresholds correspond to the bottom and top quartiles, respectively, and were calculated using the standard normal approximation that quartiles lie at approximately ±0.674 standard deviations from the mean (i.e. , 462 ± 0.674 x 84).
[0110] Similarly, for methane concentration measurements obtained using sniffers in Spain, which had a mean of 1,307 ppm and a standard deviation of 391 ppm, the inventors consider that animals with methane concentrations below approximately 1,043 ppm would fall into the bottom quartile (low emitters), and those above approximately 1,571 ppm into the top quartile (high emitters), using the same ±0.674 SD rule (i.e., 1,307 ± 0.674 x 391).Cow Genotypes
[0111] The animals located in Australia were genotyped with SNP panels, including custom genotyping-by-sequencing (GBS) and custom SNP (XT, Illumina) arrays (approximately 8,800 SNP, of which at least 6,900 overlapped with the BovineSNP50 BeadChip, Illumina, San Diego, California, USA) and then imputed to the Bovine 50K SNP chip panel using the software Flmpute as described by Haile-Mariam et al. (2020) J Dairy Sci. The animals located in Spain were genotyped with the EURO12K SNP chip (Illumina, San Diego, California, USA) and then imputed to 54,609 SNPs using BEAGLE software as described in Jimenez-Montero et al. (2013) J Dairy Sci, using the Spanish reference population provided by the Spanish Holstein Association (CONAFE, Madrid, Spain) containing more than 200,000 genotypes. A panel with 39,058 (40K) SNP shared by both populations and with a minor allele frequency greater than 0.05 was selected for analyses.1005811148Ruminal Sample Collection
[0112] A single rumen sample was collected from each cow during the EME recording periods. Ruminal fluid samples from Australian cows were collected via an oesophageal probe placed into the rumen via the mouth. An oro-ruminal sampling probe, similar to the one described by Geishauser (2019) BovPract, and a vacuum pump were used to collect samples (Moate et al. (2014) J Dairy Sci). The oesophageal probe was a smooth-polished stainless-steel device comprising two parts: (1) a 170 mm long bolus of 42 mm diameter with a 2 mm screen and (2) a 4-meter-long flexible tube of 20 mm diameter. The collected samples were allowed to drain freely through a cheesecloth layer, separating rumen solids from the ruminal fluid, and then frozen at -80 °C.
[0113] In Spain, a mechanical device was used to elevate the cow's snout and approximately 100 ml of ruminal content from each cow was extracted as described by Saborio-Montero et al. (2021) Livest Sci by inserting a tube connected to a mechanical pump (Vacubrand ME 2SI, Wertheim, Germany) via the oesophagus. The obtained samples were then secured in sterilized containers. The hose and equipment were washed after each use. The samples were filtered using four layers of sterile cheesecloth to separate the solid elements, and the liquid portion was immediately frozen using liquid nitrogen vapours. These frozen samples were then conveyed to the lab in containers filled with liquid nitrogen and kept at -80 °C until analysis.Laboratory Processing of Rumen Samples
[0114] In Australia, microbial genomic DNA was extracted from 300 pl of ruminal fluid using a ZymoBIOMICS DNA miniprep kit (Zymo Research, Irvine, California, USA) according to the manufacturer's instructions with beadbeating performed using a multispeed vortex (Biofen MSV-3500) at 3500rpm for 40 minutes. DNA quality and concentration were determined using a Nanodrop and Qubit and normalized to the same concentration. In Spain, the samples were thawed and homogenized, and the DNA was extracted from 250 pl of ruminal fluid using the DNeasy Power Soil Kit (QIAGEN, Valencia, California, USA), and DNA concentration and purity were measured using a fluorometer and spectrophotometer.1005811148
[0115] Long-read sequencing with Oxford Nanopore Technologies (ONT) was used for metagenome sequencing in Australian and Spanish populations. Australian DNA samples underwent end repair with NEBNext end repair kit (New England Biolabs, Notting Hill, Victoria, Australia), and then libraries were prepared using the Native Barcoding expansion kit (SQK-NBD112.96, Oxford Nanopore Technologies, Oxford, United Kingdom) according to the manufacturer's instructions. Libraries were then sequenced using R9.4.1 flow cells on a PromethlON24 sequencer (Oxford Nanopore Technologies, Oxford, United Kingdom) according to the manufacturer's instructions. In Spain, one microgram of DNA per sample was used to prepare libraries using the ligation sequencing kit (SQK-LSK109, Oxford Nanopore Technologies, Oxford, United Kingdom) according to the manufacturer's instructions. Twelve samples were multiplexed in each run, and the barcoded samples were pooled for adapter ligation and sequencing using an R9.4.1 flow cell on a MinlON sequencer (Oxford Nanopore Technologies, Oxford, United Kingdom). Further information regarding sampling, DNA extraction and sequencing of Spanish samples is provided by Saborio-Montero et al. (2020) J Anim Breed Genet and Saborio-Montero et al. (2021) J Dairy Sci.Bioinformatic Pipeline
[0116] Basecalling was conducted using ONT's Guppy software with high accuracy mode (HAC) version 5.0.16 in Australia and Guppy version 6.5.7 and super high accuracy mode (SUP) in Spain. The reads were quality-controlled, ensuring a Quality Score greater than seven and a sequence length greater than 150 base pairs.Taxonomic and functional annotation was conducted with the SqueezeMeta software with version 1.4 in Australia and version 1.6.1 in Spain. SqueezeMeta uses DIAMOND Blastx to conduct alignments against the GenBank® database (Benson et al. (2014) Nucleic Acids Res) for taxonomic assignment, KEGG database (Kanehisa and Goto (2000) Nucleic Acids Res) for KEGG Orthology groups (KOs) and Clusters of Orthologous Groups (COG) database for COG identifiers (Tatusov et al. (2000) Nucleic Acids Res). The euk option of SqueezeMeta was utilized to enhance eukaryotic annotation. As a result, the features obtained were the number of reads per animal assigned to operational taxonomy units (OTUs) at superkingdom, phylum, class, order, family, and genus rank levels, and KO and COG identifiers. Following the methodology of Lopez-Garcia et al. (2022) Animals, sequences from nonmicrobial clades (Animalia1005811148and Plantae) and unclassified OTUs at the family level were removed from the taxonomic database. The KOs that included genes of Bos taurus (cow) were removed by accessing the KEGG database, identifying the KOs that included "BTA:" and removing them from the dataset. The COG database included Bacteria and Archaea data; therefore, it was not necessary to remove nonmicrobial data from its database. Prevalence Filter and Relative Abundance
