Microbial signatures of autism spectrum disorders

Abstract

Aspects of the present disclosure relate to methods and compositions for diagnosing and / or treating a subject having one or more symptoms of or at risk of developing an Autism Spectrum Disorder (ASD). In some embodiments, the method comprises determining and / or modulating the level of chorismate and / or a molecule in the chorismate metabolic pathway in the subject. In some embodiments, the method comprises determining and / or modulating the level of a microbial signature (e.g., Sarcina bacteria) in the subject. In some embodiments, the method comprises determining and / or modulating the level of proton flux in the gastrointestinal tract of the subject.

Description

[Technical field]

[0001] Related Applications This application claims the benefit under 35 U.S.C. §119(e) of U.S. Provisional Patent Application No. 63 / 311,757, entitled "Microbial Signatures of Autism Spectrum Disorder," filed February 18, 2022, which is incorporated by reference in its entirety for all purposes. [Background technology]

[0002] Autism spectrum disorders (ASD) represent a group of neurodevelopmental disorders characterized by a spectrum of behavioral, speech, and social impairments. Recent reports suggest that diagnosis has risen to 1 in 44 children (2.3%) in the United States, highlighting ASD as a disorder of increasing prevalence. The exact etiology of autism is unknown but is thought to be multifactorial.

[0003] The human gut microbiota has emerged as a key factor in a wide range of diseases, including neurological conditions such as ASD. Mechanistically, there are multiple hypotheses on how microbes may be involved in ASD, including the maternal immune activation model, production of neurotransmitters such as serotonin or GABA, modulation of vagus nerve signaling, and production of molecules such as 4-ethylphenyl sulfate or short-chain fatty acids. However, it remains unclear which of these mechanisms drive the development of ASD in any case. Thus, improved methods and compositions for diagnosing and / or treating ASD are needed. Summary of the Invention

[0004] In some embodiments, aspects of the invention relate to systems and methods for identifying one or more molecular and / or microbial features associated with Autism Spectrum Disorder (ASD) in a subject. In some embodiments, aspects of the invention relate to methods and compositions for treating one or more symptoms of ASD and / or ASD comorbidities in a subject or for reducing the risk of a subject developing ASD.

[0005] A systematic meta-analysis of studies profiling the gut microbiome of individuals with ASD revealed that in some aspects, common differentially abundant metabolic pathways represented by the gut microbiome of individuals with ASD involve a common precursor molecule, chorismate.

[0006] Thus, in some embodiments, methods and compositions for treating ASD include modulating a molecule in a pathway involving chorismate (e.g., the chorismate metabolic pathway). In some embodiments, a method of treating a subject with one or more symptoms of ASD and / or ASD comorbidities includes modulating the level of chorismate in the subject. In some embodiments, a method of treating a subject with one or more symptoms of ASD and / or ASD comorbidities includes modulating the level of one or more molecules in the chorismate metabolic pathway in the subject. In some embodiments, a method of treating a subject with one or more symptoms of ASD and / or ASD comorbidities includes modulating the activity or level of one or more enzymes or metabolites in the chorismate metabolic pathway (e.g., one or more of the non-limiting examples of molecules listed in Table 1 or Table 2). In some embodiments, modulating includes increasing the level and / or activity of the enzyme (e.g., administering an agonist). In some embodiments, modulating includes decreasing the level and / or activity of the enzyme (e.g., administering an antagonist). In some embodiments, chorismate mutase is modulated. In some embodiments, chorismate lyase is modulated (e.g., inhibited). In some embodiments, isochorismate synthase is modulated (e.g., inhibited). In some embodiments, aminodeoxychorismate synthase (also known as PABA synthase) is modulated (e.g., inhibited). In some embodiments, anthranilate synthase is modulated.

[0007] In some embodiments, a modulator (e.g., an agonist or antagonist of one or more enzymes or other molecules in the chorismate metabolic pathway) is administered to a subject having one or more symptoms or one or more symptoms of ASD and / or one or more comorbidities of ASD or to a subject at risk of developing autism.

[0008] In some embodiments, the level of chorismate or one or more molecules in a pathway related to chorismate (e.g., the chorismate metabolic pathway) is regulated in the gut of a subject (e.g., in the microbiota of a subject). In some embodiments, the level of chorismate or one or more molecules in a pathway related to chorismate (e.g., the chorismate metabolic pathway) is regulated in the blood or tissues (e.g., the brain, colon, and / or small intestine) of a subject.

[0009] In some embodiments, the level of chorismate or one or more molecules in a pathway related to chorismate (e.g., the chorismate metabolic pathway) is regulated by administering a composition that modulates the expression and / or activity of one or more enzymes in the chorismate metabolic pathway (e.g., in the gut microbes of a subject or in the blood or tissues of a subject). In some embodiments, the composition comprises one or more agonists and / or antagonists of one or more enzymes in the chorismate metabolic pathway.

[0010] In some embodiments, the level of chorismate or one or more molecules in a pathway related to chorismate (e.g., chorismate metabolic pathway) is regulated by administering a composition that regulates the level of microorganisms in the gut of a subject. In certain embodiments, the microorganism produces chorismate and / or one or more chorismate metabolites. In some embodiments, the composition is a microbial composition. In some embodiments, the composition comprises one or more antibiotics, probiotics, and / or prebiotics that can regulate the relative amount of different microorganisms in the gut of a subject.

[0011] In some embodiments, the method and composition for assisting in the diagnosis of ASD in subjects with one or more symptoms of ASD and / or ASD comorbidities or at risk of developing ASD comprises determining the level of chorismate and / or one or more molecules in chorismate metabolic pathway in the sample obtained from the subject.In some embodiments, for example, based on the analysis of data from subjects diagnosed with ASD, if the level of one or more of such molecules is statistically correlated with ASD, the subject is identified as having ASD or at risk of ASD.

[0012] In some embodiments, the sample is a fecal sample. In some embodiments, the sample is a serum sample. In some embodiments, the sample is a saliva, buccal, nasal, urine, cerebrospinal fluid, or gastrointestinal sample (e.g., a swab or biopsy).

[0013] In some embodiments, the subject is a human.In some embodiments, the subject is a child.In some embodiments, the subject is an adult.In some embodiments, the subject is at risk of developing ASD and / or one or more comorbidities of ASD.In some embodiments, the subject is diagnosed with one or more signs or symptoms of ASD and / or one or more comorbidities of ASD.

[0014] Other advantages and novel features of the present invention will become apparent from the following detailed description of various non-limiting embodiments of the invention when considered in conjunction with the accompanying drawings. In cases where the specification and documents incorporated by reference include conflicting and / or inconsistent disclosure, the present specification shall control.

[0015] Non-limiting aspects of the present invention are described, by way of example, with reference to the accompanying drawings, which are schematic and are not intended to be drawn to scale. In the drawings, each identical or nearly identical component illustrated is typically represented by a single numeral. For clarity, not every component is labeled in every figure, and not every component of each aspect of the present invention is shown where illustration is not necessary to allow a person skilled in the art to understand the invention. [Brief description of the drawings]

[0016] [Figure 1] FIG. 1 shows a schematic diagram of the shikimate pathway. [Diagram 2] FIG. 2 shows a schematic diagram of the chorismate metabolic pathway. [Diagram 3] FIG. 3 shows the association of ASD status with microbiome alpha and beta diversity, according to some embodiments. FIG. 3A shows a Principal Coordinate Analysis (PCoA) from an unweighted UniFrac distance matrix of 1740 samples. FIG. 3B shows a PCoA from a weighted UniFrac distance matrix of 1740 samples. FIG. 3C shows a pseudo-F statistic from a PERMANOVA beta group significance test of ASD vs. controls from unweighted and weighted UniFrac distance metrics. Full opacity color bars refer to PERMANOVA tests with P<0.05. FIG. 3D shows the Log2 fold change of alpha diversity values ​​of ASD vs. controls for each cohort sized by significance. Box plots show the median, quartiles, and 1.5×interquartile range of data distribution. P values ​​were generated from unadjusted two-tailed t-tests. [Fig. 4A-4H]FIG. 4 shows taxonomic and functional differences between ASD and controls according to some embodiments. FIG. 4A shows the Log2 fold change of the ASD / control relative abundance of the most significant phyla. Individual points are colored by cohort and size indicates significant differences between ASD and controls (unadjusted two-tailed t-test p-value<0.05). FIG. 4B shows the Log2 fold change of the ASD / control relative abundance of the most significant genera by coloring and sizing the same individual points as FIG. 4A. FIG. 4C-4E show sample-by-sample plots of the log10 transformed relative abundance of each sample for various classes and genera stratified by ASD status. Box plots show the median, quartiles, and 1.5×interquartile range of the data distribution. FIG. 4C shows the relative abundance of the Actinobacteria class in ASD vs. control children. FIG. 4D shows the relative abundance of the Prevotella genus in ASD vs. control children. Figure 4E shows the relative abundance of Bifidobacterium genus in ASD vs. control children. Figure 4F shows the significant microbial pathways when comparing ASD to controls. For each cohort treated individually, the log2 fold change of normalized pathway abundance values ​​in ASD / control is plotted for the most commonly significant bacterial metabolic pathways. Hierarchical clustering of studies and pathways was performed using Euclidean distance. Points are colored by log2 fold change and sized by significance (unadjusted two-tailed t-test P<0.05). Figure 4G shows the ASD / control normalized log2 fold change values ​​for all pathways involving menaquinone and ubiquinone in each cohort treated individually or aggregated. Individual points are colored by Log2 ASD / control fold change and sized to indicate statistical significance between ASD and controls (unadjusted two-tailed t-test p-values ​​for unequal variances <0.05). FIG. 4H shows ASD / control normalized log2 fold change values ​​for chorismate, aromatic amino acids, and vitamin B pathways from each cohort treated individually or collectively.Individual points are colored by Log2 ASD / control fold change and sized to indicate statistical significance between ASD and control (unadjusted two-tailed t-test p-values ​​<0.05 for unequal variances). The L-tyrosine, L-tryptophan biosynthetic superpathway, L-tryptophan biosynthesis, L-tyrosine biosynthetic superpathway, and L-phenylalanine biosynthetic pathway superpathways are involved in aromatic amino acids. The adenosylcobalamin biosynthesis and tetrahydrofumarate biosynthesis superpathways are pathways involved in vitamin B. [Figure 5A-5D] FIG. 5 shows an evaluation of machine learning performance for ASD status prediction according to some embodiments. FIG. 5A shows a bar graph of the mean AUC±SEM from 50 iterations. Eleven classification algorithms were tested using the aggregated ASV level dataset. Gradient boosting classifier and Adaboost models performed best and were further tested. FIG. 5B shows the mean accuracy rate of the trained and tested gradient boosting classifier models for collapsed read counts at different taxonomic levels on both the aggregated data and each individual study. Error bars show standard error from 5-fold cross-validation repeated 10 times. FIG. 5C shows the relative abundance and SHAP values ​​for each sample of the top 25 features plotted with decreasing feature importance from the gradient boosting classifier model trained on the aggregated data at the ASV level. FIG. 5D is a cluster heat map generated and colored by the SHAP values ​​of the top 25 features from the gradient boosting classifier models trained and tested independently on the aggregated data and each cohort. Both cohorts and features were clustered using Euclidean distance hierarchical clustering. [Figures 6A-6G]FIG. 6 shows an evaluation of study design factors that affect machine learning model performance and taxonomic abundance, according to some embodiments. FIGS. 6A-6D show (top) study-by-study log2 fold changes of ASD vs. control for genus Prevotella (FIG. 6A), ratio of Prevotella to Bacteroidetes (FIG. 6B), order Desulfovibrionales (FIG. 6C), and class Deltaproteobacteria (FIG. 6D) for data sequenced from V4 and V3-V4 hypervariable regions, respectively, and (bottom) log10 transformed relative abundances of genus Prevotella (FIG. 6A), ratio of Prevotella to Bacteroidetes (FIG. 6B), order Desulfovibrionales (FIG. 6C), and class Deltaproteobacteria (FIG. 6D) in each sample from V4 and V3-V4 regions, split by ASD status (V4 n=1024, V3-V4 n=352). P values ​​were calculated by unadjusted Wilcoxon two-tailed t-test. Box plots show the median, quartiles, and 1.5× interquartile range of data distribution. Figure 6E shows the machine learning AUC, accuracy, and F1 score of gradient boosting clarifier models trained to predict ASD status from samples split by sequencing depth and variable region (sequencing depth: >140000 n=851, >6000 n=640, variable region: V3-V4 n=356, V4 n=1032, V1-V2 n=103). Figure 6F shows the AUC, accuracy, and F1 score of the gradient boosting clarifier model in predicting ASD status from samples split by control relatedness, country, and gender (control relatedness: related controls n=462, unrelated controls n=1029; country: United States n=761, Ecuador n=46, China n=684; gender: unknown n=168, male n=1021, female n=302). Bar graphs show mean ± SEM. Figure 6G shows a heatmap of the average AUC of GBC models trained and tested on binned samples in each possible age range from 1 to 16. [Figure 7A-7D]FIG. 7 shows the abundance of chorismate branch point enzymes according to some embodiments. FIG. 7A shows the log2 fold change per study of normalized enzyme number abundance of ASD / control samples from the five main chorismate branch point enzymes. FIG. 7B shows box plots of predicted and normalized chorismate lyase EC numbers per sample stratified by cohort and ASD status. Two-tailed Wilcoxon t-tests were performed between groups. Significant differences are indicated to the right of each cohort (*, p<0.05; **, p<0.01, ***p<0.001). Box plots show median, quartiles, and 1.5×interquartile range of data distribution. FIG. 7C shows box plots of predicted and normalized menF gene abundance per sample stratified by cohort and ASD status. Wilcoxon two-tailed t-tests were performed between groups (*, p<0.05; **, p<0.01, ***p<0.001). Figure 7D shows a heatmap illustrating the presence or absence of bacterial features containing the menF gene, highlighting the bacterial features present in the greatest number of studies. The heatmap indicates the presence (1) or absence (0) of bacteria in each cohort. [Figure 8A] FIG. 8A shows a plot of the Log2 fold change between ASD and control subjects for commonly modulated genera, according to some embodiments. [Figure 8B] FIG. 8B shows a histogram and KDE curves illustrating the distribution of compositional abundance of Sarcina in the gut microbiome of ASD and control subjects, according to some embodiments. [Figure 8C] FIG. 8C shows a heatmap describing the fractional abundance of Sarcina in ASD or control subjects separated by cohort and study factors, according to some embodiments. [Figure 9A] FIG. 9A shows a bubble plot illustrating common significant pathways found in the gut microbiome of ASD subjects compared to controls, according to some embodiments. [Figure 9B]FIG. 9B shows histograms and KDE curves illustrating the distribution of composition abundance of the chorismate metabolic super-pathway in the gut microbiome of ASD and control subjects, according to some embodiments. [Figure 9C] FIG. 9C shows a heatmap depicting the abundance ratios of the chorismate metabolic super-pathway in ASD or control subjects separated by cohort and study factors, according to some embodiments. [Figure 9D] FIG. 9D shows a Venn diagram describing the overlap between the chorismate metabolic super-pathway and various features linked to ASD, according to some embodiments. [Figure 10A] FIG. 10A shows a bubble plot illustrating commonly significant metabolic flux differences found in the gut microbiome of ASD subjects compared to controls, according to some embodiments. [Figure 10B] FIG. 10B shows histograms and KDE curves illustrating the distribution of proton flux by the gut microbiome of ASD and control subjects, according to some embodiments. [Figure 10C] FIG. 10C shows a heatmap depicting the abundance fraction of ASD or control subjects containing above-threshold proton flux by the gut microbiome, separated by cohort and study factors, according to some embodiments. [Figure 11] FIG. 11 shows an upset diagram comparing patient populations of three subtypes of ASD to each other, according to some embodiments. [Figure 12A] FIG. 12A shows a bar graph of the top 30 metabolites modulated in ASD subjects with increased chorismate metabolism genes when given increased chorismate, according to some embodiments. [Figure 12B] FIG. 12B shows a box plot showing the difference in flux for L-tyrosine for chorismate subtype ASD subjects when given increased chorismate in a simulated diet, according to some embodiments. [Figure 12C]FIG. 12C shows a box plot illustrating the difference in flux for folate for chorismate subtype ASD subjects when given increased chorismate in a simulated diet, according to some embodiments. [Figure 12D] FIG. 12D shows box plots illustrating the flux differences for indole for chorismate subtype ASD subjects when given increased chorismate in a simulated diet, according to some embodiments. [Figure 12E] FIG. 12E shows a box plot illustrating the difference in flux for menaquinone 8 for chorismate subtype ASD subjects when given increased chorismate in a simulated diet, according to some embodiments. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0017] Aspects of the present invention relate to methods and compositions for diagnosing and / or treating subjects with one or more symptoms of autism spectrum disorder (ASD) and / or at risk of developing ASD. In some embodiments, compositions that modulate chorismate levels in a subject are useful for treating ASD. In some embodiments, compositions that modulate chorismate metabolism (e.g., biosynthesis of molecules downstream of chorismate) in a subject are useful for treating ASD. In some embodiments, compositions that modulate the relative abundance of microorganisms (e.g., Sarcina bacteria, Clostridium ventriculi) in a subject (e.g., in the gut of a subject) are useful for treating ASD. In some embodiments, compositions that modulate the level of proton flux in the gastrointestinal (GI) tract of a subject are useful for treating ASD.

