Summer sale! Now offering 25% off DNA diet plans! Learn more

Scientific Process

This is an election year, so perhaps you are following a blog like Nate Silver maintains at FiveThirtyEight? FiveThirtyEight keeps a running tally of the likelihood that various candidates will win Presidential and Congressional elections.

But this is the Scientific Section of Gene Food, a nutrigenomics startup, why mention an election tracking model from The New York Times?

Because just as Nate Silver aggregates polling data and then makes predictions about upcoming elections, at Gene Food, we scour hundreds of peer reviewed studies on PubMed, and then weigh the strength of various genetic markers that can be influenced by nutrition, in order to score our customers into one of twenty DNA based diet types.

Just as a meteorologist predicts the weather, the model we created at Gene Food helps our customers predict how they will respond to various nutritional inputs like saturated fat, cholesterol, and refined carbohydrates.

To give you an idea of our methods for rating genes we’ll run through an example of how we assess and grade one of our polymorphisms on the site. Below we’ll run through an example for perhaps the most well known polymorphism, rs1801133, commonly known as C677T, in the MTHFR gene. Now to be clear, we list MTHFR here not because you can determine much about health and nutrition by just evaluating changes in MTHFR gene. MTHFR “mutations” are commonplace, and the research remains mixed, but we discuss it here so you have a better idea of how we score the genes used in our custom nutrition plan algorithm.

Reviewing peer reviewed research on genetics and nutrition

To start, we extract the most up to date information about a polymorphism of interest from the dbSNP website, a freely available resource provided by the National Center for Biotechnology Information in the US, which is updated and maintained by researchers from around the globe.

The above screenshot from the site provides us with some key bits of information.

Establishing clinical significance

First, it describes the possible alleles in this case G and A, and also gives information about which is the most common of the two. Here we can see that the A allele occurs in about 30% of individuals making it the minor allele. In the same image we can also see that it has a reported Clinical Significance and that a large number of publications are reported to have investigated it.

Based on this we would be able to grade this SNP as C-.

Further down the page we can extract some further information. Here, we can see that the reference polymorphisms are coming from the minus strand of the DNA molecule. To interface with 23andme and other consumer kits we need to switch this to the positive orientation.

To do this we simply switch G > C and A > T.

We can also see here that this particular polymorphisms causes an amino acid change in the MTHFR protein. Together all this information points to this polymorphism being of increased interest.

Based on this and significant research and clinical interest we would increase therefore increase the grade up to a C or C+.

For several SNPs we leave the grade here as there is no extra information available for us to base further assessment on, or there are significantly conflicting results about the impact of the polymorphism on diet and health.

But for MTHFR C677T we keep moving onwards, and switch over to dbSNPs sister site (and possibly one of the most amazing resources on the web) PubMed.

PubMed indexes a huge range of life-science journal articles. dbSNP provides a handy link at the bottom of the page to take you there directly but you can freely search for other polymorphisms, or anything else of interest.

In the image above you can see what an example search result for all papers referencing C667T looks like. 664 total PubMed hit for this MTHFR variant.

664 papers is like drinking from a fire hose, so we apply filters.

I’ve listed five key types of article that we use to assess the importance of a genetic polymorphism and also its function.

For lay people a good starting point is to select “Review” and then also “Free full text”. This will filter any articles that are behind paywalls, and show review articles which as the name suggests review the field of research. They’re a great way to get started in a field, but we’re most interested in primary research.

Our first step is to identify if the genetic change described is associated with a change in the actual function of the protein, and identify which of the alleles, C or T, is associated with any potential poor health outcomes.

Searching the PubMed index above we can find numerous articles which reference a mechanism. By reading the above article, and others, we can see that the T allele of C677T is associated with a more heat sensitive MTHFR enzyme, which impairs its function.

Next step, is to see if there are any associations with clinical outcomes. A good filter to use here is “Case Reports” which are small papers based on single individuals. Using these and other articles we can rapidly determine what health effects a particular polymorphism may have. For C677T and MTHFR there is a long list, which we cover in our blogs, for other polymorphisms however they may be much more targeted.

Identifying mechanism of action and health impact

With a mechanism of action and clear health impacts we can upgrade this SNP to B grade.

This is where most of our polymorphisms reside as further studies are either lacking or there are inconsistent results relating to effects. So what does it take to get into the A grades?

