Genotype-environment interactions for quantitative traits in Korea Associated Resource (KARE) cohorts
© Kim et al.; licensee BioMed Central Ltd. 2014
Received: 30 September 2013
Accepted: 27 January 2014
Published: 4 February 2014
Due to the lack of statistical power and confounding effects of population structure in human population data, genotype-environment interaction studies have not yielded promising results and have provided only limited knowledge for exploring how genotype and environmental factors interact to in their influence onto risk.
We analyzed 49 human quantitative traits in 7,170 unrelated Korean individuals on 326,262 autosomal single nucleotide polymorphisms (SNPs) collected from the KARE (Korean Association Resource) project, and we estimated the statistically significant proportion of variance that could be explained by genotype-area interactions in the supra-iliac skinfold thickness trait ( = 0.269 and P = 0.00032), which is related to abdominal obesity. Data suggested that the genotypes could have different effects on the phenotype (supra-iliac skinfold thickness) in different environmental settings (rural vs. urban areas). We then defined the genotype groups of individuals with similar genetic profiles based on the additive genetic relationships among individuals using SNPs. We observed the norms of reaction, and the differential phenotypic response of a genotype to a change in environmental exposure. Interestingly, we also found that the gene clusters responsible for cell-cell and cell-extracellular matrix interactions were enriched significantly for genotype-area interaction.
This significant heritability estimate of genotype-environment interactions will lead to conceptual advances in our understanding of the mechanisms underlying genotype-environment interactions, and could be ultimately applied to personalized preventative treatments based on environmental exposures.
KeywordsGenotype-environment interaction Heritability Obesity Supra-iliac skinfold thickness
Rapid advances in population genetics in recent years have led to significantly improved insight into human complex traits. Although a large number of genetic loci for many complex traits and diseases have been identified using genome-wide association studies (GWAS), the associated variants explain only a small percentage of the overall heritability . Many common, complex traits are a result of the combined effects of genes, environmental factors, and their interactions . Genotype-environment interactions (G×E) were suggested as a possible explanation for “missing heritability” , but current knowledge remains insubstantial.
G×E is defined as a phenomenon that phenotypes respond to genotypes differently according to different environmental factors. For example, a specific aspect of the environment might have a greater effect on some genotypes over others. Alternatively, there may be a change in the order of merit of a series of genotypes when they are measured under different environmental conditions . This can be expressed as the norm of reaction (NoR), which represents the profile of phenotypes produced by a genotype across different environments . Reaction norms can be depicted as several curves in two-dimensional graphs, each of which represents the response of a particular genotype to an environmental treatment, and thus the shapes of the NoRs; whether they are parallel or intersect can be used to infer important information regarding G×E .
Since G×E can obscure both genetic and environmental effects, the study of G×E is essential for improving accuracy and precision when assessing both genetic and environmental factors . It can also help illustrate how inherited characteristics render some individuals more susceptible to the negative or positive effects of specific environments. This line of investigation is important for identifying mechanisms whereby specific environmental processes might offset or exacerbate genetic risks, thereby suggesting potential targets for preventive interventions . This might ultimately allow us to provide individualized preventative advice before disease diagnosis, based on the knowledge that an individual carries a certain genotype .
Despite this importance, there are only a small number of replicated, biologically plausible, and methodologically sound examples of G×E with demonstrated clinical relevance and sufficiently high statistical power . Moreover, previous studies focused on specific genes of interest, rather than considering genome-wide genotype data, and examined variants that were influenced differentially by environmental exposure. For example, Maier (2002) reported that beryllium-exposed workers who are carriers of the Glu69 allele were more likely to develop chronic beryllium lung disease . In addition, Memisoglu et al. (2003) identified a stronger relationship between dietary fat intake and obesity in carriers of the Pro12Ala allele . One classic example of “genome-wide genotype-environment” interaction is J. Clausen’s analysis of the environmental responses of climatic races on Achillea plants. They observed that altitude affected seven distinct genotypes, but not to the same degree or in the same way, by growing genotypically identical plants (clones) in different altitudes at low, medium, and high elevations using cuttings taken from each plant . This direct experiment can be applied only to a species for which genetic replication of a sample is feasible, such as with seed crops.