[0018] According to an aspect of the present invention, evaluating the abundance of one or more genes in the chorismate metabolic superpathway in the human gut microbiome can help determine the usefulness of administering one or more therapeutic interventions for one or more of the core symptoms of autism. In some embodiments, the therapeutic intervention can include a single therapeutic intervention or a combination of two or more therapeutic interventions.

[0019] In some embodiments, the diagnosis of ASD in a subject is aided by assessing (e.g., determining and / or monitoring) the abundance of one or more genes (e.g., all of the genes or a subset of genes) in the microbiota (e.g., gut microbiota) in a sample from a subject in the chorismate metabolic superpathway.In some embodiments, the abundance of a gene can be assessed by determining the DNA and / or RNA level of the gene (e.g., using DNA and / or RNA sequence data).In some embodiments, the abundance of a gene can be assessed by determining the level of one or more gene products in a sample.

[0020] In some embodiments, diagnosis of ASD in a subject is aided by assessing (e.g., determining and / or monitoring) the level of one or more metabolites (e.g., a combination of such metabolites) of the chorismate metabolic superpathway in a sample from the subject.

[0021] In some embodiments, the method for treating a subject at risk of ASD or a subject with one or more symptoms of ASD and / or ASD comorbidities comprises administering probiotics and / or prebiotics to the subject, for example, to regulate the abundance of microorganisms that contain one or more genes in the chorismate metabolic superpathway.Thus, in some embodiments, probiotics and / or prebiotics can be used as a therapeutic intervention or in combination with one or more additional therapeutic interventions to treat one or more core symptoms of autism.

[0022] In some embodiments, administration of molecules that increase or decrease the activity of one or a combination of chorismate branch point enzymes in the gut microbiota can be used as a therapeutic intervention for one or more of the core symptoms of autism.

[0023] In some embodiments, administration of a probiotic containing a chorismate branch point enzyme with regulated activity levels may be used as a therapeutic intervention for one or more of the core symptoms of autism.

[0024] In some embodiments, administration of molecules such as chorismic acid, chorismic acid, and / or salts or esters of either thereof can be used as a therapeutic intervention for one or more of the core symptoms of autism. In some embodiments, administration of other molecules in the chorismic acid metabolic superpathway can be used as a therapeutic intervention for one or more of the core symptoms of autism. In some embodiments, administration of chorismic acid, chorismic acid, and / or salts or esters of either thereof in combination with one or more other molecules in the chorismic acid metabolic superpathway can be used as a therapeutic intervention for one or more of the core symptoms of autism.

[0025] In some embodiments, assessing (e.g., determining and / or monitoring) the activity of chorismate branch point enzyme in a subject's gut microbiota can be used as an aid in diagnosing ASD in a subject.

[0026] In some embodiments, assessing (e.g., determining and / or monitoring) the activity of chorismate branch point enzyme in a subject's gut microbiota can be used as an aid in determining an appropriate therapeutic intervention or combination of therapeutic interventions to administer to the subject.

[0027] These and other aspects of the invention are described in more detail below.

[0028] As used herein, "ASD" refers to autism spectrum disorder.

[0029] As used herein, "core symptoms of autism" include, but are not limited to, irritability, insomnia, speech and communication problems, aggression, digestive problems, socialization problems, restricted and repetitive patterns of behavior and interests, anxiety attacks, self-aggression, and mood disorders.

[0030] As used herein, "chorismate metabolic superpathway" refers to pathways for the production of chorismate from D-erythrose 4-phosphate and phosphoenolpyruvate (i.e., the shikimate pathway), as well as pathways for the conversion of chorismate to downstream metabolites (e.g., aromatic amino acids and their derivatives, ubiquinol, menaquinol, tetrahydrofolate, enterobactin). As used herein, "chorismate metabolic pathway" refers to the chorismate metabolic superpathway or a pathway within the chorismate metabolic superpathway.

[0031] As used herein, chorismate metabolic superpathway genes include, but are not limited to, any microbial genes from the human gut microbiome involved in the production of chorismate from D-erythrose 4-phosphate and phosphoenolpyruvate, and the conversion of chorismate to metabolites of ubiquinol, menaquinol, tetrahydrofolate, enterobactin, L-tryptophan, L-tyrosine, L-phenylalanine, and L-tryptophan, including, but not limited to, serotonin, melatonin, kynurenine, indole, and any intermediate molecules. A non-limiting list of gene names can be found in Table 1.

[0032] As used herein, metabolites of the chorismate metabolic superpathway include, but are not limited to, any of the intermediate molecules involved in the production of chorismate from D-erythrose 4-phosphate and phosphoenolpyruvate, or molecules derived from chorismate. A non-limiting list of metabolite names can be found in Table 2.

[0033] As used herein, "chorismate branch point enzymes" include, but are not limited to, chorismate mutase, anthranilate synthase, aminodeoxychorismate synthase (also known as para-aminobenzoic acid (PABA) synthase), isochorismate synthase, menaquinone-specific isochorismate synthase, chorismate dehydratase, and chorismate lyase.

[0034] As used herein, "therapeutic intervention" includes, but is not limited to, melatonin supplements, folic acid supplements, probiotics, fecal transplants, vitamin K supplementation, tryptophan supplementation, or any invention that modulates serotonin, kynurenine, or indole, fructooligosaccharides, shikimic acid or shikimic acid or any salt or ester thereof, chorismic acid or chorismic acid or any salt or ester thereof, and modulators (e.g., inhibitors or agonists) of chorismate branch point enzyme and / or one or more genes or proteins in Table 1 or Table 2.

[0035] Described herein are embodiments of methods and compositions for assessing a subject's risk of having or developing an Autism Spectrum Disorder (ASD), for identifying a treatment to be provided to a subject to reduce the risk of developing an ASD or to ameliorate or alleviate one or more symptoms of an ASD and / or a comorbid condition associated with an ASD, and / or for treating a subject with an identified treatment.

[0036] According to some embodiments, the method described herein includes obtaining a sample from a subject. In some embodiments, the subject is a human subject. In certain embodiments, the subject is a male. In certain embodiments, the subject is a female. The subject can be from any geographical location. In certain embodiments, the subject is from North America (e.g., the United States), Asia (e.g., China, Korea), Europe (e.g., Italy), and / or South America (e.g., Ecuador).

[0037] The subject can be of any age. In certain embodiments, the subject is a child (e.g., 17 years old or younger). In certain embodiments, the subject is an adult (e.g., 18 years old or older). In some embodiments, the subject is a human being, e.g., a human being, 1-5 years old, 1-7 years old, 1-10 years old, 1-15 years old, 1-17 years old, 1-18 years old, 1-21 years old, 1-50 years old, 1-70 years old, 1-100 years old, 2-7 years old, 2-10 years old, 2-15 years old, 2-17 years old, 2-18 years old, 2-21 years old, 2-50 years old, 2-70 years old, 2-100 years old, 5-10 years old, 5-15 years old, 5-17 years old, 5-18 years old, 5-21 years old, 5-50 years old, 5-70 years old, 5-8 years old, 5-9 years old, 5-10 years old, 5-15 years old, 5-17 years old, 5-18 years old, 5-21 years old, 5-50 years old, 5-70 ...0 years old, 5-10 years old, 5-10 years old, , 5-100, 10-15, 10-17, 10-18, 10-21, 10-50, 10-70, 10-100, 15-21, 15-50, 15-70, 15-100, 18-50, 18-70, 18-100, 21-50, 21-70, 21-100, 50-70, 50-100, or 70-100.

[0038] In some embodiments, the sample is a fecal sample. In certain embodiments, obtaining a sample includes collecting a fecal sample (e.g., collecting a stool sample, swabbing the rectal cavity). In some embodiments, the sample is a gastrointestinal sample. In certain cases, the gastrointestinal sample includes a gastric aspirate, a biopsy (e.g., mucosal biopsy, gastric tissue biopsy), intestinal fluid, an endoscopic brush, and / or a laser capture microdissection sample. In some embodiments, obtaining a sample from a subject includes performing a biopsy, an endoscopy, and / or a colonoscopy on the subject. In some embodiments, the sample is a serum, cerebrospinal fluid (CSF), urine, oral, nasal, or saliva sample. In some embodiments, obtaining a sample from a subject includes collecting a serum sample, performing a spinal tap on the subject, collecting a urine sample, and / or swabbing the cheek, throat, and / or nasal cavity of the subject.

[0039] In some embodiments, the methods described herein include analyzing a sample obtained from a subject to evaluate one or more characteristics of the subject's microbiome. In some cases, the one or more characteristics include the identity of one or more microorganisms detected in the microbiome and / or the amount or relative amount of one or more microorganisms detected in the microbiome.

[0040] In some cases, the microbiome can be a microbiome of the gastrointestinal tract of a subject (e.g., a gut microbiome including gut microbiota). In some embodiments, the sample can represent a characteristic of the gastrointestinal tract of a subject. In some cases, the sample is a fecal sample. In some cases, the sample is a gastrointestinal sample. In certain cases, the gastrointestinal sample is taken from the gastrointestinal tract (e.g., lower gastrointestinal tract, upper gastrointestinal tract). The sample can be obtained during any suitable procedure, including, but not limited to, a colonoscopy, an endoscopy, a swab, or a brush of a portion of the gastrointestinal tract.

[0041] Although some examples described herein focus on the gastrointestinal microbiome (of the upper and / or lower gastrointestinal tract), embodiments are not so limited, and other embodiments may relate to microbiomes for other parts of the subject's anatomy (e.g., sinuses, skin, lungs, oral mucosa, or others). Any of the techniques described above for obtaining samples may be used to obtain samples from these other parts of the subject's anatomy, when medically appropriate. The obtained samples may be prepared and the microbes therein may then be analyzed to determine microbial signatures. In some cases, embodiments may relate to microbiome products (e.g., chorismate and its metabolites) present in other parts of the subject's anatomy (e.g., blood, brain, etc.).

[0042] In some embodiments, the methods described herein include determining the level (e.g., abundance) of a molecule (e.g., chorismate, a molecule in the chorismate metabolic pathway) and / or a microbial feature (e.g., a microorganism) in a sample from a subject.

[0043] In some embodiments, determining the level of molecular and / or microbial features in comprises determining the abundance of one or more genes present in a sample from the subject (e.g., one or more genes in the chorismate metabolic superpathway, including but not limited to one or more genes listed in Table 1). In some embodiments, determining the abundance of one or more genes present in a sample comprises extracting DNA and / or RNA from at least a portion of the sample. Any nucleic acid extraction method known in the art may be used. In some embodiments, determining the abundance of one or more genes present in a sample further comprises amplifying at least a portion of the DNA and / or RNA to produce a plurality of amplicons. Any nucleic acid amplification method known in the art may be used. Non-limiting examples of suitable nucleic acid amplification methods include polymerase chain reaction (PCR) and isothermal amplification methods (e.g., loop-mediated isothermal amplification (LAMP), rolling circle amplification (RCA), nucleic acid sequence-based amplification (NASBA)). In some embodiments, determining the abundance of one or more genes present in a sample comprises performing a quantitative nucleic acid amplification method. A non-limiting example of a suitable quantitative nucleic acid amplification method is quantitative PCR (qPCR). In some embodiments, determining the abundance of one or more genes present in the sample further comprises sequencing one or more amplicons.Any suitable nucleic acid sequencing method can be used.In certain embodiments, the nucleic acid sequencing method is a long-read sequencing method.In certain embodiments, the nucleic acid sequencing method is a short-read sequencing method.In some embodiments, the nucleic acid sequencing method is a next-generation sequencing method.In some embodiments, the nucleic acid sequencing method is carried out using Illumina, Pacific Biosciences, Oxford Nanopore, and / or Roche 454 sequencing platform.

[0044] In some embodiments, determining the abundance of one or more genes present in the sample comprises sequencing at least a portion of the 16S ribosomal RNA (rRNA) of the microorganisms present in the sample (e.g., performing 16S rRNA gene amplicon (16S) sequencing). In some cases, performing 16S sequencing comprises amplifying at least a portion of one or more regions of the 16S rRNA genome of the microorganisms present in the sample to produce a plurality of amplicons. Any nucleic acid amplification method (e.g., PCR) may be used. In some embodiments, the one or more regions of the 16S rRNA genome include at least one hypervariable region (e.g., V4, V3-V4, V1-V2). In some cases, performing 16S sequencing further comprises sequencing one or more amplicons of the plurality of amplicons. Any suitable nucleic acid sequencing method may be used.

[0045] In some embodiments, determining the abundance of one or more genes present in sample comprises performing shotgun metagenomic sequencing to sequence the metagenomic content of sample.In some cases, the shotgun metagenomic sequencing method comprises extracting DNA from sample, fragmenting the extracted DNA into DNA fragments, and sequencing the DNA fragments.Any suitable nucleic acid sequencing method can be used.In some cases, the sequence of the resulting DNA fragments can be analyzed to identify the microorganisms present in sample.

[0046] In some embodiments, determining the abundance of one or more genes present in a sample comprises determining the level of one or more gene products (e.g., enzymes) in the sample. The level of one or more gene products may be determined according to any method known in the art, including, but not limited to, enzyme-linked immunosorbent assays (ELISAs) and other antibody-based assays, and liquid chromatography-mass spectrometry (LC-MS)-based proteomics and other protein-based quantitative assays. In some embodiments, determining the abundance of one or more genes present in a sample comprises determining the level of one or more molecules (e.g., chorismate and / or its metabolites) that directly or indirectly interact with one or more gene products (e.g., enzymes) in a metabolic pathway (e.g., as a substrate or product). The level of one or more molecules (e.g., substrates, products) may be determined according to any method known in the art, including, but not limited to, high performance liquid chromatography (HPLC), liquid chromatography-mass spectrometry (LC-MS), gas chromatography-mass spectrometry (GC-MS), and fluorescence-based assays.

[0047] In some embodiments, determining the level of a molecule and / or microbial feature in a sample from a subject comprises measuring the level of a molecule and / or microbial feature in the sample. The level of a molecule and / or microbial feature in a sample can be measured according to any method known in the art. Non-limiting examples of suitable methods include HPLC, LC-MS, GC-MS, and matrix-assisted laser desorption / ionization time-of-flight mass spectrometry (MALDI-TOF). In certain embodiments, the level of a microbial feature in a sample can be measured by plating at least a portion of the sample on different media and / or under different conditions to evaluate the growth of microorganisms and identify one or more microorganisms present in the sample.

[0048] According to some embodiments, the method includes determining that the subject has at least one subtype of ASD or is at risk of developing at least one subtype of ASD when the level of the molecular or microbial feature is higher or lower than a threshold value. In some cases, the threshold value can be determined based on the level of the molecular or microbial feature in one or more neurotypical individuals (e.g., individuals not diagnosed with ASD). In some cases, the threshold value can be determined by identifying an optimal discrimination boundary between a first population of individuals diagnosed with ASD and a second population of neurotypical individuals.

[0049] According to some embodiments, the method includes identifying the subtype of ASD that the subject has or is at risk of having. In some embodiments, the method includes determining the composition of fecal microbiome or the presence or level of one or more molecular and / or microbial features in a sample obtained from the subject. In some embodiments, the subtype of ASD is a chorismate subtype. In some cases, the chorismate subtype can be characterized by a modulated level of chorismate and / or molecules in the chorismate metabolic pathway (e.g., shikimate pathway, pathway for conversion of chorismate to aromatic amino acids and their derivatives, folate and its derivatives, ubiquinol, menaquinol, and enterobactin) in the sample (e.g., relative to one or more neurotypical individuals). In some cases, the modulated level of chorismate and / or molecules in the chorismate metabolic pathway can be above or below a certain threshold. In some embodiments, the subtype of ASD is a Sarcina subtype. In some cases, Sarcina subtype can be characterized by a modulated (e.g., elevated) level of Sarcina bacteria in the sample (e.g., relative to one or more neurotypical individuals). In some embodiments, the subtype of ASD is a proton influx subtype. In some cases, the proton influx subtype can be characterized by a modulated (e.g., elevated) level of proton influx in the gastrointestinal tract of the subject.

[0050] Chorismic Acid The chorismate hypothesis of autism In some aspects, the assessment and regulation of the level of chorismate and / or one or more molecules in the chorismate metabolic pathway in a subject (e.g., in the subject's microbiome) can be useful for the treatment and / or diagnosis of ASD. Wherever chorismate is used herein, unless the context dictates otherwise, chorismate and / or any salt or ester of either chorismate or chorismate can also be used. In general, it should be understood that reference to a compound in its ionic, protonated, salt, or ester form can encompass other forms, unless the context dictates otherwise.

[0051] In some aspects, either the relative absence or abundance of chorismate and / or one or more molecules in the chorismate metabolic pathway can be detected in a subject with one or more symptoms of ASD. In some aspects, either the relative absence or abundance of bacteria likely to contain the capacity to regulate levels of chorismate and its related molecules can be detected in a subject with one or more symptoms of ASD.