[0117] Metagenomic ruminal features (genera, KOs, and COGs) that were not present in at least 90% of the animals within each population were removed. The missing values were recorded as zeros. These zeros were imputed to non-zero values within each population to mantain the actual compositionally of the data and allow computing logarithms in further steps. This imputation was done with the geometric Bayesian-multiplicative method from the cmultRepI function of the zCompositions R package. At this point, a ruminal microbial core of 93 genera, 2,496 KOs, and 2,198 COGs remained. The KOs of the core were classified into six categories using a bash script that used the curl command to fetch data from the KEGG database on June 16th, 2024: (1) Methane metabolism (KEGG pathway ko00680); (2) Metabolism under the KEGG category 09100 exclusive of methane-related activities (termed as 'Metabolism (other than methane)'); (3) Genetic information processing (KEGG category 09120); (4) Environmental information processing (09130); (5) Cellular processes (09140); and (6) Organismal systems (09150). The annotation of the COGs of the core was obtained using the command curl in an in-house bash script to fetch the annotation information from the COG database on June 16th, 2024.
[0118] Then, relative abundance (RA) matrices (B) of dimensions samples (n) times ruminal features (p) were constructed. When the ruminal features were genera, B was populated with the proportion of each genus in relation to the sum of all genera within the same animal. Similarly, when the ruminal features were KOs, B was populated with the proportion of each KO relative to the sum of all KOs within the same animal.Likewise, when the ruminal features were COGs, B was populated with the proportion of each COG relative to the sum of all COGs within the same animal.1005811148Centered log-ratio transformation
[0119] The B matrices were centered log ratio (CLR) transformed with the unweighted option of the CLR function from the easyCODA R package as:(BnX[B ] = log\g(Bi) / where X represents the CLR transformed matrix, Btj is the entry for animal i and genus or KO j in the matrix B, and g(B£) is the geometric mean of the vector of the ruminal feature j within the animal i defined as follows, and p is the number of genera or KOs:
[0120] The ranks superkingdom, phylum, class, order, and family of the genera used for this CLR transformation were used to create additional matrices populated with the sum of the genera's absolute abundance corresponding to that rank and transformed with the RA and CLR steps previously mentioned.Principal components analysis of ruminal microbiota
[0121] The first two principal components (PCs) of the CLR-transformed matrices (X) with the CLR-transformed genera, KOs, and COGs (six PCs in total) were calculated separately for the Australian and Spanish populations with the function prcomp of the R package stats with the options center and scale. The data for these two PCs were visualized with plots, coloured by EME, and the content of Archaea, Bacteria, and Eukaryota at the superkingdom rank. These superkingdom contents were derived from the sum of the CLR-transformed genera within each respective superkingdom.
[0122] Additionally, the first five PCs of the CLR-transformed matrices (X) with genera, KOs, and COGs (15 PCs in total) were also calculated for both populations together (Australian and Spanish) with the function prcomp of the R package stats with the options center and scale. These PCs from common matrices were visualized with plots, color-coded by population, the rank of the animals according to their EME within the1005811148population (from low to high emissions), and the content of Archaea, Bacteria, and Eukaryota at the superkingdom rank obtained.Breeding Core Subsets
[0123] Univariate genomic best linear unbiased prediction (GBLUP) models were implemented using the software ASReml-SA version 4.2 to estimate the heritability of EME traits, the PCs previously described and the CLR-transformed ruminal microbial features of the core. Separate GBLUP models were run for each trait, treating it as the dependent variable (y) within each population independently. The models were not combined across the two cattle populations. In Australia, the models were structured as follows:y = Ip + Xp + Zg+ ewhere y is a vector of EME or a metagenomic ruminal features; p is the population mean; 1 is a vector of ones with the same length as y; p is a vector of fixed effects; g is a vector of random additive genetic effects; e is a vector of random residuals; X and Z are incidence matrices. The fixed effects in Australia were cohort (11 levels), DMI, DIM, energy-corrected milk (ECM) obtained with the methodology of Visscher et al. (1994) Livest Prod Sci, and daily body weight change during the experiment. The phenotypic (rP) and genetic correlation (rg) between ECM and DMI obtained with a bivariate GBLUP model without effects was 0.79 ± 0.02 (standard error) and 0.93 ± 0.13, respectively. The effect of the year in the Australian dataset was not significant on emissions, probably because the effect of the cohort captured it, as each cohort was confined to a single year. The distribution of g was assumed N(0, Go|), where G is a genomic relationship matrix built for both populations together based on their SNP genotypes, utilizing the function Gmatrix of the R package AGHmatrix following the approach of Yang et al. (2010) Nat Genet. The distribution of e was assumed N(0, log)-
[0124] In Spain, the models were of the following form:y = Ip. + xp + Zg + Sf + e
[0125] where f is a vector of random effects of the 'sniffers' nested within farms (24 levels), and S is its incidence matrix. The distribution of f was assumed N(0, Fof), where1005811148F is the herd variance. The fixed effects in Spain were lactation number (2 levels) and stage of lactation (3 levels). The other elements in the Spanish model were the same as those previously defined for Australia.
