You might think the title of this article is a little silly, but there is a good reason for excitement thanks to new insights from some recent studies that combined the power of genomics and metabolomics. What wows me is that when you couple information on genotype and molecular phenotype (aka metabolomics), you really get something special, not just for research but also for the clinic. While genomics on its own is widely used across biology, metabolomics is an incredibly powerful addition for extracting value out of that research. I believe that metabolomics puts some extra ‘Gee’ into genomics.

What genes should we focus on?
A big issue facing biologists and clinicians is the uncertainty of gene association to function in human health. What does a gene variant do or not do? Additionally, we know that the presence of a genetic mutation is not necessarily synonymous with the development of a disease. A range of factors, including epigenetics, environmental exposures, microbiome, and lifestyle choices such as diet and exercise, influence phenotype. Genotype ≠ phenotype.

dna.pngCurrently, genomic information can only suggest what diseases we might be predisposed to, but it’s an incomplete picture of health. The data may give either a false sense of security about implied good health or, at the other extreme, lead to some sleepless nights. That’s why 23andMe and companies that offer similar testing put several disclaimers in front of consumers before they can access their personal genetic results. Nature published an article in 2016, “A radical revision of human genetics: Why many ‘deadly’ gene mutations are turning out to be harmless,” that is particularly enlightening on the issue of understanding genomic information. For myself, at the moment, I believe that such personalized genetic testing is at an immature stage and am doubtful of the benefit without accompanying phenotypic information.

How do we identify variation in genes that could contribute to or cause disease?
One approach to identifying gene variants associated with disease is the genome-wide association study (GWAS). These studies have linked thousands of loci (regions on a chromosome) to human disease, but there are issues with exactly locating and identifying the specific gene and mutations linked to a trait. In addition, one needs to establish causality and not just correlation. On top of this, perhaps only a few thousand genes1 out of the 20,000 or so that we carry have so far been associated with human disease. So, how can we improve the process of identifying gene variants associated to disease and establishing causality? The answer, or at least part of the answer, to that question is quite straightforward – we need phenotypic data.

Metabolomics aims to identify and quantify all the metabolites in a sample such as a body fluid or tissue. Measurement of metabolites provides a molecular phenotype that can be used as a proxy or surrogate for a physical phenotype - a snapshot of current health status. Metabolomics can be used in combination with genetic sequencing information to improve medical interpretation of an individual’s disease risk.1

Similarly, metabolomics combined with genomic analysis has been used to identify significant associations of gene variants with metabolite concentrations in blood.2,3 In other words, the readily seen changes in metabolite levels help demonstrate what the gene, or mutation in the gene, is actually doing.

As genomics and metabolomics technology have matured, 246 gene variant associations to metabolism have been identified. The work published in 2017 by Long et al.3 used Metabolon’s Precision Metabolomics™ platform coupled with whole genome sequencing.

Deepening the search with whole genome sequencing combined with metabolomics
The concentration of metabolites in the blood can vary widely from individual to individual, with variation arising from both genetic and environmental factors. For this reason, genomics alone cannot tell the full biological story and phenotypic data is required to strongly identify significance. Researchers from Human Longevity, Health Nucleus, King’s College, Baylor College of Medicine and Metabolon demonstrated a significant level of heritability for a large number of metabolites using Precision Metabolomics to analyze blood samples, with the median heritability being quite high at 48%.3 Genetic sequence variations at 101 loci were associated with the levels of 246 metabolites, of which 90 associations of gene variant to metabolite level for 85 metabolites in plasma had not been seen before. Of the novel variants, five had previously been associated with diseases, but not with metabolite levels.

Rare variants revealed through metabolic outliers
Long, et al. focused on extreme outliers in the population in terms of metabolite levels to identify rare variants that associated to those extremely high or low levels. They identified 151 individuals who had one or more of 69 metabolites with levels consistently very different from the population mean. Additional individuals were identified using further methods to make a total of 175. They then looked for rare functional gene variants in these ‘outlier individuals’ that might explain the extreme metabolite levels. After excluding some variants that had also been identified in individuals with normal metabolite levels, they identified 14 rare variants in 10 genes. In addition, a further 14 rare variants from seven genes were identified by searching the genomes of the 1,960 study participants for rare functional variants that associated to statistically significant (abnormal but less extreme) differential levels of metabolites.