[0052] Chorismate is the end product of the shikimate pathway, which is found in microorganisms and plants but not in animals. An exemplary depiction of the shikimate pathway is shown in FIG. 1. As shown in FIG. 1, the shikimate pathway includes seven enzymatic steps: (1) 3-deoxy-D-arabino-heptulosonate-7-phosphate synthase (DAHPS) converts D-erythrose 4-phosphate (e.g., an intermediate of the pentose phosphate cycle) and phosphoenolpyruvate (e.g., an intermediate of glycolysis) to 3-deoxy-D-arabino-heptulosonate-7-phosphate (DAHP); (2) 3-dehydroquinate synthase (DHQS) converts DAHP to 3-dehydroquinate; (3) 3-dehydroquinate dehydratase (DHQ) converts 3-dehydroquinate to 3- (4) shikimate 5-dehydrogenase (SDH) converts 3-dehydroshikimate to shikimate; (5) shikimate kinase (SK) converts shikimate to shikimate 3-phosphate; (6) 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS) converts shikimate 3-phosphate to 5-O-carboxyvinyl-3-phosphoshikimate (also referred to as 5-enolpyruvylshikimate-3-phosphate (EPSP)); and (7) chorismate synthase (CS) converts 5-O-carboxyvinyl-3-phosphoshikimate to chorismate.

[0053] Chorismate can then be converted into several biologically relevant aromatic molecules, including, but not limited to, aromatic amino acids (e.g., phenylalanine, tyrosine, tryptophan) and their derivatives, folic acid and its derivatives (e.g., tetrahydrofolate), siderophores (e.g., enterobactin), ubiquinol (e.g., coenzyme Q), and menaquinol. Figure 2 shows an exemplary depiction of five chorismate-dependent pathways. In one pathway, chorismate mutase converts chorismate to prephenate, which is the precursor of phenylalanine and tyrosine. In another pathway, anthranilate synthase converts chorismate to anthranilate, which is the precursor of tryptophan. In another pathway, aminodeoxychorismate synthase (also known as para-aminobenzoic acid (PABA) synthase) converts chorismate to p-aminobenzoic acid, which is the precursor of folic acid. In another pathway, isochorismate synthase converts chorismate to isochorismate, which is a precursor of enterobactin. In another pathway, chorismate lyase converts chorismate to p-hydroxybenzoate, which is a precursor of ubiquinol and / or menaquinol. According to aspects of the present invention, the levels of one or more downstream metabolites of chorismate may be modulated in subjects with or at risk for ASD.

[0054] Due to the absence of a shikimate pathway in humans, the levels of each molecule in the shikimate pathway and the chorismate-dependent pathway are dependent on either the amount of each molecule ingested through the diet or the amount produced by gut microbes.

[0055] Certain pharmaceutical compounds that target one or more enzymes in the shikimate pathway may be linked to ASD. One example of this is glyphosate (also known as Roundup). Glyphosate targets the enzyme EPSPS, preventing the production of 5-O-(carboxyvinyl)-3-phosphoshikimic acid, which can be converted to chorismate by chorismate synthase. One study observed that an increase in the number of children with autism corresponds to an increase in the amount of glyphosate applied to corn and soybeans. See Shaw, Elevated urinary glyphosate and clostridia metabolites with altered dopamine metabolism in triplets with autistic spectrum disorder or suspected seizure disorder: A case study, Integr Med (Encinitas), 16(1):50-57 (2017). In some cases, glyphosate may kill beneficial bacteria, leaving bacteria (e.g., Clostridia) that can produce harmful molecules (e.g., toxins). In some cases, glyphosate may compete with class I EPSPS enzymes rather than class II EPSPS enzymes.Thus, in some cases, subject exposure to glyphosate may result in a reduction in species that rely on class I EPSPS enzymes (e.g., in the gastrointestinal tract of the subject).In some embodiments, subject exposure to glyphosate may result in a reduced level of chorismate (e.g., in the gastrointestinal tract of the subject).

[0056] In some embodiments, exposure of a subject to glyphosate can result in the subject exhibiting one or more symptoms of ASD and / or having a higher risk of developing ASD. In certain embodiments, the method includes determining that the subject has or is at risk of having a subtype of ASD (e.g., the chorismate subtype of ASD) based on an elevated level of glyphosate in the subject (e.g., relative to one or more neurotypical individuals).

[0057] In some cases, a subject's exposure to glyphosate may contribute to non-Hodgkin's lymphoma, mature B-cell lymphoma, mature T-cell and natural killer cell lymphoma, celiac disease, gluten intolerance, overgrowth of pathogenic bacteria, reduced beneficial microbes, impaired serotonin signaling, infertility, miscarriage, birth defects, obesity, inflammatory bowel disease, anorexia nervosa, Alzheimer's disease, Parkinson's disease, multiple sclerosis, liver disease, ADHD, amyotrophic lateral sclerosis (ALS), cancer, and / or cachexia.

[0058] Aromatic amino acids (and their downstream products) In some embodiments, certain aromatic amino acids (eg, tryptophan, tyrosine, phenylalanine) are non-limiting examples of molecules in the chorismate metabolic pathway.

[0059] In some embodiments, chorismate can be converted to tryptophan via the tryptophan biosynthetic pathway. In the first step of the tryptophan biosynthetic pathway, anthranilate synthase can convert chorismate to anthranilate, which is an upstream metabolite (i.e., precursor) of tryptophan. In some cases, modulation of the tryptophan biosynthetic pathway can be associated with ASD in an individual. Thus, in certain cases, modulation of the levels of tryptophan and / or molecules in the tryptophan biosynthetic pathway (e.g., relative to neurotypical individuals) can be associated with ASD. In some cases, lower levels of tryptophan and / or molecules in the tryptophan biosynthetic pathway (e.g., relative to neurotypical individuals) can be associated with ASD. In some cases, higher levels of tryptophan and / or molecules in the tryptophan biosynthetic pathway (e.g., relative to neurotypical individuals) can be associated with ASD. Thus, in some embodiments, the method includes determining that the subject has or is at risk of developing at least one subtype of ASD (e.g., chorismate subtype) based on the level of tryptophan and / or molecules in the tryptophan biosynthetic pathway in a sample from the subject being modulated (e.g., increased, decreased) relative to the levels of one or more neurotypical individuals. The sample can be any sample described herein (e.g., fecal sample, gastrointestinal sample, serum sample, CSF sample).

[0060] In some cases, modulation of the levels of one or more downstream metabolites of tryptophan may be associated with ASD. Tryptophan may be further converted to downstream metabolites by at least three major pathways. In some cases, tryptophan may be converted to kynurenine or its derivatives by the kynurenine pathway. In some cases, tryptophan may be converted to serotonin or its derivatives by the serotonin pathway. In some cases, tryptophan may be converted to indole by the indole pathway. Each of these pathways may be of potential interest for ASD.

[0061] In the serotonin pathway, the enzyme tryptophan hydroxylase (TPH) converts tryptophan to 5-hydroxytryptophan (5-HTP), which is then converted to serotonin (5-HT) by the enzyme aromatic amino acid decarboxylase (AAAD). The neurotransmitter serotonin has a wide range of physiological functions in the brain and periphery through activation of its 5-HT receptors. Serotonin itself can then be converted into several molecules, including melatonin.

[0062] In some cases, individuals with ASD may have aberrant (e.g., higher, lower) levels of serotonin relative to one or more neurotypical individuals.In some cases, individuals with ASD may have higher levels of serotonin (e.g., in serum samples) relative to one or more neurotypical individuals.In some embodiments, the level of serotonin in a subject may be reduced by administering probiotics and / or fructooligosaccharides.

[0063] In some cases, individuals with ASD may have altered (e.g., higher, lower) levels of melatonin relative to one or more neurotypical individuals. In some cases, individuals with ASD may have lower levels of melatonin relative to one or more neurotypical individuals. In some cases, lower levels of melatonin may lead to sleep difficulties. In certain embodiments, the level of melatonin in a subject may be increased by administering melatonin supplements.

[0064] In the kynurenine pathway, tryptophan is converted to kynurenine by tryptophan dioxygenase or indoleamine dioxygenase, which can then be converted to several molecules, including, but not limited to, kynurenic acid, xanthurenic acid, quinolinic acid, picolinic acid, and hydroxyanthranilic acid.

[0065] In some cases, kynurenine levels may be increased in individuals with ASD. In the intestine, the levels of kynurenine and other related molecules are mediated by the activity of bacterial and host IDO1 enzymes. Kynurenine and its downstream metabolites (including kynurenic acid, xanthurenic acid, quinolinic acid and picolinic acid) have a variety of known biological processes, including neurotransmission, inflammatory immune response, and intestinal mucosal protective effects. Other downstream metabolites of kynurenine, including hydroxyanthranilic acid, have neurotoxic effects.

[0066] In some cases, an individual with an ASD may have altered (e.g., higher, lower) levels of kynurenine and / or one or more of its metabolites (e.g., downstream metabolites) relative to one or more neurotypical individuals. In certain cases, an individual with an ASD may have higher levels of kynurenine and / or one or more of its metabolites (e.g., downstream metabolites) relative to one or more neurotypical individuals. In certain cases, an individual with an ASD may have lower levels of kynurenine and / or one or more of its metabolites (e.g., downstream metabolites) relative to one or more neurotypical individuals.

[0067] Indoles are the third major category of molecules derived from tryptophan. In the case of indoles, the microbiota plays a major role in producing indoles. For example, in some cases, the bacterium Clostridium sporogenes converts tryptophan into the neurotransmitter tryptamine. In some cases, indoles can trigger signaling from the AhR, which can trigger various immune responses.

[0068] In some cases, an individual with ASD may have altered (e.g., higher, lower) levels of one or more of its metabolites (e.g., downstream metabolites) relative to one or more neurotypical individuals. In certain cases, an individual with ASD may have higher levels of one or more of its metabolites (e.g., downstream metabolites) relative to one or more neurotypical individuals. In certain cases, an individual with ASD may have lower levels of one or more of its metabolites (e.g., downstream metabolites) relative to one or more neurotypical individuals.

[0069] In some embodiments, chorismate can be converted to tyrosine and / or phenylalanine. In some cases, chorismate mutase can convert chorismate to prephenate, a precursor of tyrosine and phenylalanine. In some cases, modulation of tyrosine biosynthesis and / or phenylalanine biosynthesis can be associated with ASD in an individual. Thus, in certain cases, modulation of the levels of tyrosine, phenylalanine, and / or one or more of their metabolites (e.g., downstream metabolites) relative to one or more neurotypical individuals can be associated with ASD. In some cases, lower levels of tyrosine, phenylalanine, and / or one or more of their metabolites (e.g., downstream metabolites) relative to one or more neurotypical individuals can be associated with ASD. In some cases, higher levels of tyrosine, phenylalanine, and / or one or more of their metabolites (e.g., downstream metabolites) relative to one or more neurotypical individuals can be associated with ASD. Thus, in some embodiments, the methods include determining that the subject has or is at risk for developing at least one subtype of ASD (e.g., chorismate subtype) based on the level of one or more of tyrosine, phenylalanine, and / or their metabolites (e.g., downstream metabolites) in a sample from the subject being modulated (e.g., increased, decreased) relative to the levels of one or more neurotypical individuals. The sample can be any sample described herein (e.g., a fecal sample, a gastrointestinal sample, a serum sample, a CSF sample).

[0070] Thus, in some embodiments, levels of aromatic amino acids (and / or their downstream products) can be assessed in a sample from a subject and / or modulated in a subject to aid in the diagnosis and / or treatment of an ASD in the subject.

[0071] Folic acid In some embodiments, folic acid is a non-limiting example of a molecule in the chorismate metabolic pathway. In some cases, the enzyme aminodeoxychorismate synthase (also known as PABA synthase) converts chorismate towards folic acid biosynthesis. Folic acid (also known as vitamin B9) is required for DNA and RNA synthesis and amino acid metabolism. Lack of folic acid can lead to neural tube defects. For this reason, it has historically been supplemented in some transgenic foods. A portion of the human gut microbiome possesses genes that can produce and modify folic acid. There are 12 enzymes involved in converting chorismate to folic acid, with the activity of dihydrofolate reductase (DHFR) acting as the rate-limiting step. Folic acid is also dependent on vitamin B12 (cobalamin) for the conversion of N5-methyltetrahydrofolate to tetrahydrofolate (THF), the active form of folic acid. Cobalamin also plays an important regulatory role in folate and ubiquinone biosynthesis by acting as a cofactor for key transcription factors.

[0072] In some cases, the levels of folic acid, one or more of its derivatives (e.g., tetrahydrofolic acid), and / or one or more molecules involved in folic acid biosynthesis (e.g., cobalamin) are modulated (e.g., increased, decreased) in individuals with ASD compared to neurotypical individuals. In some cases, individuals with ASD may have increased circulating autoantibodies against folic acid receptors, which are required to transport folic acid through the blood-brain barrier. This may lead to reduced uptake of folic acid in the brain and downstream consequences related to ASD pathophysiology. In some cases, a method of treating an individual with ASD, comprising administering folic acid, one of its derivatives, and / or one or more molecules involved in folic acid biosynthesis, may alleviate one or more symptoms of ASD (e.g., language and communication difficulties).

[0073] Thus, in some embodiments, folate levels can be assessed in a sample from a subject and / or modulated in a subject to aid in the diagnosis and / or treatment of ASD in the subject.

[0074] Ubiquinol and Menaquinol In some embodiments, ubiquinol and menaquinol are non-limiting examples of molecules in the chorismate metabolic pathway. In some cases, chorismate can branch towards ubiquinol and menaquinol production when it is converted to p-hydroxybenzoate by chorismate lyase. These molecules are typically found in cell membranes and are part of the electron transport chain. Ubiquinone (also referred to as coenzyme Q) is utilized for aerobic respiration, while menaquinone is commonly used by bacteria for anaerobic respiration and lactic acid fermentation. Menaquinones are also referred to as vitamin K2 and can vary depending on the length of their isoprenoid carbon chain. Menaquinone-4 (MK4) is the main form of vitamin K in the brain, and serum concentrations of MK4 can be lower in children with ASD. Approximately 75% of MK4 is produced in the intestine, which is reportedly higher in males than females. Due to the role of MK4 in neurogenesis, there are several hypotheses that it may be important for ASD pathophysiology. There are also reports of a correlation between vitamin K levels and improved autism severity. In some of the same cases, modulated levels of ubiquinol and / or menaquinol production by the gut microbiota may have a disproportionate effect on male brain development.

[0075] In some cases, modulation of ubiquinol and / or menaquinol biosynthesis may be associated with ASD in an individual. Thus, in certain cases, modulation of the levels of one or more of ubiquinol (e.g., coenzyme Q, ubiquinol of a 6-13 carbon isoprenoid chain) and / or its metabolites (e.g., downstream metabolites) relative to one or more neurotypical individuals may be associated with ASD. In certain cases, modulation of the levels of one or more of menaquinol (e.g., menaquinone-4, menaquinol of a 6-13 carbon isoprenoid chain) and / or its metabolites (e.g., downstream metabolites) relative to one or more neurotypical individuals may be associated with ASD. In some cases, lower levels of ubiquinol, menaquinol, and / or its metabolites (e.g., relative to neurotypical individuals) may be associated with ASD. In some cases, higher levels of ubiquinol, menaquinol, and / or its metabolites (e.g., relative to neurotypical individuals) may be associated with ASD. Thus, in some embodiments, the method includes determining that the subject has or is at risk for developing at least one subtype of ASD (e.g., chorismate subtype) based on a level of ubiquinol, menaquinol, and / or metabolites thereof in a sample from the subject that is modulated (e.g., increased, decreased) relative to levels in one or more neurotypical individuals. The sample can be any sample described herein (e.g., a fecal sample, a gastrointestinal sample, a serum sample, a CSF sample).

[0076] Thus, in some embodiments, ubiquinol and menaquinol levels can be assessed in a sample from a subject and / or modulated in a subject to aid in the diagnosis and / or treatment of ASD in the subject.

[0077] Siderophore In some embodiments, siderophores are non-limiting examples of molecules in the chorismate metabolic pathway. A non-limiting example of a siderophore is enterobactin. In some cases, isochorismate synthase enzymes (e.g., EntC) convert chorismate to isochorismate in the first step toward producing enterobactin from chorismate. Siderophores are typically considered virulence factors in the battle between metal-loaded humans and bacteria. However, some studies have shown the beneficial role of enterobactin on the growth of C. elegans. This phenotype in C. elegans is linked to mitochondrial function, and mitochondrial dysfunction has been observed in children with ASD.

[0078] In some cases, the level of a siderophore (e.g., enterobactin) may be modulated (e.g., increased, decreased) in an individual with ASD relative to a neurotypical individual. In some cases, a lower level of a siderophore (e.g., enterobactin) may be associated with ASD relative to a neurotypical individual. In some cases, a higher level of a siderophore (e.g., enterobactin) may be associated with ASD relative to a neurotypical individual. Thus, in some embodiments, the method includes determining that the subject has or is at risk of developing at least one subtype of ASD (e.g., chorismate subtype) based on the level of a siderophore (e.g., enterobactin) in a sample from the subject that is modulated (e.g., increased, decreased) relative to the level in one or more neurotypical individuals. The sample may be any sample described herein (e.g., a fecal sample, a gastrointestinal sample, a serum sample).

[0079] Thus, in some embodiments, levels of siderophores can be assessed in a sample from a subject and / or modulated in a subject to aid in the diagnosis and / or treatment of ASD in the subject.

[0080] Chorismate Subtypes and Interventions Some embodiments are directed to methods and compositions for diagnosing and / or treating the chorismate subtype of ASD.

[0081] According to some embodiments, a method for assisting in the diagnosis of a subject having one or more symptoms of ASD and / or ASD comorbidities comprises determining the level of chorismate and / or a molecule in the chorismate metabolic pathway in a sample obtained from the subject.In some embodiments, the method further comprises determining that the subject has or is at risk of developing ASD (e.g., chorismate subtype of ASD) if the level of chorismate and / or a molecule in the chorismate metabolic pathway is higher or lower than a threshold value.