[0126] These models were used to estimate the within-population heritability and obtain their genomic estimated breeding values (GEBVs). The heritability in Spain was estimated including the variability due to the 'sniffers' nested within farms in the denominator as follows:
[0127] In contrast, the heritability in Australia did not include the effect of farm in the denominator because all animals were from the same farm, as follows:
[0128] The GEBVs for EME and the CLR-transformed genera, KOs and COGs of the core, obtained from these univariate models were correlated. The genera, KOs and COGs that showed a heritability of at least 0.20 and an absolute value of the correlation between their GEBVs and the GEBVs of EME of at least 0.20 in both populations were considered potential genetic indicators of EME. The heritability threshold of 0.20 was selected to retain enough features for subsequent steps, based on the distribution of features’ heritability. Additionally, the PCs that did not capture the effect of the population and the superkingdoms content were also considered potential genetic indicators of EME.
[0129] The EME and the features considered potential genetic indicators of EME were included in the bivariate form of Model 1 and Model 2 to estimate their phenotypic and genetic correlations. The genetic and residual variances from the univariate models were used as priors in these bivariate GBLUPs to improve model convergence. These bivariate models were conducted using ASReml-SA version 4.2. Different combinations of the features considered potential genetic indicators of EME with bivariate converged models were evaluated as potential breeding core subsets.1005811148
[0130] Taxonomic genera, KOs and COGs considered potential genetic indicators of EME with a genetic correlation > 0.50 with EME and an absolute value of genetic correlation < 0.90 between them (to avoid redundant selection) were considered breeding core subsets A1 in Australia and S1 in Spain (Table 1.1). To visualise the genetic correlations between the taxonomic genera, KOs and COGs considered potential genetic indicators of EME, heatmaps of these correlations were constructed with the function pheatmap of the R library pheatmap using the options cutree_rows and cutree_cols to cluster the features in four groups.Table 1.1 - Breeding core subsets defined as groups of ruminal microbiota features from distinct cattle populations with the potential to be used as traits in breeding programs aimed at reducing EME. Principal components (PCs) depicted from the content of genera (genus), KEGG ontology identifiers (KOs) and Clusters of Orthologous Groups identifiers (COs) in the rumen microbiome of animals located either in Australia (A), Spain (S) of both (AS).1005811148
[0131] The breeding core subsets A2 to A6 were the PCs derived from the Australian dataset. The breeding core subsets A7 to A11 combined the PCs derived from the Australian dataset with the genera and KOs that converged with these PC. The breeding core subset S2 and S3 were the PCs derived from the Spanish dataset. The breeding core subsets S4 and S5 combined the PCs derived from the Spanish dataset with the genera Methanobrevibacter. The breeding core subsets AS1 to AS13 included features shared by the Australian and Spanish dairy cattle populations. The breeding core subset AS14 included features shared by the Australian and Spanish dairy cattle populationsand associated with EM E in a beef cattle population in the UK .
[0132] The genus Methanobrevibacter and EME are strongly genetically correlated. For this reason, Methanobrevibacterwas selected as part of the breeding core subsets, prioritising it over other genera with an absolute value of the genetic correlation greater than 0.90 with Methanobrevibacter. The KOs and COGs included in the breeding core subsets were selected over other KOs and COGs that had an absolute value of the genetic correlation > 0.90.1005811148Estimation of Selection Response of Enteric Methane Emissions
[0133] The inventors calculated three different scenarios to select for lower EME: (1) by directly selecting against EME only, (2) by indirectly selecting on the ruminal metagenomic features only, and (3) by using a combination of scenarios (1) and (2), selecting on both EME and the ruminal features. The response to selection in all scenarios was calculated with a selection index approach (Cameron (1997) CAB international) where the response variable was EME. These models incorporated the estimated heritability and genetic and phenotypic correlations. Further, the inventors calculated the response to selection in EME units and phenotypic standard deviations (op) when 30% to 1% of the population with the lowest methane emissions were selected. These percentages of the population selected represent the average for sires and dams. The proportion of the reduction in EME relative to direct selection for EME was estimated through scenarios 2 and 3. The breeding core subsets identified for each cattle population (Australia and Spain) were evaluated as indicative traits (proxies) of EME within the same population in which they were identified. Additionally, breeding core subsets that included features common to both populations were evaluated as indicative traits in both populations.Example 2 - resultsRumen Metagenome Core
[0134] The ruminal metagenome core shared between 90% of the dairy cattle populations in Australia and Spain was composed of 93 genera, 2,496 KOs, and 2,198 COGs.