In the context of the etiology of obesity, single-gene cases cannot account for latent genetic predispositions that are revealed only upon exposure to an obesogenic environment . Instead, obesity is a complex multifactorial phenotype; inter-individual variation in such phenotypes is thought to result from the action of multiple genes and environmental factors. For this reason, the traditional approach to investigating G×E, a locus-specific method, may not be effective for fully delineating the nature and the extent of the genetic polymorphisms involved in obesity-related traits. Therefore, we analyzed how much G×E contributed to variance on a genome-wide scale to better understand the genetic architecture of human complex traits, particularly those involved in obesity. Therefore, our study first estimated the heritability of the G×E component for each trait. Heritability is usually defined as the proportion of total phenotypic variation that is due to additive genetic factors, and thus it is a general and key population parameter that can help in understanding of the genetic architecture of complex traits . The term ‘interaction’ is defined as a departure from additivity in a linear model on a selected measurement scale . As such, statistical interactions are scale dependent; an interaction on the additive scale in a linear regression model may be removable by applying an appropriate transformation . However, to strictly control potential confounders, such as age, gender, and area, in this study, phenotypes were adjusted and transformed for these factors before assessing the significance of G×E. Based on the heritability analysis, we identified an obesity-related trait, in which a G×E component significantly explained the phenotypic variation, and proceeded to perform further analyses including bivariate analysis, norms of reaction, and gene functional classification to elucidate the true genetic basis of human obesity.
The U.S. National Center for Biotechnology Information (NCBI) site was used as the source of the H. sapiens genomic sequence (version GRCh37.p5).
Data collected by the Korean Association Resource (KARE) project was used for this study. The participants in the KARE project were recruited from two community-based cohorts, Ansung (rural area) and Ansan (urban city), in Gyeonggi Province of South Korea. The Ansung and Ansan cohorts consisted of 5,018 and 5,020 participants, respectively, 40−69 years old and born between 1931 and 1963. This Institutional Review Board of the Korea National Institute of Health approved this study, and all participants provided written informed consent for participation. Based on Cho et al. (2009), we excluded individuals with low call rates (< 96%), sample contamination, gender inconsistencies, cryptic relatedness, and serious concomitant illness, retaining 8,842 subjects (4,183 males and 4,659 females) .
The genomic DNA was genotyped on an Affymetrix Genome-Wide Human SNP array 5.0 containing 500,568 SNPs. Markers (GRCh37) with a high missing gene call rate (> 5%), low minor allele frequency (MAF) (<0.01), and significant deviation from the Hardy-Weinberg equilibrium (P < 10E−6) were excluded, leaving a total of 326,262 markers to be examined.
All individuals were measured for 49 quantitative traits related to obesity, blood pressure, hyperglycemia, diabetes, liver function, lung function, and kidney function. A summary of trait descriptions is provided in Yang et al. (2013) . We adjusted the phenotypes of each trait for the age effect using the model, y = b0 + b1 × age + e, and then standardized the residuals to z-scores in each of the cohorts (Ansung and Ansan) and in each gender group separately.
We defined three environmental factors in each statistical model: gender, geographical area, and age. For gender, males were coded as A and females as B. The two cities were designated as 1 for Ansung and 2 for Ansan. Age was classified into three different groups: those born in 1931−45 (A), 1946−55 (B), or 1956−63 (C), representing individuals whom experienced the Korean War (1950−53) in their childhood or when older, in their early childhood, and those born after the war, respectively. Most Koreans suffered severe nutritional deficiency during the war. Hypothesizing that nutrition affects phenotypic characteristics, the age groups were defined to determine if different nutritional statuses at a young age interacted with genotypes of a specific trait.
Additive genetic relationships and unrelated individuals
Where x ij refers to the number of copies of the reference allele for the ith SNP of the jth individual, and p i is the frequency of the reference allele. We estimated the additive genetic relationships between all pairs of individuals from SNP data, and removed one from each pair of individuals with an estimated relatedness > 0.025. Finally, we retained 7,170 “unrelated” individuals for analysis, consisting of 3,261 male and 3,909 female samples and 2,928 rural (Ansung) and 4,242 urban (Ansan) residents (Additional file 1: Table S1). The reason for exclusion was to avoid the estimate of genetic variance being driven by phenotypic correlations for parent-offspring pairs and siblings, which could have then provided a biased estimate of total genetic variance, for example confounding due to shared environmental effects .