[0082] In some embodiments, the molecule in the chorismate metabolic pathway is the upstream or downstream metabolite of chorismate.Non-limiting examples of the upstream metabolite of chorismate include D-erythrose-4-phosphate, phosphoenolpyruvate, 3-deoxy-D-arabino-heptulosonic acid-7-phosphate, 3-dehydroquinic acid, 3-dehydroshikimic acid, shikimic acid, shikimic acid 3-phosphate, and 5-O-(carboxyvinyl)-3-phosphoshikimic acid.Non-limiting examples of the downstream metabolite of chorismate include aromatic amino acids (e.g., tryptophan, tyrosine, phenylalanine) and their metabolites (e.g., serotonin, melatonin, kynurenine, indole), folic acid and its derivatives (e.g., tetrahydrofolic acid), ubiquinol (e.g., coenzyme Q), menaquinol, and siderophores (e.g., enterobactin).

[0083] In some embodiments, determining the level of chorismate and / or molecules in the chorismate metabolic pathway comprises determining the abundance of one or more genes in the chorismate metabolic super-pathway present in a sample obtained from a subject. In certain embodiments, determining the abundance of one or more genes comprises amplifying at least a portion of one or more nucleic acids present in the sample to form a plurality of amplicons. The one or more nucleic acids can be amplified according to any nucleic acid amplification method known in the art (e.g., PCR, qPCR) to form a plurality of amplicons. In some embodiments, determining the abundance of one or more genes further comprises sequencing one or more amplicons of the plurality of amplicons. The one or more amplicons can be sequenced according to any nucleic acid sequencing method known in the art (e.g., long-read sequencing, short-read sequencing).

[0084] In some cases, determining the abundance of one or more genes in the chorismate metabolic super-pathway present in a sample obtained from a subject includes sequencing at least a portion of the 16S ribosomal RNA (rRNA) of the microorganisms present in the sample (e.g., performing 16S rRNA gene amplicon (16S) sequencing). In some cases, determining the abundance of one or more genes in the chorismate metabolic super-pathway present in a sample obtained from a subject includes performing shotgun metagenomic sequencing to sequence the metagenomic content of the sample.

[0085] In some embodiments, determining the level of chorismate and / or molecules in the chorismate metabolic pathway comprises measuring the level of chorismate and / or molecules in the chorismate metabolic pathway (e.g., using HPLC, LC-MS, GC-MS, and / or MALDI-TOF).

[0086] Some embodiments are directed to the method of treating the subject who has one or more symptoms of ASD and / or ASD comorbidities or is at risk of developing ASD.The examples of ASD comorbidities include but are not limited to obsessive-compulsive disorder (OCD), attention deficit disorder (ADD), attention deficit hyperactivity disorder (ADHD), epilepsy, schizophrenia, sleep disorder, gastrointestinal disorder, obesity, irritability, anxiety, depression, and eating disorder.

[0087] In certain embodiments, the subject has a chorismate subtype of ASD. In some cases, a sample obtained from a subject may include one or more molecular and / or microbial features of compositional abundance that indicate that the subject may be likely to respond to a chorismate-based intervention. In certain cases, the one or more molecular and / or microbial features include enterobacteriales, L-arginine degradation products, and / or enterobactin. In certain cases, the one or more molecular and / or microbial features include one or more metabolites of the chorismate metabolic superpathway, including, but not limited to, shikimate, chorismate, tryptophan, serotonin, melatonin, phenylalanine, tyrosine, folate (e.g., tetrahydrofolate), ubiquinol (e.g., coenzyme Q), and menaquinol. In some cases, a sample obtained from a subject may include one or more genes (e.g., one or more genes in the chorismate metabolic superpathway) of abundance that indicate that the subject may be likely to respond to a chorismate-based intervention.

[0088] In some embodiments, the method of treating a subject comprises regulating the level of chorismate and / or molecules in the chorismate metabolic pathway in the subject to alleviate one or more symptoms of ASD and / or ASD comorbidities or reduce the risk of developing ASD.In some embodiments, regulating the level comprises increasing the level.In some embodiments, regulating the level comprises decreasing the level.In some cases, the level is regulated in the intestine of the subject.

[0089] In some embodiments, modulating the level of chorismate and / or molecules in the chorismate metabolic pathway comprises administering to the subject a composition comprising chorismate, chorismic acid, and / or a salt or ester of either. In some embodiments, modulating the level of chorismate and / or molecules in the chorismate metabolic pathway comprises administering to the subject a composition comprising a chorismate prodrug. In some embodiments, modulating the level of chorismate and / or molecules in the chorismate metabolic pathway comprises administering to the subject a composition comprising a chorismate metabolite (and / or another molecule involved in the chorismate metabolic pathway). In certain embodiments, the composition comprises chorismate, chorismic acid, and / or a salt or ester of either, and one or more chorismate prodrugs. In certain embodiments, the composition comprises chorismate, chorismic acid, and / or a salt or ester of either, and one or more chorismate metabolites (and / or one or more other molecules involved in the chorismate metabolic pathway). In certain embodiments, the composition comprises one or more chorismate prodrugs and one or more chorismate metabolites (and / or one or more other molecules involved in the chorismate metabolic pathway). The one or more chorismate metabolites may include one or more upstream chorismate metabolites (e.g., shikimic acid) and / or one or more downstream chorismate metabolites (e.g., tryptophan, phenylalanine, tyrosine, serotonin, melatonin, indole, ubiquinol, menaquinol, e.g., menaquinone-4, tetrahydrofolate, enterobactin). The one or more other molecules involved in the chorismate metabolic pathway may include one or more enzymes (e.g., one or more chorismate branch point enzymes) and / or one or more molecules involved in the regulation of the chorismate metabolic pathway (e.g., adenosylcobalamin, cobalamin (also known as vitamin B12)).

[0090] In some embodiments, the composition may be provided as a pharmaceutical composition comprising one or more pharma- ceutically acceptable buffers, salts, additives, or other agents. In some embodiments, the composition (e.g., pharmaceutical composition) may be administered enterally, parenterally, or via any other suitable route. In some embodiments, the composition (e.g., pharmaceutical composition) may be administered orally, rectally, intravenously, intramuscularly, via inhalation, and / or topically. In some cases, the composition (e.g., pharmaceutical composition) may be in tablet form, gelatin capsule form, capsule form, liquid form, or any other form suitable for oral administration. In some cases, the composition (e.g., pharmaceutical composition) may comprise a coating or other system configured to provide a pH-dependent and / or time-dependent release of the active ingredient of the composition. In certain cases, the composition (e.g., pharmaceutical composition) may comprise a capsule comprising a pH-dependent colonic release coating. In some cases, the composition (e.g., pharmaceutical composition) is configured for intravenous administration.

[0091] In some embodiments, the composition (e.g., pharmaceutical composition) comprises a pharma- ceutically acceptable carrier (e.g., a diluent, adjuvant, excipient, or base with which the composition is administered). Such pharmaceutical carriers can be sterile liquids, such as water and oils, including those of petroleum, such as mineral oil, vegetable oil, such as peanut oil, soybean oil, and sesame oil, animal oil, or oil of synthetic origin. Saline solutions and aqueous dextrose and glycerol solutions can also be employed as liquid carriers.

[0092] In some embodiments, the composition (e.g., pharmaceutical composition) is administered to the subject in an effective amount, e.g., in an amount sufficient to alleviate (e.g., delay or reduce the onset, progression, and / or severity of) one or more symptoms of ASD and / or ASD comorbidities. In some embodiments, the composition (e.g., pharmaceutical composition) is administered to the subject in an amount sufficient to reduce the risk of the subject developing ASD. In some cases, regulating the level of chorismate and / or molecules in the chorismate metabolic pathway includes modifying the subject's diet to increase chorismate intake.

[0093] In certain embodiments, modulating the levels of chorismate and / or molecules in the chorismate metabolic pathway comprises administering to the subject a composition (e.g., a pharmaceutical composition) that modulates (e.g., increases, decreases) the expression and / or activity of one or more enzymes in the chorismate metabolic pathway. In some embodiments, the composition (e.g., a pharmaceutical composition) comprises an agonist, antagonist, activator, and / or inhibitor of one or more enzymes. In some embodiments, the one or more enzymes comprise a chorismate branch point enzyme. In some embodiments, chorismate mutase is modulated. In some embodiments, chorismate lyase is modulated (e.g., inhibited). In some embodiments, isochorismate synthase is modulated (e.g., inhibited). In some embodiments, aminodeoxychorismate synthase (also known as PABA synthase) is modulated (e.g., inhibited). In some embodiments, anthranilate synthase is modulated.

[0094] In certain embodiments, modulating the level of chorismate and / or molecules in the chorismate metabolic pathway comprises administering to the subject a composition (e.g., a pharmaceutical composition) that modulates (e.g., increases, decreases) the amount of microorganisms that produce chorismate and / or molecules in the chorismate metabolic pathway in the gastrointestinal tract of the subject. In some embodiments, the composition (e.g., a pharmaceutical composition) comprises one or more antibiotics, prebiotic compounds, and / or probiotic compounds. In certain cases, the one or more antibiotics comprise vancomycin, fidaxomicin, and / or metronidazole.

[0095] In some embodiments, the method for reducing the risk of developing ASD and / or treating one or more symptoms of ASD and / or ASD comorbidities comprises providing a microorganism that modulates the microbiome in a manner that mimics the typical profile of chorismate-related genes and / or restores the typical profile of chorismate-related genes.In some embodiments, providing a supplement of one or more molecules related to chorismate is useful for directing the microbiome toward the typical profile of chorismate-related genes, and thus can help treat one or more symptoms of autism.

[0096] The composition (e.g., pharmaceutical composition) can be administered to subjects of any age (e.g., children, adults). In some embodiments, the composition (e.g., pharmaceutical composition) can be administered chronically. In certain embodiments, the composition (e.g., pharmaceutical composition) can be administered for more than 30 days, 3 months, 6 months, 9 months, 1 year, 3 years, 5 years, 10 years, or 20 years. In certain embodiments, the composition (e.g., pharmaceutical composition) may be administered for a period of 1-3 months, 1-6 months, 1-9 months, 1 month-1 year, 1 month-3 years, 1 month-5 years, 1 month-10 years, 1 month-20 years, 3-6 months, 3-9 months, 3 months-1 year, 3 months-3 years, 3 months-5 years, 3 months-10 years, 3 months-20 years, 6-9 months, 6 months-1 year, 6 months-3 years, 6 months-5 years, 6 months-10 years, 6 months-20 years, 1-3 years, 1-5 years, 1-10 years, 1-20 years, 3-5 years, 3-10 years, 3-20 years, 5-10 years, 5-20 years, or 10-20 years. In some embodiments, the composition (e.g., pharmaceutical composition) is administered daily, weekly, monthly, or at shorter or longer time intervals. In some embodiments, the composition (e.g., pharmaceutical composition) is administered until one or more biomarkers indicative of ASD risk are reduced. In some embodiments, the composition (e.g., pharmaceutical composition) is administered in an amount sufficient to alleviate one or more symptoms of ASD and / or ASD comorbidities or to reduce the risk of developing ASD.

[0097] microorganisms According to some embodiments, a method for aiding in the diagnosis of a subject having one or more symptoms of ASD and / or ASD comorbidities includes determining the abundance of one or more microbial features (e.g., microbes) and / or their byproducts (e.g., toxins) in a sample obtained from the subject. In some embodiments, the method further includes determining that the subject has or is at risk for developing an ASD (e.g., the Sarcina subtype of ASD) if the abundance of the one or more microbial features (e.g., microbes) and / or their byproducts (e.g., toxins) is higher or lower than a threshold value.

[0098] Non-limiting examples of microorganisms include Sarcina, Prevotella (e.g., P. stercorea), Bifidobacteria, Actinobacteria, Bacteroidetes, Proteobacteria, Eggerthella lenta, Enterobacteriaceae, Aeromonas, Agrobacterium, Enterococcus, Actinobacillus, Lactobacillus, and / or Streptococcus. In certain embodiments, the microorganism is a Sarcina bacterium. In certain embodiments, the microorganism includes Sarcina ventriculi (also known as Clostridium ventriculi). In certain embodiments, the microorganism includes Clostridium perfringens. In some cases, a subject whose sample has elevated levels of Sarcina bacteria (e.g., relative to a neurotypical individual) can be said to have the Sarcina subtype of ASD.

[0099] Some embodiments are directed to a method of treating a subject having one or more symptoms of ASD and / or co-morbidities of ASD or at risk of developing ASD. In certain embodiments, the subject has a Sarcina subtype of ASD. In some embodiments, the method comprises modulating the level of one or more microorganisms (e.g., S. ventriculi and / or other Sarcina bacteria) in the subject to alleviate one or more symptoms of ASD and / or co-morbidities of ASD or reduce the risk of developing ASD. In some cases, the level of one or more microorganisms is modulated in the gut of the subject. In some embodiments, modulating the level of one or more microorganisms comprises decreasing the level. In certain cases, modulating the level comprises administering to the subject a composition comprising one or more antibiotics and / or phage therapy. In some embodiments, modulating the level comprises increasing the level. In certain cases, modulating the level of one or more microorganisms comprises administering to the subject a composition comprising one or more prebiotics and / or probiotics.

[0100] In some embodiments, the composition (e.g., a composition comprising one or more antibiotics, bacteriophages, prebiotic compounds, and / or probiotic compounds) may be provided as a pharmaceutical composition comprising one or more pharma- ceutically acceptable buffers, salts, additives, or other agents. In some embodiments, the composition (e.g., pharmaceutical composition) may be administered enterally, parenterally, or via any other suitable route. In some embodiments, the composition (e.g., pharmaceutical composition) may be administered orally, rectally, intravenously, intramuscularly, via inhalation, and / or topically. In some cases, the composition (e.g., pharmaceutical composition) may be in tablet form, gelatin capsule form, capsule form, liquid form, or any other form suitable for oral administration. In some cases, the composition (e.g., pharmaceutical composition) may include a coating or other system configured to provide a pH-dependent and / or time-dependent release of the active ingredient of the composition. In certain cases, the composition (e.g., pharmaceutical composition) may include a capsule comprising a pH-dependent colonic release coating. In some cases, the composition (e.g., pharmaceutical composition) is configured for intravenous administration.

[0101] In some embodiments, the composition (e.g., pharmaceutical composition) comprises a pharma- ceutically acceptable carrier (e.g., a diluent, adjuvant, excipient, or base with which the composition is administered). Such pharmaceutical carriers can be sterile liquids, such as water and oils, including those of petroleum, such as mineral oil, vegetable oil, such as peanut oil, soybean oil, and sesame oil, animal oil, or oil of synthetic origin. Saline solutions and aqueous dextrose and glycerol solutions can also be employed as liquid carriers.

[0102] In some embodiments, the composition (e.g., pharmaceutical composition) is administered to the subject in an effective amount, e.g., in an amount sufficient to alleviate (e.g., delay or reduce the onset, progression, and / or severity of) one or more symptoms of ASD and / or ASD comorbidities. In some embodiments, the composition (e.g., pharmaceutical composition) is administered to the subject in an amount sufficient to reduce the risk of the subject developing ASD.

[0103] The composition (e.g., pharmaceutical composition) can be administered to subjects of any age (e.g., children, adults). In some embodiments, the composition (e.g., pharmaceutical composition) can be administered chronically. In certain embodiments, the composition (e.g., pharmaceutical composition) can be administered for more than 30 days, 3 months, 6 months, 9 months, 1 year, 3 years, 5 years, 10 years, or 20 years. In certain embodiments, the composition (e.g., pharmaceutical composition) may be administered for a period of 1-3 months, 1-6 months, 1-9 months, 1 month-1 year, 1 month-3 years, 1 month-5 years, 1 month-10 years, 1 month-20 years, 3-6 months, 3-9 months, 3 months-1 year, 3 months-3 years, 3 months-5 years, 3 months-10 years, 3 months-20 years, 6-9 months, 6 months-1 year, 6 months-3 years, 6 months-5 years, 6 months-10 years, 6 months-20 years, 1-3 years, 1-5 years, 1-10 years, 1-20 years, 3-5 years, 3-10 years, 3-20 years, 5-10 years, 5-20 years, or 10-20 years. In some embodiments, the composition (e.g., pharmaceutical composition) is administered daily, weekly, monthly, or at shorter or longer time intervals. In some embodiments, the composition (e.g., pharmaceutical composition) is administered until one or more biomarkers indicative of ASD risk (e.g., levels of Sarcina bacteria) are reduced. In some embodiments, the composition (e.g., pharmaceutical composition) is administered in an amount sufficient to alleviate one or more symptoms of ASD and / or ASD comorbidities or to reduce the risk of developing ASD.

[0104] Proton Flux According to some embodiments, the method for assisting in the diagnosis of a subject with one or more symptoms of ASD and / or ASD comorbidities comprises determining the level of proton flux in the gastrointestinal tract of the subject from a sample obtained from the subject.In some embodiments, the method further comprises determining that the subject has or is at risk of developing ASD (e.g., proton flux subtype of ASD) when the level of proton flux is higher than a threshold value.