[0135] The eukaryote genera in the core included genera of the family Neocallimastigaceae, such as Anaeromyces, Neocallimastix, and Piromyces, the eukaryote genus Entodinium, and the bacteria in phyla Bacteroidetes, Firmicutes and Proteobacteria. Firmicutes bacteria in the core included genera Selenomonas and Oribacterium. The archaea in the core were of the order Methanomassiliicoccales and the family Methanobacteriaceae, including the genera Methanobrevibacter and Methanosphaera. The functional features of the ruminal microbial core were involved in diverse functions.1005811148
[0136] Thirty-three KOs were involved in methane metabolism, 847 in metabolism (other than methane), 183 in genetic information processing, 286 in environmental information processing, 206 in cellular processes, and 48 in organismal systems.
[0137] The functional categorization of the rumen microbiome based on COG annotations revealed a complex array of metabolic and cellular processes. The most predominant category was translation, ribosomal structure, and biogenesis, with 224 entries highlighting the high protein synthesis activity within the microbiome. Categories such as general function prediction only (183 entries), amino acid transport and metabolism (174 entries), and carbohydrate transport and metabolism (173 entries) indicated the versatility and metabolic richness of the microbial community. Cell wall / membrane / envelope biogenesis (152 entries) and energy production and conversion (150 entries) were also well-represented, underscoring essential functions for microbial growth and maintenance. Post-translational modification, protein turnover, and chaperones (102 entries), along with replication, recombination, and repair (120 entries) further demonstrated the dynamic nature of the microbial ecosystem. Inorganic ion transport and metabolism (79 entries) and signal transduction mechanisms (83 entries) illustrated the complex regulatory and metabolic interactions. Categories with fewer entries, such as RNA processing and modification, chromatin structure and dynamics, and the cytoskeleton, each with only one entry, pointed to specialized and less common functions. This detailed functional profiling emphasized the multifaceted and intricate nature of the rumen microbial community.Principal Components of Rumen Metagenome Core
[0138] The first two PCs calculated from the matrices of ruminal metagenome genera, KOs, and COGs tended to separate higher EME in Australia in the same direction as a higher abundance of Archaea and Eukaryota and a lower abundance of Bacteria (Figure 1). Similarly, in the principal component analysis incorporating both populations, PC2 (when using genera) and PC3 (when using KOs and COGs) did not cluster by population rather they tended to separate by the abundance of Archaea, Bacteria, and Eukaryota (Figure 2). In the principal component analyses of matrices constructed with both populations, the effect of the population was captured by the first principal component.1005811148Breeding core subsets
[0139] The heatmaps constructed with the genetic correlations between taxonomic genera, KOs and COGs considered potential genetic indicators of EME showed strong correlations between these features (Figure 3). The absolute value of multiple of these genetic correlations were superior to 0.80. The breeding core subset A1 was composed of the genera Methanobrevibacter and Entodinium, along with K02416, COG1151, COG0417, COG0667 and COG0076 (Table 1.1). The breeding core subsets A2 to A6 corresponded of five out of six PCs derived from the Australian dataset. These five PCs were those that, when included with EME in bivariate models, resulted in model convergence.
[0140] The breeding core subsets S1 to S5 were tested in the Spanish dataset. The S1 included Methanobrevibacter, K02416 and COG1100. The PCs that converged in a bivariate model with EME in Spain were the first PC of the matrices constructed with genera, KOs and COGs. The breeding core subset S2 combined the PC1 derived the genera and the PC1 from the KOs in the Spanish dataset. The breeding core subset S3 was the PC1 derived the COGs in the Spanish dataset. The S4 and S5 combined the S2 and S3 with Methanobrevibacter.
[0141] The breeding core subsets AS1 to AS13 included features shared by both Australian and Spanish dairy cattle populations. These breeding core subsets included Methanobrevibacter, the superkingdoms Bacteria and Eukaryota, and four out of 15 PCs derived from matrices constructed with the data from both populations. These four PCs were PC2 and PC3 from the genera matrix, PC3 from the matrix of KOs, and PC3 from the matrix of COGs. The breeding core subset AS14 included the genus Methanobrevibacter, K00937 and K01681 shared by both Australian and Spanish dairy cattle populations and a beef cattle population in the UK.Genetic Parameters of Traits Aimed at Reducing Enteric Methane Emissions
[0142] The heritability of EME was 0.27 ± 0.13 (methane production with SFe) in Australia and 0.17 ± 0.11 (methane concentration with sniffers) in Spain. The heritability of all genera, KOs and COGs included in the breeding core subsets were between 0.20 ± 0.11 and 0.33 ± 0.12 in Australia and between 0.20 ± 0.11 and 0.41 ±10058111480.12 in Spain. The heritability of the PCs and Bacteria and Eukaryota as superkingdoms in the breeding core subsets was between 0.01 ± 0.09 and 0.20 ± 0.12 in Australia and between 0.30 ± 0.12 and 0.44 ± 0.12 in Spain. The genetic correlations between EME and ruminal microbial features in the breeding core subsets ranged between |0.56| ± 0.32 and |0.99| ± 0.23 in both populations.