G×E estimation and bivariate analysis
To estimate the variance of G×E effects (), we can specify the mixed linear model (MLM) as y = Xβ + g + ge + ϵ with V = A g + A ge + I , where g is an n × 1 vector of the aggregate effects of all the autosomal SNPs for all individuals, A g is the genetic relationship matrix (GRM) between individuals estimated from SNPs, and ge is a vector of genotype-environment interaction effects for all individuals, with A g = A ge for pairs of individuals in the same environment, and A ge = 0 for the pairs of individuals in different environments. The environmental effects were fitted as fixed effects in the model: a vector of fixed effects (β) with its incidence matrix (X). Because GCTA estimates the variance of the genotype-environment interaction for one environmental factor, three different models were defined separately and analyzed for each environmental factor: gender, age, and area (i.e., gender was fitted as an environmental factor to calculate genotype-gender interactions). The phenotypes were corrected previously for age and gender, and standardized to z-scores in each area cohort data separately to eliminate the necessity to include the other two fixed effects (in this example, age and area). The phenotypic variance () was partitioned into the variance explained by the genetic (), G×E (), and residual variance. The variance explained by all autosomal SNPs by restricted maximum likelihood analysis of MLM was estimated by var(g) = A g and var(ge) = A ge , relying on the GRMs. The proportions of variance explained by all autosomal SNPs (narrow-sense heritability) and by G×E were defined as and , respectively. The log-likelihood ratio test (LRT) statistic was calculated to assess the significance of heritability estimates as twice the difference in log-likelihood between the full (h2 ≠ 0) and reduced (h2 = 0) models, where h2 refers to the heritability estimate. The bivariate REML option from this software was used to estimate the genetic correlation between two traits (i.e., SUP in area 1 comprises one trait, and SUP in area 2 comprises the other trait).
GWAS and functional classification
We used the PLINK-G×E option to test for differences in the association of a trait between two regression coefficients of two different environments using linear regression analysis . The Database for Annotation, Visualization and Integrated Discovery (DAVID) v. 6.7 was used to perform gene functional classification and gene ontology analyses.
Results and discussion
Variance explained by a genotype-environment interaction component in the supra-iliac skinfold thickness (SUP) trait
Analysis of genotype-environment interactions (G×E)
Genotype × area interaction
1.8E − 02
2.4E − 02
3.2E − 04
3.0E − 02
Genotype × sex interaction
3.0E − 02
3.3E − 02
3.7E − 03
4.8E − 02
Genotype × age interaction
3.6E − 02
1.0E − 02
2.7E − 02
Even though there were some effects of gender and area on the expression of this trait, this does not necessarily contribute to G×E. Nevertheless, the statistically significant heritability estimate of genotype-area interaction from the SUP trait suggests a change in the direction or magnitude of the effect of a genotype in different areas. As such, the genotypes may have a greater or lesser effect on the risk of abdominal obesity in different environmental settings (rural vs. urban area). Together with a very recent study that explored G×E for diabetes-related traits in a European-American population , the present study is the first to examine the statistically significant heritability of a G×E on the genome scale.
The proportion of phenotypic variance due to additive genetic effects (VG/VP) was also estimated from this G×E model (Additional file 1: Tables S2-S4). We compared the estimates of VG/VP with or without the G×E component in the model across all 49 traits (Additional file 1: Figure S3). There was a significant positive correlation between Vg estimates (r = 0.70 for E as area) with or without G×E in the model. The discrepancy came mostly from the amount of VG×E, since the total variance was decomposed into one additional component for the former model, and this caused some difference in estimating the proportion of variance explained by the genetic component.
Norms of reaction on genotype groups
Phenotypic values depend on the genotype groups (G) and environmental factors (E) of two areas. For the SBP trait (Figure 1A), G had the main effect, particularly for genotypes 2 and 3; E also had a main effect, but there was no interaction between G and E. In contrast, for the SUP trait (Figure 1B), G and E were found to have main effects and an interaction. The genotypes affected phenotypic values in completely different directions and with different slopes, based on the change in area. This graphical representation supports the fact that SUP has a strong effect on genotype-area interaction compared with the control trait of SBP. However, it must be emphasized that this method is an oversimplification to facilitate and clarify discussion.
Significant and non-significant genetic variants
We also performed a genome-wide association (GWA) analysis to test genome-wide SNPs for a difference in association between the two environments with SUP and SBP traits . This single SNP association analysis revealed that the most significant SNPs were rs206942 on chromosome 6 (P = 2.74E-6) and rs189317 on chromosome 8 (P = 8.01E-06) for SUP and SBP, respectively. We confirmed that the associated genetic effect does not necessarily interact with the environment or parity. For example, Cornes et al. found that the fat mass- and obesity-associated common variant (rs9939609 of the FTO gene) showed no evidence for G×E . This same variant showed a similar result in our association study: PG = 0.001719 and PG×E = 0.7766 for the SUP trait, where PG and PG×E represent the significance of the genotype and genotype-area interaction, respectively. For the SBP trait, rs17249754, known for its relationship with blood pressure , had the highest significance for genotype (PG = 5.04E-09), but no evidence of genotype-area interaction (PG×E = 0.4735).