[0105] Some embodiments are directed to a method of treating a subject having one or more symptoms of ASD and / or comorbidities of ASD or at risk of developing ASD. In certain embodiments, the subject has a proton flux subtype of ASD. In some embodiments, the method comprises regulating the gastrointestinal pH of the subject to alleviate one or more symptoms of ASD and / or comorbidities of ASD or to reduce the risk of developing ASD. In some embodiments, regulating the pH comprises increasing the pH. In some embodiments, regulating the pH comprises decreasing the pH. In some embodiments, regulating the gastrointestinal pH comprises administering to the subject a composition or treatment configured to regulate the gastrointestinal pH of the subject. In some embodiments, the composition or treatment comprises one or more proton pump inhibitors. Non-limiting examples of proton pump inhibitors include omeprazole, lansoprazole, pantoprazole, rabeprazole, esomeprazole, dexlansoprazole, and derivatives thereof. In some embodiments, the composition or treatment comprises bismuth subsalicylate or a derivative thereof formulated for colonic release.

[0106] In some embodiments, the composition (e.g., a composition comprising one or more proton pump inhibitors and / or bismuth subsalicylate or derivatives thereof) may be provided as a pharmaceutical composition comprising one or more pharma- ceutically acceptable buffers, salts, additives, or other agents. In some embodiments, the composition (e.g., pharmaceutical composition) may be administered enterally, parenterally, or via any other suitable route. In some embodiments, the composition (e.g., pharmaceutical composition) may be administered orally, rectally, intravenously, intramuscularly, via inhalation, and / or topically. In some cases, the composition (e.g., pharmaceutical composition) may be in tablet form, gelatin capsule form, capsule form, liquid form, or any other form suitable for oral administration. In some cases, the composition (e.g., pharmaceutical composition) may include a coating or other system configured to provide a pH-dependent and / or time-dependent release of the active ingredient of the composition. In certain cases, the composition (e.g., pharmaceutical composition) may include a capsule comprising a pH-dependent colonic release coating. In some cases, the composition (e.g., pharmaceutical composition) is configured for intravenous administration.

[0107] In some embodiments, the composition (e.g., pharmaceutical composition) comprises a pharma- ceutically acceptable carrier (e.g., a diluent, adjuvant, excipient, or base with which the composition is administered). Such pharmaceutical carriers can be sterile liquids, such as water and oils, including those of petroleum, such as mineral oil, vegetable oil, such as peanut oil, soybean oil, and sesame oil, animal oil, or oil of synthetic origin. Saline solutions and aqueous dextrose and glycerol solutions can also be employed as liquid carriers.

[0108] In some embodiments, the composition (e.g., pharmaceutical composition) is administered to the subject in an effective amount, e.g., in an amount sufficient to alleviate (e.g., delay or reduce the onset, progression, and / or severity of) one or more symptoms of ASD and / or ASD comorbidities. In some embodiments, the composition (e.g., pharmaceutical composition) is administered to the subject in an amount sufficient to reduce the risk of the subject developing ASD.

[0109] The composition (e.g., pharmaceutical composition) can be administered to subjects of any age (e.g., children, adults). In some embodiments, the composition (e.g., pharmaceutical composition) can be administered chronically. In certain embodiments, the composition (e.g., pharmaceutical composition) can be administered for more than 30 days, 3 months, 6 months, 9 months, 1 year, 3 years, 5 years, 10 years, or 20 years. In certain embodiments, the composition (e.g., pharmaceutical composition) may be administered for a period of 1-3 months, 1-6 months, 1-9 months, 1 month-1 year, 1 month-3 years, 1 month-5 years, 1 month-10 years, 1 month-20 years, 3-6 months, 3-9 months, 3 months-1 year, 3 months-3 years, 3 months-5 years, 3 months-10 years, 3 months-20 years, 6-9 months, 6 months-1 year, 6 months-3 years, 6 months-5 years, 6 months-10 years, 6 months-20 years, 1-3 years, 1-5 years, 1-10 years, 1-20 years, 3-5 years, 3-10 years, 3-20 years, 5-10 years, 5-20 years, or 10-20 years. In some embodiments, the composition (e.g., pharmaceutical composition) is administered daily, weekly, monthly, or at shorter or longer time intervals. In some embodiments, the composition (e.g., pharmaceutical composition) is administered until one or more biomarkers indicative of ASD risk (e.g., elevated proton flux) are reduced. In some embodiments, the composition (e.g., pharmaceutical composition) is administered in an amount sufficient to alleviate one or more symptoms of ASD and / or ASD comorbidities or to reduce the risk of developing ASD.

[0110] Techniques and systems for assessing a subject's risk of having or developing ASD In some embodiments, once one or more characteristics (e.g., abundances of molecular and / or microbial features) are determined from a sample from a subject, the one or more characteristics can be analyzed by a risk analysis means to assess the subject's risk of having or developing an ASD and / or to identify a treatment to be provided to the subject. The risk analysis means can be implemented in software, executable instructions executed by, for example, one or more servers, one or more laboratory devices (e.g., laboratory devices that receive, prepare, and analyze samples such as fecal samples), one or more personal or mobile computing devices, or other devices.

[0111] In some such embodiments that operate by risk analysis means, the analysis of the subject's risk of having or developing ASD and / or the analysis for identifying treatment can be performed in some embodiments using one or more rules.For example, when a rule or a combination of rules is met, the subject can be determined to have a higher or lower risk of ASD, or can be found to be a candidate or not for a certain treatment.Such rules can in some cases relate to the level of one or more molecules and / or types of microorganisms detected in the subject's microbiome, or the amount or relative amount of molecules and / or microorganisms.As one particular example, if the subject's stool sample demonstrates a certain relative abundance of a certain microorganism in the stool sample, the patient can be found to be at increased risk of ASD.

[0112] In some embodiments, in addition to or as an alternative to the rule-based analysis, the risk analysis means may utilize a trained machine learning model to assess the risk of a subject having or developing ASD or to identify a treatment to be provided to the patient. In some embodiments implementing a machine learning model, such a machine learning model may be trained based on one or more characteristics (e.g., abundance of molecular and / or microbial features) of a population of people diagnosed with ASD and a population of people not diagnosed with ASD. As a result of the training, it may learn a relationship between a sample characteristic (e.g., abundance of molecular and / or microbial features) and the presence or absence of ASD and / or the success or lack of success of certain characteristics. In some such embodiments, once one or more characteristics are obtained by analyzing a sample, the one or more characteristics may be input into a machine learning model. In some such embodiments, the machine learning model may respond by outputting a classification of the subject's microbiome or the subject into a particular class of a set of classes. Each class may be associated with a risk or a particular risk level of having or developing ASD, and / or being or not being a candidate for a particular treatment. In other embodiments, each class may additionally or alternatively relate to a subtype of ASD. Here, each subtype of ASD corresponds to a risk level, the type of ASD that the subject may have or develop, the probable combination of symptoms that the subject may experience, the severity of symptoms, or other characteristics of the ASD condition that the subject may have or be at risk of developing.In some embodiments, rather than outputting a single class, the machine learning engine may output a set of one or more classes and, for each class, a confidence level associated with the class indicating the likelihood calculated by the machine learning engine that the subject will be correctly classified into that class.The likelihoods may then be evaluated by the risk analysis means, respectively, to identify one or more of the classes that the subject will be classified into.For example, the class with the highest likelihood may be identified as the class that the subject falls into.As another example, only those classes with a likelihood above a first threshold may be further analyzed and a decision may be made among the classes that are the class(es) with which the subject will be identified to match, e.g., the most likely may be identified, or one or more classes that exceed a threshold amount may be more likely than other classes.

[0113] In aspects using a risk analysis means, once a subject is identified by the risk analysis means as meeting some criterion(s) (e.g., by rules analysis, machine learning analysis, e.g., classification, or another analysis), the risk analysis means may then initiate a diagnostic or treatment workflow consistent with the criterion(s). For example, the workflow may include outputting a determination of met criterion(s) to a clinician (e.g., a laboratory technician, doctor, nurse, etc.) and associating risk, diagnosis, treatment recommendation, or other information associated with the satisfaction of those criterion(s). As another example, the workflow may include prescribing a treatment, e.g., initiating a prescription order for a treatment associated with the criterion(s) or the risk or treatment recommendation. As another example, the workflow may include ordering a follow-up analysis, such as developmental monitoring or screening, or other evaluation by a physician (e.g., a psychologist, neurologist, or other physician) to confirm the risk or treatment recommendation or to obtain additional information about the risk to inform the treatment plan. In the embodiment of identifying the treatment to be provided to the subject to reduce the risk of developing ASD or ASD comorbidities or to ameliorate or alleviate symptoms, various suitable treatments can be utilized. In some embodiments, the treatment can be related to a microbial feature, such as a microbial feature used to assess the subject's risk of having or developing ASD. Such treatment can include a treatment to introduce one or more specific microorganisms or one or more compounds (e.g., drugs, medicines, etc.) into the patient's microbiome to promote the growth of a specific microorganism. Such treatment to promote growth can include administration of probiotics or prebiotics. Such treatment can also or alternatively include a treatment to remove microorganisms from the subject's microbiome or to attempt to prevent the growth of a specific microorganism, for example, by antibiotic or phage treatment.Such treatments may further, additionally or alternatively, include procedures to introduce or increase the prevalence or remove or decrease the prevalence of microbial molecules or other compounds associated with the microorganism (e.g., beneficial compounds produced or derived from the microorganism, or toxins or harmful or non-beneficial compounds produced or derived from the microorganism). Treatments based on microbial characteristics may include treatments to encourage the patient's microbiome to become more similar to the target microbiome, for example, by having a microbial composition that is more similar to the composition of the target microbiome than the patient's current microbiome composition. EXAMPLES

[0114] Example 1 In this example, we conducted a quantitative meta-analysis of existing ASD microbiome data. We assessed the effect that 16S ribosomal RNA gene amplicon (16S) sequencing method, gender, geographic location, control type, and age have on detecting differences in the ASD microbiome, as well as how these factors may influence the outcome of machine learning prediction of ASD status. Data were derived from 13 published case-control cohorts and one curated cohort from the American Gut Project (AGP). Cohorts were analyzed individually and as a whole to draw conclusions about how the ASD gut microbiome differs from developmentally and neurotypical individuals.

[0115] Included studies and sample demographics Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA), we curated a dataset consisting of 14 different case-control cohorts. The resulting dataset had a total of 1,740 samples, of which 888 were diagnosed with ASD and 852 were neurotypical controls. The largest cohort was manually curated (n=524) from participants from the AGP, along with other large cohorts from Dan et al. (n=286) and Chen 2020 (n=246). Control subjects were selected from the AGP in a manner that rigorously controlled for technical variability, finding the best match based on various selected variables. This approach resulted in a control cohort that largely reflected the demographics of ASD subjects from the AGP. In total, the majority of the sample was from males. This follows the general trend of increased ASD rates in males. Studies typically included participants between the ages of 2 and 6 years. Notable exceptions are Kong et al. and Kang et al., which encompassed older age groups. Thirteen of the cohorts represented samples from the United States (n=1079) and China (n=866), and one study, Zurita et al., 2019, was conducted in Ecuador. In addition, a large number of samples (n=1280) were processed targeting the 16S V4 hypervariable region, 357 samples were sequenced from the 16S V3-V4 region, and 103 were processed from the V1-V2 region. Most studies utilized age- and sex-matched samples from unrelated controls, but two studies utilized siblings of children with ASD and two studies utilized mothers.

[0116] Assessing differences in ASD microbiome diversity To address the question of whether microbiome diversity is modulated in ASD, multiple alpha and beta diversity metrics were evaluated in both per-study and aggregated dataset approaches. For beta diversity, two commonly utilized distance metrics, unweighted UniFrac and weighted UniFrac, were tested. These metrics are phylogenetically informed metrics that were not used in all of the original reports of the datasets. For aggregation, significant clustering by study was observed for both unweighted and weighted UniFrac distance metrics, but the effect was noticeably stronger for unweighted UniFrac distance (unweighted UniFrac PERMANOVA P<0.001, pseudo F=42.15; weighted UniFrac PERMANOVA P<0.001, pseudo F=18.54) (Figures 3A-3B). This observation may indicate that low abundance study-specific microbes may increase study separation and drive study divergence. Although less striking than the study effect, significant clustering by ASD status was observed for unweighted UniFrac distances (PERMANOVA P<0.001, pseudo F=3.99) but not for weighted UniFrac distances (PERMANOVA P=0.162, pseudo F statistic=1.48). Strong separation was also observed in relation to the 16S hypervariable region used, highlighting the importance of taking into account which variable regions were used. Differences related to sequencing depth and geographic location in unweighted UniFrac distances were also observed, but these results were confounded by the effects of variable region and study.

[0117] Given the strong confounding effect on beta diversity results driven by which study the samples came from, each cohort was further analyzed individually. Eight of the 14 cohorts showed statistically significant differences in the ASD gut microbiome compared to neurotypical controls from either weighted or unweighted UniFrac distance metrics (PERMANOVA P<0.05) (Figure 3C). These results indicated that although there is variability in detecting differences in beta diversity between ASD and controls, it can be detected in the majority of studies and when analyzing studies in aggregate.

[0118] Each study was also analyzed for the effect of ASD status on microbial diversity within the samples. Four alpha diversity metrics were utilized: Faith's phylogenetic diversity, Shannon's diversity, evenness, and observed features. For all metrics, the study effect was large, requiring analysis of each of the 14 cohorts individually. For each of the four alpha diversity metrics utilized, there were at least three studies showing a statistically significant difference in alpha diversity values ​​between ASD and neurotypical controls (Student's t-test, P<0.05) (Figure 3D). Seven of the 14 cohorts showed a statistically significant difference in alpha diversity between ASD and neurotypical controls in at least one of the four metrics (Figure 3D). In general, studies showed mildly reduced alpha diversity in ASD subjects (median ASD / control differences in observed features = +0.04%, Even = -2.5%, Shannon = -0.09%, and Faith = -1.2%). These results suggested an unclear association between ASD and alpha diversity. Possible reasons for this difference are described later in this example.

[0119] Differences in microbial relative abundance between ASD and controls Differences in microbial relative abundance between ASD and control subjects were assessed at a per-study level and data were aggregated at different phylogenetic levels. Few phyla were consistently significantly different. The most commonly significant phylum was Bacteroidetes, with four studies reaching statistical significance (Figure 4A). Although statistical significance was not reached, the majority of cohorts showed increased Actinobacteria and Proteobacteria (Figure 4A). Specific phylum ratios were also examined. The ratio of Firmicutes to Bacteroidetes was found to be elevated in ASD individuals in Berding et al., 2020, Cao et al., 2021, Dan et al., 2020, David et al., 2021, Huang et al., 2021, Fouquier et al., 2021, Kong et al., 2019, Liu et al., 2019, and Son et al., 2015 (Figure 4A). In contrast, the ratio of Firmicutes to Bacteroidetes was decreased in Zou et al., 2021, Chen et al., 2020, and Kang et al., 2017 (Figure 4A). No significant difference in this ratio was found between the US and Chinese samples.

[0120] The most common genus with significant differences in its relative abundance was Prevotella, for which five cohorts reached statistical significance (unadjusted P<0.05) (Figure 4B). However, studies were divided on whether Prevotella was increased or decreased in ASD subjects (Figure 4B). Common ASD associations were also inconsistently increased or decreased at the class and order level, but most studies showed increased abundance of the class Enterobacteriales in ASD subjects.

[0121] When the data were analyzed in aggregate, few taxa also generalized across the dataset. A few notable exceptions included the class Actinobacteria, which was significantly higher in children with ASD compared to controls (Wilcoxon t-test P=0.0014, Fig. 4C). Despite inconsistent study-to-study findings for Prevotella, a general trend was found pointing to a reduction in ASD samples compared to controls (Wilcoxon t-test, P=0.044, Fig. 4D). One of the strongest findings in the aggregate data confirmed a previous association with increased Bifidobacterium in ASD subjects (Wilcoxon t-test, P=0.00015, Fig. 4E).

[0122] Chorismate links several microbial pathways commonly modulated in ASD Despite taxonomic inconsistency between studies, functional pathways encoded by the microbiomes of ASD subjects may be more generalizable. To test this possibility, metabolic pathway abundances were assessed for each sample and compared for ASD and control subjects. After assessing the most commonly significant pathways, two major clustering patterns of pathways commonly increased versus commonly decreased in ASD subjects were observed (Figure 4F). Of particular interest, several of the commonly identified pathways shared a common molecular precursor, chorismate.

[0123] Chorismate is a product of the shikimate pathway, which is absent in humans. Chorismate can be converted into several important molecules, including menaquinol and ubiquinol. Eighteen pathways involving the biosynthesis of various menaquinols and ubiquinols that were significant in three or more studies were identified (Figure 4G). These pathways included the biosynthesis of menaquinols 6-13. These were greatly increased in ASD, except for Zou et al., which showed a significant decrease in ASD subjects (Figure 4G).

[0124] Menaquinone and ubiquinone are essential molecules involved in bacterial anaerobic and aerobic respiration, respectively, and can be synthesized from chorismate.

[0125] Aromatic amino acids (AAA) are also products originating from chorismate. From the analysis, L-tryptophan biosynthesis was significantly modulated in four studies (P<0.05) and increased in ASD subjects in nine of the fourteen studies (Figure 4F). However, pathways involving other aromatic amino acids L-tyrosine and phenylalanine were less commonly significant, while L-tyrosine degradation was increased in ASD in eight of the fourteen cohorts (Figure 4H). It was also observed that pathways involving other amino acids, such as L-threonine biosynthesis and L-glutamate and L-glutamine biosynthesis, were significantly decreased in five cohorts each (Figure 4F). These results support the previously noted differences in amino acid metabolism in ASD patients and extend the analysis to new cohorts.