[0143] There were high genetic correlations between the PCs and the other features of the breeding core subsets. In Australia, the genus Methanobrevibacter showed genetic correlations between |0.47| ± 0.30 and |0.89| ± 0.72 with eight PCs. The genus Entodinium showed genetic correlations between |0.75| ± 0.17 and |1.00| ± 0.33 with almost all PCs in Australia. Also in Australia, the K02416 showed genetic correlations between |0.54| ± 0.72 and |0.97| ± 0.30 with the PCs and multiple COGs showed genetic correlations of at least |0.80| with the PCs.
[0144] In Spain, the genetic correlations of the PCs with Bacteria and Eukaryota as superkingdoms were between |0.85| ± 0.10 and |1.00| ± 0.01, with the genus Methanobrevibacter between |0.76| ± 0.20 and |0.87| ± 0.11 and with K02416 and COG1100 between |0.66| ± 0.18 and |1.00| ± 0.20. In Spain, the Bacteria and Eukaryota superkingdoms and the PCs that separated these superkingdoms (Figures 1 and 2) were strongly genetically correlated (rgbetween |0.64| ± 0.30 and |1.00| ± 0.01;Figure 3).Table 2.1 - heritability (main diagonal), phenotypic correlations (above the main diagonal) and genetic correlations (below the main diagonal) between methane production in Australia and ruminal microbiota features. PC: principal components depicted from the content of genera (genus), KEGG ontology identifiers (KOs) and Clusters of Orthologous Groups identifiers (COs) in the rumen microbiome of animals located either in Australia (A) or Australia and Spain (AS). * Ko and Co found in Aus or Both followed by an arbitrary consecutive number. NC: No converged.1005811148Table 2.1 - continued1005811148Table 2.2 - heritability (main diagonal), phenotypic correlations (above the main diagonal) and genetic correlations (below the main diagonal) between methane1005811148concentration in Australia and ruminal microbiota features. PC: principal components depicted from the content of genera (genus), KEGG ontology identifiers (Ko) and Clusters of Orthologous Groups identifiers (Co) in the rumen microbiome of animals located either in Spain (S) or Australia and Spain (AS). * Ko and Co found in Aus or Both followed by an arbitrary consecutive number. NC: No converged.Table 2.3 - heritability (main diagonal), phenotypic correlations (above the main diagonal) and genetic correlations (below the main diagonal) between methane production (MeP) in Australia and methane concentration (MeC) in Spain with ruminal microbiota features shared between both dairy cattle populaitons located in Australia1005811148and Spain and beef cattle in the UK (ASU) followed by an arbitrary consecutive number for of KEGG ontology identifiers (Ko).Selection Response on Enteric Methane Emissions
[0145] The inventors estimated the phenotypic reduction of EME per generation that could be achieved by: 1) directly selecting on EME, 2) indirectly selecting on the breeding core subsets and 3) combining the direct and indirect selection. The phenotypic standard deviation of EME was 61.44 g / d in Australia and 388 ppm in Spain. When selecting the top 30 to 1% of the population, direct selection on EME were estimated to decrease between 0.33 and 0.75 phenotypic standard deviations of EME per generation in Australia and between 0.20 and 0.45 in Spain (Table 2.4).Table 2.4 - Estimated phenotypic reduction per generation of enteric methane emissions (EME) in two dairy cattle populations located in Australia and Spain using a direct selection against EME. Top: The top percentage of the population with the lowest methane emissions selected as parents. The trait in Australia is enteric methane production in grams per day (g / d). The trait in Spain is enteric methane concentration in parts per million (ppm). oP: Reductions in number of phenotypic standard deviaitons.1005811148
[0146] Larger EME reductions compared with direct selection were estimated in Australia by selecting on some breeding core subsets, with the larger reductions with the breeding core subsets A 1 and G (Table 2.5).1005811148Table 2.5 - Proportion of the reduction in enteric methane emissions (EME) in two dairy cattle populations located in Australia and Spain relative to only direct selection for EME, estimated through selection based on the ruminal microbiome (Microbes) or a combination of CH4 and the microbes. BCS: breeding core subsets, i.e. groups of ruminal microbial features. NA: Not applicable.1005811148
[0147] In Spain, larger EME reductions than direct selection on EME were estimated when selecting on the PC1 of the matrix constructed with genera and the PC1 from the KOs (breeding core subset S2) and when selecting for Bacteria as a superkingdom (breeding core subset V).