To get an A- grade the first thing we want to see are well controlled “Clinical Trials” If we filter our previous search for clinical trials we get 30 results (again significantly larger than most other polymorphisms). This is where we start doing a deep dive into the data of the paper and determine the relative impact of a polymorphism on health. Clinical trials often report useful metrics like relative risk, that allow us to determine the likelihood of a particular polymorphism impacting on health.

Great, for MTHFR C677T we can see plenty of evidence, now what about those really high grades?

Here we’re interested in our final two article types, “Meta Analyses” and “Systematic Reviews”. The approach of both these study types is to synthesize previous studies and attempt to produce an overall picture.

When dealing with genetics it isn’t uncommon to see studies reporting significantly different effects, this is because there are a whole host of other factors that come into play such as other polymorphisms, an individuals lifestyle and even environment. One way to account for this is to perform a single large study, but this is often unfeasible or prohibitively expensive.

This is where a meta analysis becomes very useful, and such studies often contain figures like the above. The solid vertical line represents the equivalent of no effect on the outcome (in this figure the effect of the T allele of C677T on total cholesterol). Each study is represented by a point on the graph, those to the right show a positive effect, and those to the left a negative association. The authors then determine the average effect across all of these studies and represent it as the red dashed line. In the above image we can therefore see that the T allele is associated with increased total cholesterol.

So for MTHFR C677T we have a described mechanism of action at the protein level, a defined impact on health which has been well studied and validated by gold standard studies.

It is therefore given a grade of A+, however, we then adjust the rating based on scientific consensus. 

Adjusting for scientific consensus

Our view internally is that while there are many unfounded claims made about MTHFR circulating on the internet, MTHFR C677T’s potential impact on heart health is backed by a large volume of high level research and therefore we score it highly by Science Grade even as we emphasize that hyper-focusing on just one gene is never the right approach.

However, we agree that there has been an overemphasis placed on MTHFR in many health circles. We also recognize that we aren’t the only organization that has looked at the research on MTHFR.

While there is a growing awareness that these methylation genes play a role in inflammatory markers like homocysteine, MTHFR alone isn’t the only gene responsible, and so we adjust our internal scoring based on the opinions of other scientists who have drawn their own conclusions.

In this case, 23andme’s official position is that MTHFR variants cannot yet be said to account for many of the claims some in the health world make.

Based on 23andme’s official position which wants to see more MTHFR research before drawing conclusions, we lower the official Science Grade to a B+ so as to reflect scientific consensus.

How We Use Genes in our Custom Nutrition Plan

Our custom nutrition plan product is an attempt to synthesize a wide range of SNPs which we have rated as having a high impact on nutritional health and wellbeing.

While we are strong believers in the profound effect that individual polymorphisms within genes can have on an individual’s general health and wellbeing, polymorphisms, genes, proteins and the complex systems of our body do not exist in isolation and so we shouldn’t assess them or make changes to our lifestyle based on a single polymorphism. In other words, we do not hyper focus on any one gene.

Understanding how all these polymorphisms, genes and systems interact is incredibly complex, and it becomes ever more complex as new and exciting discoveries are made.

To understand how it all works the best place to start is at the beginning, the raw data file that you provided to us or that we receive when you purchase one of our test kits. This file contains information about hundreds of thousands of polymorphisms scattered throughout your genome. We extract the information we require from this file and use this to build your report.

For each polymorphism we have identified the risk and non-risk allele, and also assigned a Science Grade based on a thorough review of the literature as described above.

We then assign each polymorphism to relevant categories ranging from macronutrients such as fat, carbohydrate or protein, down to micronutrients such as individual B vitamins. Then using a proprietary algorithm, we weigh all the individual SNPs, and any associated haplotypes in a particular category and assign you to a specific class. So, for example you may be recommended to limit your B6 intake, or avoid saturated fat.

We then build all of this information up into our main focus which is the overall dietary plan. Here, by incorporating all your polymorphisms we assign you to a particular diet type with associated recipes and strategies you can use to maximize your well-being.

We believe that this approach of giving you a top level overview, but also allowing you to delve further into specific categories, genes and even SNPs is the best way to provide you with the specific information to inform your dietary and supplementary choices.

Facebook icon Twitter icon Instagram icon Pinterest icon Google+ icon YouTube icon LinkedIn icon Contact icon Info icon