[0126] Other molecules derived from chorismate include siderophores and folate, which were also found to be modulated in several ASD studies. Enterobactin is a metal-chelating siderophore that was identified to be significantly modulated in Dan et al., Huang et al., and Zou et al. (Figure 4H). Tetrahydrofolate biosynthesis was further identified to be significantly decreased in four studies (Figure 4H). In addition, adenosylcobalamin biosynthesis (also known as vitamin B12) was decreased in 10 of 14 cohorts, with four cohorts reaching concordant significance (Figure 4F). Vitamin B12 catalyzes the conversion of homocysteine ​​to methionine and tetrahydrofolate, which serves as a rate-limiting step for folate levels.

[0127] Evaluating machine learning performance for predicting ASD status The relationship between the microbiome and ASD is nonlinear and may not be likely to be identified by reductionist approaches. Thus, various machine learning approaches were utilized to assess whether the ASD gut microbiome is discriminable from that of neurotypical individuals. Performance and accuracy in predicting ASD status was evaluated using 11 different classification algorithms on an aggregate of 14 datasets (Figure 5A). All 11 models performed at least slightly better than chance as determined by having an area under the receiver operating characteristic curve greater than 0.5 (Figure 5A). The top performing models, the gradient boosting classifier (GBC) and the Adaboost classifier (ABC), were able to discriminate ASD individuals from controls with an area under the curve (AUC) greater than 0.6 (GBC; AUC = 0.62 ± 0.03, ABC; AUC = 0.61 ± 0.02) (Figure 5A). Using the F1 score of each model to evaluate both precision and recall, the GBC and ABC models were found to perform at 0.62 ± 0.03 and 0.62 ± 0.02, respectively. The performance of each model was evaluated by assessing the classification accuracy rate of each model. Both the GBC and ABC models reported the highest percent accuracy rate of 62% ± 0.03 and 61% ± 0.02, respectively. For all metrics, GBC performed best in classifying ASD status and was therefore selected for downstream analysis.

[0128] Since the initial machine learning model comparisons were performed on amplicon sequencing variants (ASVs), it was of interest to see how taxonomic resolution impacted model performance and accuracy. A randomized train-test split of 50 iterations was performed utilizing the GBC model with read count data either aggregated or uncollapsed at seven different taxonomic levels (kingdom, phylum, class, order, family, genus, species, ASV). Increasing the taxonomic resolution improved the model accuracy, with only a slight gain past the genus level classification (Figure 5B). A similar trend was observed when performing the same analysis with both Adaboost and decision tree classifiers. When trained and tested on each dataset individually, the GBC model performed better than chance (AUC>0.5) on all datasets except three, with five having an AUC above 0.70 (Figure 5B). These results indicate that the ASD gut microbiome may contain information that would be beneficial in assessing ASD status.

[0129] We next considered the important features in discriminating individuals with ASD from controls. The top three features from the GBC models trained and tested on the aggregated ASV-level dataset were ASVs from the genus Sarcina (family Clostridiaceae), Eggerthella lenta, and Prevotella stercorea (Figure 5C). Among the top 25 features, the increased relative abundance of ASVs from Sarcina, Pseudomonas, and Uruburuella suis was important for ASD classification, while the increased abundance of ASVs from the class TM7-3, phylum Bacteroidetes, Coporobacillus cateniformis, and Clostridium clostridioforme was important for control classification (Figure 5C). To provide more context to the findings, the top 25 features from the aggregated dataset were further evaluated for their importance to predicting ASD status within the GBC models generated with each individual dataset. Hierarchical clustering of the importance scores of these features for each dataset revealed two main groupings (Figure 5D). Of particular interest was the group of features that contained similarly high importance scores for all datasets. This group encompassed ASVs from the genera Eggerthella lenta, Sarcina, Bilophila, Veilonella parvula, and Megasphaera. These top features highlight that despite heterogeneity of results across studies, there are several microbial features that can be consistently utilized for ASD classification.

[0130] Assessing the impact of study design on study outcomes To address the variability in ASD microbiome results reported in the literature and highlighted in this example, we sought to evaluate how study design factors may impact the findings of ASD microbiome studies. Some of these factors include, but are not limited to, 16S sequencing method, geographic location, sex, age, and relatedness of control subjects to ASD subjects. We evaluated the effect of different technical options for 16S sequencing method, sequencing depth of the 16S hypervariable region, and choice.

[0131] Of the 16S hypervariable regions included in the data, one study targeted the V1-V2 region, five studies utilized the V3-V4 region, and eight studies used the V4 region. Due to the limited number of studies conducted utilizing that variable region, V1-V2 samples were excluded from this analysis. It was found that studies utilizing the V3-V4 variable region had a significantly higher relative abundance of the genus Prevotella in ASD subjects (P=4.61E-3), while studies targeting the V4 region had a significantly lower relative abundance of this genus in ASD subjects (P=5.6E-4) (Figure 6A). This was also consistent when comparing the ASD / control relative abundance ratios for Prevotella from a study-by-study perspective (P=0.011) (Figure 6A). Also, the ratio of the genus Prevotella to the phylum Bacteroidetes, order Desulfovibrionales, and class Deltaproteobacteria was found to differ between the V3-V4 and V4 regions (Wilcoxon t-test per study P = 0.017, 0.011, and 0.011, respectively) (Figures 6B-6D). Contrary to per study, when analyzing the data aggregated, it was noted that significant differences could be found for Desulfovibrionales and Deltaproteobacteria in the V3-V4 studies, but no differences were found in the V4 samples (both were reduced in the V3-V4 ASD with P = 0.015).

[0132] Despite identifying differences between taxonomic associations with ASD by sequenced variable region, machine learning models trained on both datasets performed well above chance, with the V3-V4 data outperforming the V4 data (average AUC of 0.65 and 0.70 for the V4 and V3-V4 models, respectively) (Figure 6E). These results suggest that there are notable taxonomic differences that can occur depending on the 16S hypervariable region sequenced, however, machine learning models built for ASD classification are feasible when utilizing either the V4 or V3-V4 hypervariable regions.

[0133] A split in the dataset was noted regarding sequencing depth, where six studies were sequenced at a relatively low sequencing depth (sequencing around 6,000 reads per sample) and eight at a higher sequencing depth (sequencing a minimum of 14,000 reads per sample). As assessed by AUC, accuracy rate, and F1 statistic, there was a noticeable difference in the performance of models trained on samples from a cohort with a minimum of 14,000 reads per sample versus samples from a cohort with a minimum of 6,000, as shown in Figure 6E. Models trained on data from studies using high sequencing depth had an average AUC of 0.75, whereas studies with low sequencing depth had an average AUC of 0.59. However, from a taxonomic perspective, no significant effect of sequencing depth on the difference in relative abundance of ASD to controls was found. These results suggest that increased sequencing depth may be important for accurate ASD classification, but does not impact taxonomic outcomes.

[0134] We next assessed how aspects related to the demographics of study participants may modulate the ASD microbiome results. For the effect of geographic location, the cohort was split by which country each study was from. No statistically significant differences were found in the relative abundance ratios of ASD vs. controls from US vs. Chinese studies. However, striking differences were found in the ability of the GBC model to classify ASD and control samples depending on the country (Figure 6F). The model was found to perform better in Chinese samples, with a mean AUC of 0.78 versus 0.6 AUC in US samples. These results suggest that geographic location may be an important factor in ASD classification using the gut microbiome.

[0135] Some studies utilized age- and sex-matched controls, while others used household controls such as siblings or mothers. No statistically significant differences were found in the relative abundance ratios of ASD vs. controls from related vs. unrelated individuals as controls. A striking difference was observed in the ability of the GBC model to classify ASD and controls from cohorts utilizing unrelated controls compared to related controls (Figure 6F). The model performed better in samples from the cohort with unrelated controls, with a mean AUC of 0.71, compared to the cohort with related controls, which had a mean AUC of 0.54. This indicates that there may be greater differences in the microbial profiles of ASD individuals compared to controls when the controls are unrelated to the ASD individuals.

[0136] We also evaluated how gender affects the microbial profiling and diagnostic accuracy of the machine learning model for ASD and control individuals. Samples were divided by gender, and the AUC of the model for predicting ASD status in females was compared to males (Figure 6F). A greater accuracy rate was observed for predicting ASD status from male samples, with a mean AUC of 0.67 compared to an AUC of 0.55 from female samples (Figure 6F). Taxonomic differences between ASD vs. controls on a study-by-study basis were not examined due to the limited number of females in each dataset. However, the difference in the prediction accuracy rate of ASD status from male samples compared to females may indicate the influence of gender on microbial profiling and subsequent prediction accuracy.

[0137] The microbial composition of the gut microbiome is known to transition during human development. Therefore, we assessed how the age of the sample affects the ability to predict ASD status. To do this, we subsetted the data with all possible age ranges from 1 to 16 years and compared the performance of the GBC model in predicting ASD status (Figure 6G). We observed that samples from ages 2 to 7 years obtained the greatest prediction accuracy rate and that the AUC of the model decreased concomitantly as the age of the individuals increased (Figure 6G). To verify that the findings were not simply due to the corresponding sample size available in each age range, we first determined the effect of sample size on the GBC model performance and then plotted the corresponding sample size. We observed that samples from younger subjects performed better than expected by the sample size in predicting ASD status.

[0138] Branch-point enzymes of chorismate metabolism In addition to assessing differences in the pathways derived from chorismate listed above, we assessed whether six branch point enzymes of chorismate metabolism were modulated in the ASD gut microbiome. Chorismate lyase, the enzyme that constrains chorismate in the direction of ubiquinol biosynthesis, was observed to be upregulated in the ASD gut microbiome in 11 of 14 cohorts, and of these, two reached statistical significance (Figure 7A). It was observed that there was a significantly greater difference in the ASD gut microbiome compared to controls for the branch point enzyme chorismate lyase than for any of the other branch point enzymes. To see if ASD samples had a greater number of chorismate lyase enzymes than their control counterplots, normalized enzyme counts were calculated for each sample and the counts for ASD individuals and controls were compared for each study (Figure 7B). It was shown that five cohorts showed statistically significant differences in the predicted number of chorismate lyase enzymes in the ASD gut microbiome (Figure 7B).

[0139] Finally, one of the most common significant pathways involving chorismate was the biosynthesis of various menaquinols. We wanted to explore whether this could be due to an increase or decrease in the branch point enzyme responsible for constraining chorismate to the production of menaquinol, the menaquinol-specific isochorismate synthase. Isochorismate synthase has been demonstrated to convert chorismate to both enterobactin and menaquinol biosynthesis and is classified under the same enzyme number (EC 5.4.4.2). We further explored isochorismate synthase menF, which is found in several bacterial species and constrains chorismate to menaquinol biosynthesis. We examined the predicted gene count of menF to assess whether it was more or less enriched in ASD fecal samples compared to controls. Three studies showed statistically significant differences in the predicted gene abundance of menF (Figure 7C). Although not reaching significance, the majority of studies showed an increase in menaquinol-specific isochorismate synthase in the ASD gut microbiome (Figure 7C), a trend comparable to that observed for the entire menaquinol biosynthetic pathway mentioned earlier (Figure 4H).

[0140] The next goal was to identify which bacteria contained the menF gene. Bacteria containing the menF gene from each cohort were identified. Heat maps of the bacteria were created to assess their prevalence in each cohort (Figure 7D). Four of the bacteria with the menF gene were important bacteria for the ASD classification of controls from the machine learning model (Figure 5C). The increased importance of these bacteria for the machine learning classification model further elucidated that chorismate, metabolites related to chorismate, or chorismate-converting enzymes are important in the ASD gut microbiome and may represent promising therapeutic targets.

[0141] Discussion of results In this example, we took a systematic approach to reanalyzing existing ASD gut microbiome data and provide a thorough report of commonalities and differences between datasets. We tested the extent to which aspects of study design affect the taxonomic findings and accuracy rates of machine learning models designed for ASD classification. Due to the various choices in different data processing approaches and the variability these choices introduce, it was important to reanalyze the data using a consistent methodology. To address differences in microbiome alpha diversity, beta diversity, and relative abundance at different taxonomic levels, we reexamined the common question of how ASD microbiomes differ from control individuals across datasets. We found mixed results on whether differences in alpha diversity were observed, with somewhat more consistency in differences in beta diversity in ASD individuals compared to controls. We extended what has been commonly reported by analyzing data based on factors such as variable region, control relationship, age, sex, and geographic location, as well as by utilizing machine learning models to estimate the impact of these factors on classification of ASD status using microbiome data.

[0142] This example contextualizes the ASD microbiome technology field by evaluating how both technological choices and cohort demographics modulate the results. Microbial composition is known to vary among different human populations and based on technological choices. Of note, significant differences in the diagnostic performance of machine learning models based on data from US cohorts vs. Chinese cohorts were observed. One possible explanation is the collection of more severe ASD cohorts in China than in the US. Supporting this are reports suggesting that the prevalence of ASD in China may be underestimated and that the true prevalence rate may be closer to that of the US. This may indicate that children with ASD diagnoses in China may represent more severe cases than their Western counterparts. Alternatively, the results may also be a result of different countries representing populations with different socioeconomic status, diet, living conditions, and environmental factors.

[0143] The increased ability to predict ASD status at younger ages was also found to be particularly interesting. This suggests that developing diagnostic screening tools based on the microbiome may facilitate earlier diagnosis, which could be of great interest given the limited diagnostic tools available for ASD to date. It is also possible that these results represent a critical window of age in ASD pathology. However, many confounding factors exist, including bias in the study from which the sample was derived, or the potential that ASD cases aged 2-7 years represent more severe cases than those in older children.

[0144] The most important factor for study differences in taxonomic associations with ASD was found to be the choice of 16S hypervariable region. Discordant results were found for the ratio of Prevotella to Bacteroidetes in ASD vs. controls in the V3-V4 cohort versus the V4 cohort, the genus Prevotella, the order Desulfovibrionales, and the class Deltaproteobacteria, both in the study and in the sample relative abundance. It was particularly interesting that Prevotella appeared to be increased in ASD in the V4 dataset, whereas it was decreased in ASD in the V3-V4 cohort. The V3-V4 region represents a larger portion of DNA (~300bp vs. ~150bp in these studies), so one possible explanation for this could be the different taxonomic assignments of reads based on longer or shorter read lengths.

[0145] Despite the large study variability, several consistent trends in the ASD gut microbiome were identified across cohorts. The greatest taxonomic consistency was the enrichment of Actinobacteria at both the phylum and class level, as well as increased Bifidobacterium and a general decrease in Prevotella in ASD individuals. The genus Bifidobacterium (phylum Actinobacteria) is important for early human development in life, but it is subsequently replaced by obligate anaerobic bacteria. In some cases, the increased abundance of Bifidobacterium may be an indicator of an underdeveloped gut microbiome.

[0146] Several consistencies between predicted functional pathways encoded by the ASD gut microbiome were also found. Chorismate is a common branch point for the synthesis of several biologically relevant molecules including AAA, menaquinone, ubiquinone, enterobactin, and folate. Several pathways implicated in these downstream metabolites were found to be modulated in the ASD gut microbiome. Interestingly, chorismate biosynthesis is dependent on the enzyme 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS), which is a target of glyphosate. In some cases, glyphosate levels may be linked to ASD rates and a modulated microbiome.

[0147] method Literature search strategy and selection criteria To identify applicable studies evaluating autism spectrum disorders and the gut microbiome, a systematic literature search of the PubMed database from inception to July 15, 2021 was performed. The following search terms were used as free text words only or as keywords: ASD (or autism spectrum disorder or autism) and (gut microbiome or gut microbiota or fecal microbiome). No other search features or advanced language restrictions were used. In this meta-analysis, articles were included if (a) fecal microbiome samples of autistic patients and / or those of neurotypical individuals were taken; (b) the data and respective metadata were retrievable and interpretable; and (c) an autism diagnosis was given by a health care professional. From the literature search, a total of 244 studies were assessed for inclusion. Of the 244 studies, 36 studies were read in detail to indicate the cohort in which the role of the microbiome in ASD was studied and to assess their inclusion eligibility. A total of 12 studies met the inclusion criteria described above and were included in this analysis. One dataset was identified that had not yet been published but was deposited in the Sequence Read Archive (SRA) with its metadata made public during manuscript preparation. Finally, open-source data generated from the American Gut Project (AGP) were utilized to cultivate additional datasets of children with ASD and controls. A propensity score algorithm was designed to identify the best matching control subject for each ASD subject. Together, these 14 cohorts yielded a total of 888 ASD samples and 852 control samples of neurotypical individuals from a wide range of ages and geographic locations.

[0148] Exclusion criteria With the goal of minimizing known technical effects, studies that utilized only shotgun metagenomic sequencing, Roche 454 sequencing, and / or 16S rRNA pyrosequencing were not included. Therefore, 15 studies were excluded from the analysis due to incompatible sequencing methods. Eight other studies met the inclusion criteria but did not have publicly available metadata. All eight authors were contacted but did not respond to requests for metadata and therefore could not be included in the analysis.