To determine if variances captured by SNPs differ between areas, we performed bivariate analysis, considering SUP (or SBP as control) in area 1 as one trait and SUP (SBP) in area 2 as the other trait (Additional file 1: Table S6). For SBP, the genetic correlation between areas was 1.00 (s.e. = 0.31), suggesting that the same genetic signals explained the variance in SBP in different areas. In contrast, there was a negative genetic correlation (rg = -0.26, s.e. = 0.26) between areas for the SUP trait, suggesting that the genetic factors for this trait in two areas are not positively correlated (P = 0.002, rejecting the null hypothesis rg = 1). Although the significance does not survive multiple testing correction, this result may imply that different genetic signals are associated with abdominal obesity in different areas.
Gene functional classification and gene ontology analysis
Lack of statistical power in heritability estimates
G×E might not appear in heritability estimates due to the lack of statistical power, particularly if a small fraction of individuals experience adverse exposure, and population stratification in the opposite direction of the allelic effect . However, this specific analysis was exempt from these limitations, since the environmental factor (area) divided the population into approximately equal sample sizes. In addition, we concluded from a previous study  and PCA plot Yang et al. )  that the population structure of KARE could be disregarded, and thus did not preclude our analysis of the interaction.
Although the ‘nature versus nurture’ debate forced the admission that both genetic and environmental factors contribute to phenotypic variation, scientists continue to consider their interaction. Most genetic epidemiology studies have not considered G×E effects, simply because of the difficulty in assessing these effects in quantitative genetic models and the lack of sufficient statistical power to provide sufficient proof . However, based on the findings of the current study, the lifestyle and environmental factors associated with increased risk of obesity could eventually be specified for each individual, and preventive medical and public health strategies could be developed for population subgroups with an emphasis on high-risk individuals. We observed that the SUP trait contains a significant G×E component. This result may be of paramount importance due to increasing evidence that obesity is reaching epidemic proportions worldwide.
Norm of reaction
Supra-iliac skinfold thickness
Systolic blood pressure
Single nucleotide polymorphism
Korean Association Resource
Genome-wide Complex Trait Analysis.
This work was supported by a grant (PJ009019) from Next-Generation BioGreen 21 Program, Rural Development Administration, Republic of Korea. This was also supported by the Bio & Medical Technology Development Program of the National Research Foundation (NRF) funded by the Ministry of Science, ICT & Future Planning (2012M3A9D1054622). We are grateful to the Korea Association Resource (KARE) project, funded by the Korean National Institute of Health, Republic of Korea, for permission to use data. The authors have no conflict of interest.
- Manolio TA, Collins FS, Cox NJ, Goldstein DB, Hindorff LA, Hunter DJ, McCarthy MI, Ramos EM, Cardon LR, Chakravarti A: Finding the missing heritability of complex diseases. Nature. 2009, 461 (7265): 747-753. 10.1038/nature08494.PubMedPubMed CentralView ArticleGoogle Scholar
- Murcray CE, Lewinger JP, Gauderman WJ: Gene-environment interaction in genome-wide association studies. Am J Epidemiol. 2009, 169 (2): 219-226.PubMedPubMed CentralView ArticleGoogle Scholar
- Frazer KA, Murray SS, Schork NJ, Topol EJ: Human genetic variation and its contribution to complex traits. Nat Rev Genet. 2009, 10 (4): 241-251.PubMedView ArticleGoogle Scholar
- Falconer DS, Mackay TFC, Frankham R: Introduction to quantitative genetics (4th edn). Trends Genet. 1996, 12 (7): 280-10.1016/0168-9525(96)81458-2.View ArticleGoogle Scholar