[0148] Multiple breeding core subsets were found to boost the estimated response to selection in EME in Australia when selecting for both EME and the microbial features together. Between them, the higher reductions (1.48 times the reductions achieved with direct emissions only) was estimated where selecting for the first PC of the matrix constructed with genera in Australia in combination with Methanobrevibacter, Entodinium and K02416 (breeding core subset A7; Table 2.5). Combining EME and the breeding core subsets in Spain was estimated to reduce between 1.05 and 1.39 times the reductions achieved with EME only, being the highest when selecting for Bacteria as a Superkingdom along with Methanobrevibacter (breeding core subset AS12). Selecting for EME and ruminal microbial features shared between two dairy cattle populations located in Australia and Spain and a beef cattle population from the UK (breeding core subset AS14) was estimated to increase the EME reductions (Table 2.5).Discussion
[0149] In a study of the rumen microbiome that included 829 dairy cows from Australia and Spain, the inventors evaluated a ruminal microbiome core of 93 genera, 2,496 KOs, and 2,198 COGs. The aim was to estimate the response to selection of these features on enteric methane emissions and to connect populations with different emission traits. The features in subsets of this core, denominated breeding core subsets, were heritable (h2> 0.20) and genetically correlated with EME (rg> 0.50). The breeding core subsets included combinations of two taxonomic superkingdoms, two genera in Australia and one in Spain, three KOs, six COGs in Australia and one in Spain, as well as principal1005811148components derived from matrices constructed with the core (nine PCs in Australia and seven in Spain).
[0150] The first principal components of matrices constructed with ruminal features of both populations captured the effect of the population and therefore, they were not evaluated as genetic indicative traits of EME. The PCs that distinguished lower bacterial content and higher eukaryotic content also tended to separate larger EME in Australia (Figure 1A). For this reason, the bacteria or eukaryota genera were expected to serve as an indicative trait of EME even when they are not methanogens.
[0151] Selecting fewer ruminal microbiota features is likely to be more practical for implementation in a breeding program aimed at reducing EME than selecting multiple features. Principal components and taxonomic superkingdoms aggregate information from numerous ruminal microbiota features. Heatmaps of the genetic correlations between taxonomic genera, KOs, and COGs identified as potential genetic indicators of EME show multiple features with absolute genetic correlations greater than 0.80 (Figure 3). These strong genetic correlations suggest that selecting a small number of these features could capture similar host genetic effects as selecting many.
[0152] The inventors evaluated three scenarios for selecting against EME: (1) direct selection against EME only, (2) indirect selection on ruminal metagenomic features only, and (3) a combination of scenarios 1 and 2. In scenario 2 in Australia, breeding core subset A1 and subsets A7 to A11, used as selection criteria, resulted in EME reductions comparable to or greater than those achieved through direct selection on EME (Table 2.5). This result is likely due to the strong genetic correlations between their features (see Table 1.1) and EME (rgranging from |0.68| to |0.99|). In Spain, all breeding core subsets in scenario 2 achieved EME reductions similar to those from direct selection on EME. In both populations, combining direct selection on EME with any breeding core subset (scenario 3) was estimated to achieve larger reductions than direct selection on EME alone.
[0153] To illustrate potential real selection scenarios, the inventors assumed selecting on the breeding core subset A7 in Australia. This subset included the first principal component of the matrix constructed with the genera information in Australia along with the genera Methanobrevibacter and Entodinium and K02416. In dairy cattle, the1005811148proportion of males and females in the population to select as parents is often different and a higher selection pressure is usually achieved in males, driving increased rates of genetic gain. For this reason, the inventors hypothetically assume selecting the top 1% of males and 30% of females, which means selecting approximately the top 15% of the population. In this scenario, the estimated EME reductions per generation would be 26.7 g / d (0.43 op) through direct selection (scenario 1; Figure 4; Table 2.6), 33.7 g / d (0.55 op) selecting on the breeding core subset A7 only (scenario 2) and 39.6 g / d (0.64 op) by combining direct and indirect selection (scenario 3). The average carbon dioxide equivalent (CO2e; kg / year) reduction per cow can be calculated as follows:number of oP* oP( ) * 365 days * GWPC02e (kg / year) = - - —v JL * 1.000 9
[0154] where L is a generation interval of 3 years and GWP is a conservative global warming potential of methane relative to carbon dioxide of 28 (IPCC, 2021). In this case, direct selection on EME would reduce 91 kg / year of CO2e (9.7 kg of methane) per cow on average, while selecting on the microbiome would reduce 115 kg of CO2e (12.3 kg of methane) and selecting on both EME and the microbes would reduce 135 kg / year of CO2e (14.5 kg of methane) per cow on average.
[0155] In another scenario, selecting the top 50% of females and 10% of males (averaging 30%) and using breeding core subset A7, the following outcomes were estimated: using direct selection on EME (scenario 1) would result in a reduction of 0.32 op per generation, equivalent to 68 kg / year of CO2e (7.3 kg of methane) per cow on average (Figure 4; Table 2.6). Figure 4 further suggests that selection based on certain breeding core subsets may reduce EME more effectively than direct selection for EME itself - even under the hypothetical scenario where methane could be measured on all farms. This outcome is consistent with the fact that EME originates from microbes rather than from the cow directly. If animals have both EME records and microbial profiles, even greater reductions in EME may be achieved.
