Larry Parnell (Twitter, Blog), Yu-Chi Lee, Chao-Qiang Lai, & Jose Ordovas (Twitter) took what must have been an enormous amount of time and put together a database of gene-environment interactions related to lipids, cardiovascular disease, and diabetes. I had no idea this many had been studied so far!
Their open access paper is available here. The data is taken from 154 published papers & 1 unpublished observation from studies using at least 20 adult subjects, and they found 554 GxE interactions that were significantly related to disease and disease biomarkers, and 1439 that were not. The significant interactions consisted of 146 SNPs/variants related to 88 genes. See the paper for # of SNPs classified by phenotype and environmental characteristics.
The authors also listed the results of pathway analysis in a table for some of these phenotypes and environmental characteristics. This gives a sense of what the genes that are the most sensitive to a phenotype or environment are involved in (or what has been focused on most so far), and notably there is much overlap with the PPAR genes.
They also analyzed 13 interactions that replicate in studies and found 9 markers in 7 different genes. Many studies didn’t replicate, reinforcing the need for a lot of study on these interactions in their various contexts.
The last table provides an example comparison to recent research of genes differentially expressed after consumption of olive oil or other anti-inflammatory compounds. The genes identified and compared against this new database suggest allele-specific effects on HDL-C, LDL-C, and total cholesterol by dietary MUFA, which olive oil contains. These studies found that some PPAR network genes are allele-specific, allowing further comparisons. This identifies “central players” as they state of phenotypes and environmental factors that are allele-specific sensitive to diet, exercise, or other factors.
They write that there is a “growing interest in the use of allele-specific pathway fluxes and differential networks,” so cataloging this data will help to advance the study of these fluxes in the diet/exercise sense.
Their resulting data was deposited into the Nutritional Phenotype Database here.
More information about GxE interactions
Larry was kind enough to provide some additional information for us to consider when looking at this data (I also asked for some other variants he considered interesting besides the PPARs):
“One important point to keep in mind: GxE interactions are only a portion of disease risk. Genetics is another component and the environment alone is a third (regardless of genotype – if you smoke, your risk of lung/esophageal cancer increases and if you eat a lot of calories, your risk of obesity, dyslipidemia and heart disease rises).
A second important point is the GxE catalog for any given phenotype or environmental factor is far from complete. That point combined with the multifaceted interplay of numerous alleles with a complex environment means it is difficult at this time to enact a regimen of personalized nutrition. Consider, for example, how complex food alone is without even looking at other important environmental factors such as sleep, alcohol intake, sunlight/seasonality, tobacco use, exercise, etc.
Variants of interest in the GxE catalog. There are many interesting variants in this catalog. APOA2 is perhaps of the most universal significance as the genetic association replicated in three populations, the first time such was observed [PubMed ID 19901143]. APOE is also remarkable because people tend to think it is deterministic for early-onset Alzheimer but GxE interactions show that this gene can “sense” the environment. Third, FTO (“fat gene”) variants interact with physical activity to alter BMI. Fourth, a number of the SNPs found in a recent publication describing 95 loci for blood lipids also show GxE interactions [Biological, clinical and population relevance of 95 loci for blood lipids (Nature 2010 466:707)], and this gives an indication of the additional complexity of the genetics of blood lipid traits. That genetic variants in the CLOCK gene participate in GxE interactions is highly interesting because of the emerging connections between circadian rhythm and obesity. Lastly, LIPC promoter variant rs1800588 is also quite intriguing. In fact, any GxE SNP that maps to a gene promoter region allows one to speculate on regulation of gene expression via the interacting environmental factor(s).
Lastly, I’d say that the breadth of the genes (with their variants), the associated phenotypes and the modifying environmental factors collected from the literature is quite exciting. The published catalog concerns only nutrition-based phenotypes but there are many published examples for other conditions and diseases. For example, there are numerous GxE interactions involving cancer phenotypes. In short, GxE interactions are not outliers but probably much more common.”
If you have purchased a genetic test through services like 23andMe, you’ve probably been a bit disappointed like me of the lack of diet, exercise, sleep, and other lifestyle related SNPs to compare to so far (in SNPedia/Promethease too). But this database provides much new data to play with.
If you have 23andMe results, I found that a nice way to browse this data instead of individually looking up SNPs is to use SNPTips (plugin for Firefox only at this point) and load this spreadsheet (significant SNPs only from the database- thanks to Larry for sending this).
Yu-Chi Lee, Chao-Qiang Lai, Jose M Ordovas, & Laurence D Parnell (2011). A Database of Gene-Environment Interactions Pertaining to Blood Lipid
Traits, Cardiovascular Disease and Type 2 Diabetes Data Mining in Genomics & Proteomics : 10.4172/2153-0602.1000106