Interesting associations
Overall, approximately one in 10 unrelated individuals had metabolite blood levels that associated with rare genetic variants. Many gene-metabolite pairs were associated with inherited metabolic disorders (IMDs), which should probably not come as a surprise. What might be surprising is that some of these were novel, and that some outlier metabolite levels in heterozygous individuals associated to autosomal recessive IMDs or other pediatric diseases. Since these associations were observed in heterozygous individuals, one would expect them to be clinically and phenotypically normal.

One individual, heterozygous for a rare variant in SLC6A3, suffers from adult-onset Parkinson’s disease. Variants in this gene have been shown to cause infantile parkinsonism dystonia when homozygous, with reduced dopamine reuptake observed. It is intriguing to think that elevated levels of dopamine sulfate detected in this individual by Precision Metabolomics may have resulted from this defective gene. We should consider the “possibility that the heterozygous variant may translate into adult-onset clinical symptoms.” Thus, there is the real possibility that late-onset phenotypes, previously thought to occur in childhood, could be present and result in adult disease.

Identifying unknown, unidentified metabolites using associations to genes of known function               

A number of unidentified metabolites were associated with genes of known function. The authors attempted to identify these unknown metabolites using liquid chromatography-mass spectrometry (LC-MS) data from the metabolomics experiments in combination with the corresponding genetic information. 

How do you identify these unknown and unnamed metabolites? Metabolon has a great deal of institutional expertise in, and developed proprietary methods for, interpreting mass spectral data for metabolite identification. In addition, knowledge derived from our already existing, extensive metabolite data aids in forming a putative identification of an unknown metabolite. For example, an unknown compound might have similar but not identical MS/MS data to a known metabolite. An example unknown metabolite designated as X-12511 was associated with N-acetyltransferase 8 (NAT8). Analysis of the LC-MS data, combined with this gene to metabolite association, gave the confident structural assignment of an acetylation product of 2-aminooctanoic acid, N-acetyl-2-aminooctanoic acid.

Many gene variants with large effects were identified in this study. There was a wide variance in metabolite levels from the extreme of IMDs through to abnormal outlier metabolite profiles in adults, supporting the role of rare gene variants in common diseases. More than one-third of unidentified metabolites were successfully mapped to genetic loci, with some of these unknowns being subsequently identified when analyzed using metabolomics LC-MS data. Lastly, and of incredible interest to me, the authors stated that, “Our data underscore the metabolic consequences of multiple rare variants and leaves open the possibility that they may translate into adult-onset clinical symptoms.”

We cannot rely on genomics to tell us about the state of our health. Genomics can only tell us about the potential risk of developing a disease sometime in the future. We need measurements that tell us about the state of our health now. We are all individuals; our metabolisms are different one from the other, and they can confer upon us disease just as easily as they can protect us from it. Metabolomics is a powerful tool to help us understand that. One of the most compelling conclusions to draw from genomics studies that have also collected metabolomics data is that we should consider using metabolic screening more routinely to understand chemical uniqueness and its impact on individual health.

If you would like to learn more about metabolomics coupled with genomics, you can download our free eBook Bringing the Genome to Life with Metabolomics: A “Sentinel” for the Genome-Phenotype Relationship.


1. Guo, L. et al. Plasma metabolomics profiles enhance precision medicine for volunteers of normal health. Proc. Natl. Acad. Sci USA 112, E4901-E4910 (2015)
2. Shin, S.Y et al. An atlas of genetic influences on human blood metabolites. Nat. Genet. 46, 543-550 (2014)
3. Long, T, et al Whole-genome sequencing identifies common-to-rare variants associated with human blood metabolites. Nat. Genet. 49, 568-578 (2017)