[0156] Selection based on the microbiome (scenario 2) would achieve a reduction of 0.41 op per generation, or 86 kg / year of CO2e (9.2 kg of methane) per cow on average. Combining selection on both EME and the microbiome (scenario 3) would lead to a reduction of 0.48 op per generation, corresponding to 100 kg / year of CO2e (10.8 kg of1005811148methane) per cow on average. The robustness of the predicted responses under parameter uncertainty in this example was evaluated using a Monte Carlo sensitivity analysis (1,000 iterations; Figure 5). Despite variation in heritability and genetic correlation estimates by accounting for standard errors, the relative ranking of strategies (Methane, Microbes, and Methane and Microbes) was maintained, and predicted responses remained significantly different across all three strategies (p < 2.0 x 10'16). The distribution of simulated selection responses indicates that reductions close to the estimated values are the most frequent and thus likely to be observed in practice (Figure 5). In other words, the results were statistically robust, supporting the conclusion that greater reductions in methane emissions could be achieved by selecting for microbial features. Moreover, combining selection on methane emissions and microbial traits could lead to even larger reductions.Table 2.6 - estimation of average enteric methane emissions (EME) and carbon dioxide equivalent reductions per cow per year in Australia using the breeding core subset A7.Scenario 1: by directly selecting against EME only. Scenatio 2: by indirectly selecting on the ruminal metagenomic features only. Scenario 3: by using a combination of scenarios (1) and (2), selecting on both EME and the ruminal features. Parameters: phenotypic standard deviation (a_P) = 61.44; days in a year = 365; global warming potential (GWP) of methane relative to carbon dioxide = 28; generation interval (L) = 3; kg to g = 1000.1005811148
[0157] The results suggest that combining the PCs with genera, KOs and COGs could increase the EME reductions in breeding programs if all of them are heritable and genetically correlated with EME (Table 2.5). Although the PCs contained information of genera, KOs, and COGs, combining them with the original genera, KOs, and COGs add extra weight to those with higher heritabilities and genetic correlations with EME.Bacteria and Eukaryota superkingdoms and some taxonomic genera, KOs and COGs showed strong genetic correlations with the PCs, suggesting that these traits could be capturing similar genetic effects.
[0158] The results suggest that selecting on ruminal metagenomic features could achieve large reductions in EME. It is also important that the inventors found a consistency in some features in both Australia and Spain. The findings suggest that genomic selection on a common group of ruminal metagenome features shared between Australia and Spain would lead to reductions in EME in each of the populations, irrespective of the final EME trait (Table 2.5). Based on the result of combining populations and EME traits through the rumen metagenome, an international collaboration between the dairy, beef, and other ruminant industries could be achievable to reduce global EME. With this approach, the different instruments or traits in one population would not directly influence genetic selection in another population; rather, they would serve to confirm the association between EME and the breeding core subset under independent conditions.
[0159] Faecal and oral microbiomes could potentially serve as alternatives to the rumen microbiome.Conclusions
[0160] The inventors identified a ruminal metagenome core that could be used to reduce EME from dairy cattle in breeding programs and obtain larger reductions than direct selection on EME. A combination of direct selection on EME and indirect selection1005811148on the core could produce even larger reductions. These results suggest that rumen metagenome could have candidate traits to be improved with genomic selection programs along with enteric methane emission records. The results also suggest that sharing the ruminal metagenome core between reference populations and ruminal livestock industries could facilitate the prediction of EME irrespective of each population's EME trait.1005811148
Claims
CLAIMS1. A method of determining the likelihood that a ruminant is low or high methane emitting, comprising:- determining the abundance of taxonomic or functional features in a microbiome sample of the ruminant, wherein the features are selected from:a) Methanobrevibacter, K02416, COG1100b) Methanobrevibacter, Entodinium, K02416, COG1151, COG0417, COG0667, CGG0076;c) Methanobrevibacter, K00937, K01681; ord) Archaea, Bacteria and / or Eukaryota at the superkingdom taxonomic level;or a combination thereof,- calculating a score that ranks the ruminant as being a high or low methane emitter, based on the abundance of the features relative to the mean abundance of the features in a reference population and wherein the features Methanobrevibacter, Entodinium, COG0417, COG0667, COG0076, CGG1100, Archaea and Eukaryota are positively correlated with EME and the features K02416, COG1151, K00937, K01681 and Bacteria, are negatively correlated with EME,thereby determining the likelihood that a ruminant is low or high methane emitting.
2. A method of stratifying a population of ruminants into low or high methane emitters, comprising:- obtaining or having obtained nucleic acid sequences from microbiome samples of the individuals in the population,- determining the abundance of taxonomic or functional features in the samples, wherein the features are selected from:a) Methanobrevibacter, K02416, CGG11001005811148b) Methanobrevibacter, Entodinium, K02416, COG1151, COG0417, COG0667, CGG0076;c) Methanobrevibacter, K00937, K01681; ord) Archaea, Bacteria and / or Eukaryota at the superkingdom taxonomic level;or a combination thereof,- calculating a score that ranks each ruminant in the population as being a high or low methane emitter, based on the abundance of the features relative to the mean abundance of the features in the population, and wherein the features Methanobrevibacter, Entodinium, COG0417, COG0667, COG0076, CGG1100, Archaea and Eukaryota are positively correlated with EME and the features K02416, COG1151, K00937, K01681 and Bacteria, are negatively correlated with EME,thereby stratifying a population of ruminants into low or high methane emitters.
3. The method of claim 1 or 2, wherein the abundance is an absolute abundance of the taxonomic or functional feature, or a relative abundance calculated as the absolute abundance expressed as a proportion of the total abundance of features of the same type within the sample of the ruminant.
4. The method of any one of claims 1 to 3, wherein the method further comprises estimating the genetic correlation between the nucleic acid features and EME in the reference population prior to obtaining the score.