[0149] Compiling a cohort from the American Gut Project The AGP dataset contains samples from over 15,000 participants and therefore represents a diverse dataset for compiling targeted cohorts. All available 16S data and metadata were downloaded from Qiita.ucsd.edu (https: / / qiita.ucsd.edu / study / description / 10317). We first identified fecal samples from subjects diagnosed with ASD by a physician, and compiled a control cohort based on the features of each ASD subject. Thus, we adapted a method based on propensity score matching, which consists of two main steps: scoring and matching. For each sample in the ASD cohort, a corresponding control sample was selected based on having a perfect match of certain metadata features (run date, sample type, and sex) as well as having the closest “similarity score” calculated from other metadata features (age, BMI, sea section, country, diabetes, antibiotic history, probiotic frequency, cooked food frequency, allergies, cooking method frequency, red meat frequency, fermented food consumption, whole grain frequency, vitamin B supplementation frequency, plant protein frequency, vitamin D frequency, vegetable frequency, epilepsy, or seizure disorder). Perfect matching allowed the chosen features to be controlled. For example, to control for technical variability between sequencing runs, it was required that the control sample come from the same sequencing run date as the ASD sample. On the other hand, the similarity score was simply the conditional probability of cohort assignment given the vector of metadata features. In this way, the matched control samples represented a subset of the control cohort with a metadata feature distribution similar, if not identical, to that of the ASD samples.

[0150] To calculate the similarity score, each feature was first binarized using the OneHotEncoder class from scikit-learn version 0.24.2 (https: / / scikit-learn.org / ). These features were then fitted using scikit-learn's logistic regression class, where the dependent variable was cohort status (i.e., 1 for ASD cohort; 0 for control cohort). The similarity score was the output of a sigmoid function, which was also the predicted probability of being assigned to the ASD cohort.

[0151] During the matching process, the similarity scores from each sample in the ASD cohort were compared with the similarity scores from every control sample whose "perfect match" variables were the same. The control sample with the closest similarity score was then placed in the control cohort and removed as a possible control sample for subsequent matching rounds. After matching every ASD sample, we obtained a control cohort of unique samples whose sample size was the same as that of the ASD cohort.

[0152] Data Acquisition For data publicly deposited in the NCBI Sequence Read Archive (SRA), sample metadata was downloaded from each project using the SRA Run Selector tool (https: / / www.ncbi.nlm.nih.gov / Traces). Data tables were systematically downloaded, primers trimmed, and samples processed using a custom script (process_experiment.py) available on github (https: / / github.com / mortonjt / GetData). In cases where data and sample metadata were not deposited in the SRA, they were provided directly by the authors of the original study and processed using the identical parameters described below.

[0153] Standardized 16S processing pipeline One goal was to collect and reprocess each study in a systematic and harmonized manner to reduce errors imposed by different processing methods and allow better conclusions to be drawn about the ASD gut microbiome and the dataset as a whole. Each study was individually processed using the same processing steps with QIIME 2 (v.2020.8) software. All samples were also processed together in the same pipeline to draw conclusions and uncover trends about the aggregate data as a whole. The first step in the systematic processing pipeline was primer removal and demultiplexing when required, utilizing a custom script process_experiment.py. All sequences were then trimmed to a length of 150 bp. After trimming, sequences were denoised and filtered using the Deblur QIIME 2 2020.8 plugin. Taxonomic classification was performed by the QIIME 2 q2-feature-classifier plugin with a GreenGenes-trained naive Bayes classifier (gg-13-8-99-nb-classifier.qza) pre-fitted on the full-length 16S provided by QIIME 2 (QIIME 2 v.2020.8). Phylogenetic placement was assigned by the SEPP fragment insertion plugin from QIIME 2 (v.2020.8). SEPP was chosen to minimize the effect of hypervariable regions by assigning short reads to the reference lineage. All features not aligned to the tree were removed by the q2-fragment-insetion filter-features QIIME 2 2020.8 plugin. The resulting tables were ASV-level read counts, and these were used for all downstream analyses.

[0154] Prior to alpha and beta diversity analysis, sequences from each cohort were independently rarefied to the depth required to retain the maximum number of reads possible, while filtering out any samples with very low read counts. For aggregated data, the resulting tables were rarefied to a sequencing depth of 6000 by the core-phylogenetic QIIME 2 v.2020.8 plugin. Alpha diversity was assessed by Faith's phylogenetic diversity, Shannon's diversity, species evenness, and observed features from the core-phylogenetic QIIME 2 v.2020.8 plugin. Unweighted UniFrac and weighted UniFrac distance matrices were also calculated using the core-phylogenetic QIIME 2 v.2020.8 plugin and used to calculate beta diversity distances and corresponding principal coordinate analysis (PCoA) plots.

[0155] Analysis of alpha diversity, beta diversity, taxonomic and functional differences For both aggregate data and per-study level analyses, all samples containing less than 6,000 reads per sample were filtered prior to all fold change, relative abundance, pathway abundance, and machine learning comparison analyses. Analyses were performed using custom Python (version 3.6.0) scripts. Statistical significance of beta diversity distances between groups was calculated using QIIME 2 beta-group-significance, and reported p-values ​​were calculated using permutation multivariate analysis of variance (PERMANOVA). Differences between alpha diversity of ASD and control subjects were performed on a per-cohort basis only due to the large variability between studies. Alpha diversity log2 ratios of ASD vs. controls were performed with NumPy version 1.19.1 (https: / / numpy.org / ) from the mean values ​​of ASD and control subjects calculated using SciPy version 1.5.2 (https: / / scipy.org / ). The significance of the log2 fold changes for each cohort was assessed by unadjusted two-tailed t-tests with assumed unequal variances in SciPy version 1.5.2.

[0156] For the log2 fold change per study at each taxonomic level mentioned, samples were first normalized for relative abundance by dividing the ASV read count by the total reads per sample. Relative abundance was summed at each taxonomic level using Pandas version 0.25.3 (https: / / pandas.pydata.org / ). At the per study level, the log2 ratio of the mean values ​​of ASD and control subjects was calculated using SciPy version 1.5.2 and log transformed with NumPy version 1.19.1. Statistical significance was evaluated in the same manner as that of alpha diversity described above. Log10 transformed relative abundance in each sample for the highlighted significant taxa was performed with Pandas version 0.25.3. To avoid the infinity case due to the log10 transformation, the minimum non-zero normalized relative abundance value was added to all samples utilizing NumPy version 1.19.1. The normalized relative abundances for each sample were then log10 transformed with NumPy version 1.19.1 before plotting.

[0157] Microbial functional profiles were inferred from the 16S data using the Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (Picrust2) software (v.2019.1). The resulting pathway abundance tables were analyzed for differences between ASD and control subjects as described above. All figures were generated using ggplot2 version 3.3.3 in R version 4.0.5.

[0158] Machine learning classification of ASD status To predict ASD status from the microbiome, we used 11 different classification models, evaluation metrics, and auxiliary functions from the scikit-learn package version 0.24.2 (https: / / scikit-learn.org / ). Unless otherwise stated, we estimated the mean and variance of the different performance metrics using 5-fold cross-validation repeated 10 times. No hyperparameter optimization was performed. Reported results encompassed the area under the curve (AUC) of the ROC curve for each model, as well as the accuracy rate and F1 score.

[0159] During the initial evaluation of the different machine learning models, an aggregated ASV-level dataset was used, with relative abundances normalized to sum to one before training and evaluating each machine learning model. Next, for each study (embracing the aggregated dataset), the dataset was collapsed at different levels of the taxonomic hierarchy (e.g., genus) by grouping each taxon (e.g., Clostridium), summing their raw read counts, and normalizing relative abundances to sum to one. This processed dataset was then used to train and test GBC, ABC, and decision tree classifiers from scikit-learn. Reported results include the area under the curve (AUC) of the ROC curve for each model, as well as accuracy and F1 score.

[0160] To determine the importance of each feature, the Shapley additive explanations (SHAP) package (https: / / github.com / slundberg / shap) was applied to gradient boosting classifier models trained on the aggregated ASV dataset. The top 25 most important features by cumulative SHAP value were displayed using the beeswarm plot function of the SHAP package. The absolute SHAP values ​​from each of the top 25 features were averaged for each study and displayed as a cluster map using the seaborn package version 0.11.0 (https: / / seaborn.pydata.org / ). To determine the effect of different metadata variables, the entire dataset was separated by their corresponding metadata features. Each partition of the dataset was further split using 5-fold cross-validation, and this data was used to train and evaluate gradient boosting classifiers trained to predict ASD or control status. Model performance was evaluated using the AUC of the ROC, F1 score, and accuracy rate. The entire process was repeated 50 times to estimate the standard error of each performance metric.

[0161] To estimate the predictability of ASD in different age ranges, we evaluated the predictability of ASD across different age ranges from 1 to 16 years. That is, we used only the subset of data that fell into each age bin (e.g., 1 to 3 years). To ensure sufficient data within each age bin, the lower and upper limits of each bin had to be separated by 2 years. We trained and evaluated a GBC model of 30 cases using 5-fold cross-validation on this subset of data. The mean AUC of the ROC curve for each age bin was displayed using seaborn heatmaps. Age ranges that were invalid (e.g., 3 to 1 year) were excluded and displayed as grey squares.

[0162] Data availability The data utilized in this example were from publicly available datasets deposited in the Sequencing Run Archive (SRA, https: / / www.ncbi.nlm.nih.gov / sra). The data can be accessed using the following study identifiers: PRJNA282013, PRJNA578223, PRJNA642975, PRJNA453621, PRJNA644763, PRJNA687773, PRJNA624252, PRJEB27306, PRJEB27306, PRJNA529598, PRJEB42687. Other datasets included data from David et al. 2021 (available at http: / / files.cgrb.oregonstate.edu / David_Lab / ASD_study1 / ) and data from the AGP available from Qiita with study identifier 10317 (https: / / qiita.ucsd.edu / study / description / 10317).

[0163] Code Availability Data processing scripts are available on Github (https: / / github.com / mortonjt / GetData). Notebooks for both data analysis and visualization are also available on Github (https: / / github.com / erxw / ASD_16S_metaanalysis).

[0164] Example 2 In this example, further meta-analysis was performed on additional cohorts. Whereas the dataset used in Example 1 consisted of 1,740 samples from 14 cohorts, the dataset used in this example consisted of 3,589 samples from 22 cohorts of 16S sequencing data for ASD and control subjects. A meta-analysis framework was established to identify commonly modulated taxa occurring in the subjects' microbiota, functional differences between gut microbiomes, and predicted metabolic fluxes.

[0165] Sarcina Taxonomic analysis identified potential biomarkers of ASD status in the presence of sequencing data from the genus Sarcina. This example outcome suggests that around 14% of all ASD subjects contained the threshold abundance of the species, whereas only 6.4% of control subjects contained a similar abundance of the species.

[0166] Figure 8A shows commonly modulated genera between ASD and control subjects. The compiled 16S sequencing data was collapsed at the genus level and partitioned among different metadata characteristics for the samples to compare differences between ASD and controls based on various factors. Log2-transformed fold changes of ASD / control compositional comparisons were plotted for each factor, and significance according to an adjusted rank sum test for multiple hypothesis testing was represented by the size of each point. Box plots displaying interquartile ranges (IQRs) indicate the general distribution of the data. Notably, the IQRs for Sarcina were overall above 0, suggesting a biomarker of ASD status that is resilient to the various studied factors.

[0167] The Sarcina biomarker may be Clostridium ventriculi or a close relative of this species. Other close relatives include Clostridium perfringens, which is known to contain several toxins that can induce ASD-like symptoms, including tachycardia, dehydration, anorexia, GI problems, tremors, lethargy, depression, lethargy, hypersensitivity, hyperactivity, irritability, seizures, neuroinflammation, permeable gut / blood-brain barrier, and oligodendrocyte cell death. Thus, the toxin(s) or bacteria may be targets for therapeutic intervention in ASD.

[0168] Figure 8B shows a histogram of the compositional abundance (when present) of Sarcina in the gut microbiome for ASD and control subjects. Figure 8B also shows the overlaid kernel density estimation (KDE) curves for ASD and control subjects. The optimal discriminant boundary for classifying ASD vs. control was determined and plotted in Figure 8B.

[0169] Figure 8C is a heatmap depicting the abundance fraction of ASD or control subjects that contain above-threshold abundances of Sarcina in their gut microbiome. The dataset is separated by cohort and study factors. In addition, the difference between the abundance fraction of ASD and the abundance fraction of controls is plotted, showing the cohorts and study factors in which this feature was highly enriched in ASD subjects.

[0170] Chorismic Acid A meta-analysis of the expanded dataset of this example largely supported the findings from Example 1, strengthening the claim and suggesting that chorismate is likely the key molecule that is decreased in the gut of many ASD subjects. Reanalysis of the most commonly significant pathways showed a high frequency of pathways involving chorismate.

[0171] Chorismate metabolic pathways have been identified as present in most, if not all, gut microbiomes, and ASD subjects had a higher proportion of their microbiome directed towards these pathways. Identification of discriminant boundaries based on chorismate metabolism identified cohorts and study factors in which chorismate metabolism was particularly good at discriminating between ASD and control subjects. These findings suggest that interventions based around chorismate may be particularly relevant in places such as Italy, may not discriminate between the sex of ASD subjects, and may not be related to gastrointestinal symptoms.

[0172] FIG. 9A is a bubble plot showing the most common significant pathways found in a cross-sectional cohort of ASD gut microbiomes and controls. Each point is colored by the Log2 compositional fold change between ASD and control subjects and sized by whether the association passed a significance threshold of P<0.05 in a rank sum test without adjustment for multiple hypothesis testing. Pathways involving chorismate are indicated by arrows. These pathways include the menaquinol-7 biosynthetic superpathway, the chorismate metabolic superpathway, the degradation of aromatic compounds via beta-ketoadipate, the menaquinol-12 biosynthetic superpathway, the menaquinol-11 biosynthetic superpathway, the demethylmenaquinol-8 biosynthetic superpathway, the menaquinol-8 biosynthetic superpathway, the menaquinol-13 biosynthetic superpathway, and enterobactin biosynthesis.

[0173] Figure 9B shows a histogram of the compositional abundance (when present) of the chorismate metabolic superpathway in the gut microbiome for ASD and control subjects. Figure 9B also shows the overlaid KDE curves for ASD and control subjects. The optimal discriminant boundary for classifying ASD vs. controls was determined and plotted in Figure 9B.

[0174] Figure 9C is a heat map that describes the abundance ratio of ASD or control subjects that contain chorismate metabolic superpathway with abundance above threshold in the subject's gut microbiome.The data set is divided by cohort and study factor.In addition, the difference between ASD abundance ratio and control abundance ratio is plotted, showing the cohort and study factor that this feature is highly enriched in ASD subjects.

[0175] FIG. 9D shows a cross-comparison of chorismate metabolism and other microbiome features. Discriminant thresholds were identified for various features linked to ASD status, and ASD subjects identified with high levels of each category were compared to ASD subjects identified with increased levels of chorismate metabolism. FIG. 9D shows a Venn diagram describing the overlap between subjects categorized as having increased abundance of each listed feature. As shown in FIG. 9D, strong overlap was identified in patients with increased enterobacteriales gene, menaquinol-11 gene, aromatic compound degradation, and L-arginine degradation. Since patients with increased enterobacteriales are almost entirely represented by patients with increased chorismate metabolism, chorismate metabolism phenotypes may often be driven by enterobacteriales. Thus, in some embodiments, the compositional abundance of enterobacteriales, L-arginine degradation, and / or enterobactin may be used to assess whether a patient is likely to respond to chorismate intervention.

[0176] Metabolic flux differences The differences in metabolic flux between ASD and control subjects were analyzed using the computational tool MICOM. MICOM uses metabolic models built around metabolic pathways and known nutrient requirements within microbes to simulate how the collective species identified in the microbiome interact with each other given the diet to be ingested. These studies utilized a standard "Western diet." A customized "Western diet" was also developed that increased the amount of chorismate available in the diet.

[0177] Figure 10A is a bubble plot showing the most common significant pathways found in a cross-sectional cohort of ASD gut microbiomes and controls. Each point is colored by the Log2 fold change between ASD and control subjects and sized by whether the association passed a significance threshold of P<0.05 in a rank sum test without adjustment for multiple hypothesis testing. The most common metabolic flux differences between ASD and control subjects included nicotinamide mononucleotide (NMN), 3-methyl-2-oxopentanoic acid, L-phenylalanine, nicotinamide, fumarate, chorismate, L-tyrosine, L-serine, glycyl-L-tyrosine, and glycylphenylalanine.

[0178] When examining different features for their potential to correspond to subgroups of ASD subjects, proton flux appears to be a promising candidate, where 14% of the microbiomes of ASD subjects indicated a high influx of free protons. This phenotype was particularly prominent in the Ding2020 and Kong2019 cohorts, as well as in studies carried out in Ecuador and Korea that used the V3V4 variable region to profile the gut microbiome. Figure 10B shows a histogram with overlaid KDE curves describing the distribution of proton influx (when present) by the gut microbiome of ASD and control subjects. The optimal discriminant boundary for classifying ASD vs. controls was determined and plotted in Figure 10B.

[0179] Figure 10C is a heat map depicting the abundance fraction of ASD or control subjects that contain above threshold influx of protons by their gut microbiome. The data set is divided by cohort and study factor. In addition, the difference between the abundance fraction of ASD and the abundance fraction of controls is plotted, showing the cohort and study factor where this feature is highly enriched in ASD subjects.