5. The method of claim 1, wherein the reference population is the same population of which the ruminant is a member.
6. The method of any one of claims 1 to 5, wherein the features are as defined in a).
7. The method of any one of claims 1 to 5, wherein the features are as defined in b).10058111488. The method of any one of claims 1 to 5, wherein the features are as defined in a) and in b).
9. The method of any one of claims 1 to 8, the method further comprising directly detecting the methane emission of the ruminant.
10. The method of claim 9, wherein the methane emission is determined using a non-dispersive infrared methane detector.
11. The method of claim 9, wherein the methane emission is determined using a sulphur hexafluoride (SFe) tracer method.
12. The method of any one of claims 1 to 8, wherein a ruminant identified as low methane emitting according to the method is then selected for breeding.
13. The method of claim 12, wherein a ruminant identified as low methane emitting by a method of any one of claims 1 to 8, and also by a method of directly detecting methane emission, is selected for breeding.
14. A ruminant offspring of the ruminant selected for breeding in claim 13.
15. A method of determining predictive features capable of distinguishing a low methane emitting ruminant from a high methane emitting ruminant, comprising:a. identifying an initial set of features in nucleic acid sequences from a microbiome sample of a ruminant, wherein the initial set of features comprise operational taxonomic units (OTUs), KEGG Orthology groups (KO) features and Clusters of Orthologous Groups (COG) features; b. pre-treating the initial set of features, comprising:i. filtering the initial set of features;ii. imputing zero values in the filtered features to non-zero values, thereby generating core features;iii. constructing abundance matrices from the relative abundance of the core features and methane emission traits of the animal; and1005811148iv. transforming the matrices obtained in step iii.;c. optionally performing principal component analysis of the transformed matrices, thereby identifying principal components; andd. estimating heritability of the core features, genetic and phenotypic correlation between the core features and the the methane emission traits,thereby determining the features, wherein the features comprise one or more of OTUs, KEGG features, COG features and PCs.
16. The method of claim 15, wherein in step a. the nucleic acid sequences are taxonomic assigned at the genus rank and functionally annotated as KOs and COG identifiers to produce absolute abundance matrices of dimensions animals times rumen features (OTUs, KOs or COGs) populated with the number of reads assigned to each feature.
17. The method of claims 15 or 16, wherein the step of imputing zero values (when pre-treating the initial set of features at step b.) comprises assigning a non-zero value to zeros recorded in the filtered absolute abundance matrix.
18. The method of any one of claims 15 to 17, wherein the step of filtering the initial set of features comprises removal of features that comprise genes of the ruminant.
19. The method of any one of claims 15 to 18, wherein the step of filtering comprises removal of microbial features that do not appear in at least about 95% of animals, more preferably in at least 90% of animals.
20. The method of any one of claims 15 to 19, wherein the step of constructing the relative abundance matrixes comprises obtaining the relative abundance of each feature with respect to other features of the same type within each sample.
21. The method of any one of claims 15 to 20, wherein the step of transforming the abundance matrixes comprises obtaining a log-ratio transformation that is applied to the relative abundances of genus-level taxa, KOs, and COGs.100581114822. The method of any one of the preceding claims, wherein the ruminant is a bovine, an ovine or a caprine.
23. The method of claim 22, wherein the bovine is cattle.
24. The method of claim 22, wherein the ovine is sheep.
25. The method of claim 22, wherein the caprine is a goat.
26. The method of claim 23, wherein the cattle are dairy cows.
27. The method of any one of claims 15 to 26, wherein the features are any one or more of:a) Methanobrevibacter, K02416, COG 1100b) Methanobrevibacter, Entodinium, K02416, COG1151, COG0417, COG0667, COG0076;c) Methanobrevibacter, K00937, K01681; ord) Archaea, Bacteria and / or Eukaryota at the superkingdom taxonomic level; or a combination thereof.
28. A method of selecting a ruminant for inclusion in a breeding program, the method comprising:- obtaining or having obtained nucleic acid sequences from a microbiome sample of a ruminant being considered for inclusion in a breeding program,- determining the abundance of taxonomic or functional features in the sample, wherein the features are selected from:a) Methanobrevibacter, K02416, CGG1100b) Methanobrevibacter, Entodinium, K02416, COG1151, COG0417, COG0667, CGG0076;c) Methanobrevibacter, K00937, K01681; or1005811148d) Archaea, Bacteria and / or Eukaryota at the superkingdom taxonomic level;or a combination thereof,- calculating a score that ranks the ruminant as being a high or low methane emitter, based on the abundance of the features relative to the mean abundance of the features in a reference population and wherein the features Methanobrevibacter, Entodinium, COG0417, COG0667, COG0076, CGG1100, Archaea and Eukaryota are positively correlated with EME and the features K02416, COG1151, K00937, K01681 and Bacteria, are negatively correlated with EME- selecting the ruminant for inclusion in a breeding program when the score indicates that the ruminant is a low methane emitter or- determining not to include the ruminant in a breeding program when the score indicates that the ruminant is a high methane emitter.
29. A kit for use in accordance with the method of any one of the preceding claims, wherein the kit comprises means for enabling detection and / or quantification of the abundance of any of the features listed in claim 1 a) to d), or a combination thereof.1005811148