[0180] Metabolic flux analysis further supports the assertion based on the functional analysis of ASD gut microbiome (regulating chorismate pathway molecules for ASD treatment) because metabolites related to chorismate are frequently identified as molecules with differential uptake by the gut microbiota of ASD subjects. These molecules include L-phenylalanine, L-tyrosine, and chorismate. The results further suggest that NMN, fumarate, and / or protons may be important for ASD pathology. For example, the increased proton flux found in some ASD subjects may indicate that their gut pH is more acidic. A possible intervention for these subjects may be to reduce GI pH. Thus, in some embodiments, therapeutic intervention comprises administering one or more compositions to reduce GI pH of subjects with or at risk of ASD.

[0181] Comparison of major subtypes Having identified several promising biomarkers for ASD, subjects within each group were compared to each other to identify whether they were distinct subtypes of ASD. For this analysis, ASD subjects who exceeded the threshold set for the chorismate metabolic superpathway were compared to ASD subjects who contained the biomarker "Sarcina" and to patients who contained high levels of proton influx. Results found very little overlap in these ASD populations, suggesting that they may be distinct subtypes of ASD that may require a diagnostic approach to identify each subtype to identify which targeted therapeutic may work best for the individual. For the largest subtype, it was estimated that at least 18% of ASD subjects could benefit from a chorismate-based intervention. Furthermore, it was estimated that at least 13.8% of ASD subjects could benefit from an intervention based on GI pH adjustment, and 13.5% of ASD subjects could benefit from an intervention targeting Clostridium ventriculi.

[0182] FIG. 11 shows an upset diagram comparing patient populations of the three most major subtypes of ASD from this example. The bottom left bar graph shows the number of patients of each subtype with the corresponding percentage. The bottom right indicates the participation in the specific subtype that corresponds to each bar graph above. Each bar graph shows the number of ASD subjects with participation in each subtype, and each bar is colored depending on the country of the ASD subjects. The percentage of the total population represented by each subpopulation is plotted above each bar.

[0183] Chorismic Acid Intervention Having identified a subpopulation of ASD with a higher capacity to metabolize chorismate and molecules derived from chorismate, we assessed the impact that increasing chorismate in the diet of these patients would have on the metabolic uptake of their microbiome. For this analysis, the dietary flux of chorismate was increased by an order of magnitude and the MICOM simulation was re-run with all subjects in the meta-analysis study. The total flux of each metabolite through the gut microbiota was compared to the flux obtained from a standard diet. The flux differences in the ASD chorismate subtype were compared to all other ASD subjects and the most statistically significant metabolites between these groups were ranked. The analysis identified strikingly significant differences in several metabolites that may be key to ASD pathology, including indole, folate, chorismate, L-tyrosine, and menaquinone8, a derivative of tryptophan that has behavioral impacts. These studies also identified differences in other interesting metabolites, including 3-methyl-2-oxopentanoic acid and NMN, which are known neurotoxins. The average flux of each of these metabolites was reduced by chorismate intervention in these subjects, but did not change significantly in the larger population. This reduced flux of these metabolites by the gut microbiota may indicate that the increased availability of chorismate eliminates the need for certain species to degrade downstream metabolites of chorismate. This further indicates that the intervention of chorismate in this patient population may have an intended mechanism of action by reducing the amount of degradation of various metabolites downstream of chorismate, as a compensation for the reduced availability of chorismate due to glyphosate exposure or other mechanisms.

[0184] Figure 12A shows the top 30 metabolites modulated in ASD subjects with increased chorismate metabolism genes when fed increased chorismate. A bar graph of -Log10 (FDR adjusted p-value) based on average rank sum statistical comparison between flux differences with and without the chorismate modified Western diet for chorismate and non-chorismate ASD subtypes is shown.

[0185] 12B-12E show box plots showing the distribution of flux differences for chorismate subtypes or other subjects with ASD when given increased chorismate in their simulated diet. Above each box plot is a significance marker based on an unpaired corrected Welch's T-test.

[0186] Simulated chorismate intervention studies indicate that chorismate can be utilized to modulate the levels of key metabolites (e.g., L-tyrosine, folate, indole, menaquinone8) in certain ASD subjects. They demonstrate that administration of chorismate may likely reduce gut microbial reliance on uptake of downstream metabolites of chorismate. From previous findings of these downstream metabolites in various symptoms or comorbidities of ASD, it suggests that chorismate may intervene through these mechanisms.

[0187] Example 3 In this example, chorismate is administered to treat two separate rodent models of ASD: prenatal exposure to either glyphosate (also known as Roundup) or valproic acid.

[0188] Autism spectrum disorders emerge from distinct etiologies: altered microbiome, viral infection, and genetics. A subset of ASD patients may possess alterations on the chorismate metabolic pathway encoded by their gut microbiota. Chorismate supplementation may alleviate ASD symptoms and may be a possible treatment for at least the “chorismate subtype” of ASD. In this example, chorismate is administered to treat two separate rodent models of ASD: prenatal exposure to either glyphosate (also referred to as Roundup) or valproic acid. Glyphosate has been shown to directly interfere with the synthesis of chorismate by the shikimate pathway and induce ASD-like behaviors in rodents. Although the exact mechanism of action of valproic acid remains unclear, rodents born to valproic acid-exposed mothers also display many ASD-like behavioral phenotypes.

[0189] This example focuses on the primary behavioral symptom of rodent models of ASD, social deficits, which are measured using a three-chamber assay described below. Secondary measures encompass the metabolic and microbial profiles of the gut microbiome.

[0190] Experimental Method Induction - Valproic acid Mating-day-designated pregnant C57B 1 / 6 females are administered vehicle (0.9% saline) or valproic acid (VPA) (500 mg / kg; 500 μL; intraperitoneally) on gestation day 12.5. Male pups are weaned at 3 weeks of age (P21); female pups are euthanized.

[0191] Induction - Glyphosate Mating-day-designated pregnant C57B1 / 6 females are administered regular drinking water or glyphosate-spiked water (175 mg / kg; approximately 0.1%) from embryonic day 5 (E5) through postnatal day 21 (P21) for a total of 37 days. Male pups are weaned at 3 weeks of age (P21); female pups are euthanized.

[0192] Treatment – ​​Chorismic Acid At weaning (P21), animals are group housed, weighed, monitored daily, and fecal samples are collected weekly. Half of the animals in each group receive either regular drinking water or chorismate in the drinking water. Drinking water is changed every 24 hours.

[0193] Lead Out - Social The primary behavioral readout is the sociality of the animals using a three-chamber assay, in which a rectangular arena contains three chambers and subject mice are allowed to walk freely between the chambers.

[0194] At P56, mice are allowed to habituate within the chambers for 5 min. Then, a pair of empty small wire cages are placed on either side of the left and right chambers. A stranger mouse is introduced into one of the wire cages. The subject mouse is allowed to freely explore between the three chambers for 10 min. Video tracking records the amount of time the subject mouse spends in each chamber.

[0195] At P66, animals are euthanized. Blood (plasma), colon contents (flash frozen), cecal contents (flash frozen), and brains (flash frozen) are collected at the time of euthanasia.

[0196] Expected Results Previous studies have shown that control subject mice (controls for induction and treatment) prefer to spend 2–5× more time in the chamber with the stranger mouse than in the chamber with an empty cage.

[0197] These studies also show that control mice with prenatal exposure to valproic acid or glyphosate exhibit key ASD-like behavioral phenotypes, spend roughly equal amounts of time between chambers, and do not display social preference.

[0198] In contrast, control mice with prenatal exposure to valproic acid or glyphosate but treated with chorismate should show social preferences similar to those of control mice and spend 2–5× more time in the chamber with the stranger mouse than in the chamber with an empty cage.

[0199] It is also hypothesized that chorismate is necessary for normal neurogenesis but is not a substrate for promoting sociality, i.e., control mice treated with chorismate (induced controls) do not show increased sociality when compared to control mice (induced and treated controls).

[0200] [Table 1-1] [Table 1-2]

Table 1-3

[0201]

Table 2-1

Table 2-2

Claims

1. 13. A method for aiding in the diagnosis of a subject having one or more symptoms of autism, the method comprising determining a level of chorismate and / or a molecule in the chorismate metabolic pathway in a sample obtained from the subject.

2. The method of claim 1, wherein the molecule in the chorismate metabolic pathway is an upstream or downstream metabolite of chorismate.

3. 3. The method of claim 2, wherein the downstream metabolite of chorismate is tryptophan, tyrosine, phenylalanine, or a metabolite thereof.

4. 4. The method of claim 3, wherein the downstream metabolite of chorismate is a downstream metabolite of tryptophan, wherein the downstream metabolite of tryptophan is serotonin, melatonin, kynurenine, and / or indole.

5. The method of claim 2, wherein the downstream metabolite of chorismate is folic acid or a derivative thereof.

6. The method of claim 2, wherein the downstream metabolite of chorismate is a siderophore.

7. The method of claim 6 , wherein the siderophore is enterobactin.

8. 3. The method of claim 2, wherein the downstream metabolite of chorismate is ubiquinol or menaquinol.

9. The method according to any one of claims 1 to 8, wherein the sample is a stool sample.

10. The method according to any one of claims 1 to 8, wherein the sample is a gastrointestinal sample.

11. The method of claim 10, wherein the gastrointestinal sample is a gastric aspirate, a biopsy, an intestinal fluid, an endoscopic brush, and / or a laser capture microdissection sample.

12. The method of any one of claims 1 to 8, wherein the sample is a saliva, buccal, nasal, urine, serum, or cerebrospinal fluid sample.

13. 13. The method of any one of claims 1 to 12, wherein determining the level of chorismate and / or molecules in the chorismate metabolic pathway comprises amplifying and / or sequencing one or more portions of one or more nucleic acids present in the sample.

14. 14. The method of claim 13, wherein the one or more portions of the one or more nucleic acids comprise one or more regions of 16S ribosomal RNA present in the sample.

15. 14. The method of claim 13, wherein amplifying and / or sequencing one or more portions of one or more nucleic acids of a sample comprises performing quantitative PCR on at least a portion of the sample.

16. 14. The method of claim 13, wherein amplifying and / or sequencing one or more portions of one or more nucleic acids of the sample comprises performing shotgun metagenomic sequencing on at least a portion of the sample.

17. 17. The method of any one of claims 1 to 16, wherein determining the level of chorismate and / or molecules in the chorismate metabolic pathway comprises using HPLC, LC-MS, GC-MS, and / or MALDI-TOF.

18. 18. The method of any one of claims 1 to 17, further comprising determining that the subject has or is at risk of developing ASD if the level of chorismate and / or molecules in the chorismate metabolic pathway is above or below a threshold value.

19. 20. The method of claim 18, further comprising regulating levels of chorismate and / or molecules in the chorismate metabolic pathway in the subject to alleviate one or more symptoms of ASD and / or a comorbidity of ASD or to reduce the risk of developing ASD.

20. 20. The method of claim 19, wherein regulating the levels of chorismate and / or molecules in the chorismate metabolic pathway comprises administering to the subject a composition comprising chorismate, chorismic acid, and / or a salt or ester of either.

21. 21. The method of any one of claims 19 to 20, wherein modulating the levels of chorismate and / or molecules in the chorismate metabolic pathway comprises administering to the subject a composition comprising a chorismate prodrug.

22. 22. The method of any one of claims 19 to 21, wherein modulating the levels of chorismate and / or molecules in the chorismate metabolic pathway comprises administering to the subject a composition comprising a chorismate metabolite.

23. 23. The method of any one of claims 19 to 22, wherein modulating the level of chorismate and / or molecules in the chorismate metabolic pathway comprises administering to the subject a composition that modulates expression and / or activity of one or more enzymes in the chorismate metabolic pathway.

24. 24. The method of claim 23, wherein the composition comprises an agonist, antagonist, activator, and / or inhibitor of one or more enzymes in the chorismate metabolic pathway.

25. 25. The method of any one of claims 23 to 24, wherein the one or more enzymes comprise chorismate lyase.

26. 26. The method of any one of claims 19 to 25, wherein modulating the level of chorismate and / or molecules in the chorismate metabolic pathway comprises modulating the level of chorismate.

27. 27. The method of any one of claims 19 to 26, wherein modulating the level of chorismate and / or molecules in the chorismate metabolic pathway comprises modulating the level of molecules in the chorismate metabolic pathway.

28. A method of treating a subject having one or more symptoms of or at risk of developing ASD, the method comprising regulating levels of chorismate and / or a molecule in the chorismate metabolic pathway in the subject.

29. 29. The method of claim 28, wherein the levels are regulated in the intestine of the subject.

30. 30. The method of any one of claims 28-29, wherein modulating levels of chorismic acid and / or molecules in the chorismic acid metabolic pathway comprises administering to the subject a composition comprising chorismic acid, chorismic acid, a chorismic acid prodrug and / or a chorismic acid metabolite.

31. 31. The method of any one of claims 28 to 30, wherein modulating the level of chorismate and / or molecules in the chorismate metabolic pathway comprises administering to the subject a composition that modulates expression and / or activity of one or more enzymes in the chorismate metabolic pathway.

32. 32. The method of claim 31 , wherein the one or more enzymes comprises a chorismate branch point enzyme.

33. 33. The method of claim 32, wherein the chorismate branch point enzyme comprises chorismate lyase.

34. The method of any one of claims 31 to 33, wherein the composition comprises an agonist, antagonist, activator, and / or inhibitor of one or more enzymes in the chorismate metabolic pathway.

35. 35. The method of any one of claims 28 to 34, wherein modulating the level of chorismate and / or molecules in the chorismate metabolic pathway comprises administering to the subject a composition that modulates the amount of microorganisms that produce chorismate and / or molecules in the chorismate metabolic pathway in the gastrointestinal tract of the subject.

36. A method comprising determining the abundance of one or more microbial signatures in a sample obtained from a subject to predict and / or diagnose the ASD status of the subject.

37. 37. The method of claim 36, wherein one or more microbial features include Sarcina bacteria.

38. The method of any one of claims 36 to 37, wherein one or more microbial characteristics include Clostridium ventriculi.

39. 39. The method of any one of claims 36-38, further comprising determining that the subject has or is at risk for developing ASD if the abundance of one or more microbial features is higher than a threshold.

40. 40. The method of claim 39, further comprising reducing the abundance of one or more microbial traits in the subject to alleviate one or more symptoms of ASD and / or a comorbidity of ASD or to reduce the risk of developing ASD.

41. 41. The method of any one of claims 36-40, wherein reducing the abundance of the one or more microbial characteristics comprises administering one or more antibiotic and / or phage therapy to the subject.

42. 42. The method of any one of claims 36-41, further comprising determining that the subject has or is at risk for developing ASD if the abundance of one or more microbial features is lower than a threshold value.

43. 43. The method of claim 42, further comprising increasing the abundance of one or more microbial signatures in the subject to alleviate one or more symptoms of ASD and / or a comorbidity of ASD or to reduce the risk of developing ASD.

44. 44. The method of claim 43, wherein increasing the abundance of one or more microbial traits comprises administering to the subject one or more prebiotics and / or probiotics.

45. 13. A method for aiding in the diagnosis of a subject having one or more symptoms of autism, the method comprising determining a level of proton flux in the gastrointestinal tract of the subject from a sample obtained from the subject.

46. 46. ​​The method of claim 45, further comprising determining that the subject has or is at risk for developing ASD if the level of proton influx is above a threshold value.

47. 47. The method of claim 46, further comprising regulating gastrointestinal pH in the subject to alleviate one or more symptoms of ASD and / or a comorbidity of ASD or to reduce the risk of developing ASD.

48. 11. A method for identifying a subtype of a subject having ASD, the method comprising determining the composition of the fecal microbiome or the presence of one or more microbial features in a sample obtained from the subject.

49. 1. A method comprising: determining one or more microbial features in a sample obtained from a subject; and using the one or more microbial features as biomarkers for medical decision-making regarding therapeutic intervention.

50. A method comprising: training a machine learning model to correlate a microbial signature with risk or risk level of ASD, and / or candidacy or non-candidacy for treatment of ASD in a subject.

51. A non-transitory computer-readable storage medium having executable instructions encoded thereon that, when executed by at least one processor, cause the at least one processor to perform any one or any combination of the aforementioned methods.

52. An apparatus comprising: at least one processor; and at least one non-transitory computer-readable storage medium having executable instructions encoded thereon that, when executed by the at least one processor, cause the at least one processor to perform any one or any combination of the aforementioned methods.

53. determining the abundance of one or more molecular or microbial features in a sample obtained from the subject; administering one or more treatments based on the absence and / or low abundance of one or more molecular or microbial features to alleviate one or more symptoms of ASD and / or ASD comorbidities or to reduce the risk of developing ASD; A method comprising:

54. A method comprising administering one or more probiotics and / or prebiotics based on the absence and / or low abundance of one or more microbial hallmarks to alleviate autism comorbidities and phenotypes.

55. The method includes targeting one or more microbial features based on the presence and / or excess of the one or more microbial features to alleviate one or more symptoms of ASD and / or a comorbidity of ASD in a subject or to reduce the subject's risk of developing ASD.

56. 57. The method of claim 56, wherein targeting one or more microbial characteristics comprises administering one or more antibiotic and / or phage therapies to the subject.

57. A method comprising targeting one or more microbial molecules associated with one or more microbial signatures to alleviate one or more symptoms of ASD and / or a comorbidity of ASD in a subject.

58. 58. The method of claim 57, wherein the one or more molecules comprise one or more toxins.

59. 59. The method of any of claims 1-58, wherein the subject has one or more signs or symptoms of ASD or is at risk of developing ASD.

Images