- Via S, Lande R: Genotype-environment interaction and the evolution of phenotypic plasticity. Evol. 1985, 39 (3): 505-522. 10.2307/2408649.View ArticleGoogle Scholar
- Fuller T, Sarkar S, Crews D: The use of norms of reaction to analyze genotypic and environmental influences on behavior in mice and rats. Neurosci Biobehav Rev. 2005, 29 (3): 445-456. 10.1016/j.neubiorev.2004.12.005.PubMedView ArticleGoogle Scholar
- Ottman R: Gene–environment interaction: definitions and study designs. Prev Med. 1996, 25 (6): 764-10.1006/pmed.1996.0117.PubMedPubMed CentralView ArticleGoogle Scholar
- Leve LD, Kerr DCR, Shaw D, Ge X, Neiderhiser JM, Scaramella LV, Reid JB, Conger R, Reiss D: Infant pathways to externalizing behavior: evidence of Genotype × Environment interaction. Child Dev. 2010, 81 (1): 340-356. 10.1111/j.1467-8624.2009.01398.x.PubMedPubMed CentralView ArticleGoogle Scholar
- Hunter DJ: Gene–environment interactions in human diseases. Nat Rev Genet. 2005, 6 (4): 287-298.PubMedView ArticleGoogle Scholar
- Dempfle A, Scherag A, Hein R, Beckmann L, Chang-Claude J, Schäfer H: Gene–environment interactions for complex traits: definitions, methodological requirements and challenges. Eur J Hum Genet. 2008, 16 (10): 1164-1172. 10.1038/ejhg.2008.106.PubMedView ArticleGoogle Scholar
- Maier LA: Genetic and exposure risks for chronic beryllium disease. Clin Chest Med. 2002, 23 (4): 827-10.1016/S0272-5231(02)00029-1.PubMedView ArticleGoogle Scholar
- Memisoglu A, Hu FB, Hankinson SE, Manson JAE, De Vivo I, Willett WC, Hunter DJ: Interaction between a peroxisome proliferator-activated receptor γ gene polymorphism and dietary fat intake in relation to body mass. Hum Mol Genet. 2003, 12 (22): 2923-2929. 10.1093/hmg/ddg318.PubMedView ArticleGoogle Scholar
- Clausen J, Keck D, Hiesey W: Experimental studies on the nature of species. III. Environresponses of climatic races of Achillea. Experimental Studies on the Nature of Species III Environresponses of Climatic Races of Achillea. 1948, Carnegie Institution of Washington, Publ. 581Google Scholar
- Bouchard C: Gene–environment interactions in the etiology of obesity: defining the fundamentals. Obesity. 2008, 16: S5-S10.PubMedView ArticleGoogle Scholar
- Visscher PM, Hill WG, Wray NR: Heritability in the genomics era—concepts and misconceptions. Nat Rev Gen. 2008, 9 (4): 255-266. 10.1038/nrg2322.View ArticleGoogle Scholar
- Satagopan JM, Elston RC: Evaluation of removable statistical interaction for binary traits. Stat Med. 2012, 32 (7): 1164-1190.PubMedPubMed CentralView ArticleGoogle Scholar
- An P, Mukherjee O, Chanda P, Yao L, Engelman CD, Huang CH, Zheng T, Kovac IP, Dubé MP, Liang X: The challenge of detecting epistasis (G × G interactions): genetic analysis workshop 16. Gen Epidemiol. 2009, 33 (S1): S58-S67. 10.1002/gepi.20474.View ArticleGoogle Scholar
- Cho YS, Go MJ, Kim YJ, Heo JY, Oh JH, Ban H-J, Yoon D, Lee MH, Kim D-J, Park M: A large-scale genome-wide association study of Asian populations uncovers genetic factors influencing eight quantitative traits. Nat Gen. 2009, 41 (5): 527-534. 10.1038/ng.357.View ArticleGoogle Scholar
- Yang J, Lee T, Kim J, Cho M-C, Han B-G, Lee J-Y, Lee H-J, Cho S, Kim H: Ubiquitous polygenicity of human complex traits: genome-wide analysis of 49 traits in Koreans. PLoS Gen. 2013, 9 (3): e1003355-10.1371/journal.pgen.1003355.View ArticleGoogle Scholar
- Yang J, Lee SH, Goddard ME, Visscher PM: GCTA: a tool for genome-wide complex trait analysis. Am J Hum Gen. 2011, 88 (1): 76-82. 10.1016/j.ajhg.2010.11.011.View ArticleGoogle Scholar
- Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, Maller J, Sklar P, De Bakker PI, Daly MJ: PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Gen. 2007, 81 (3): 559-575. 10.1086/519795.View ArticleGoogle Scholar
- Durnin J, Womersley J: Body fat assessed from total body density and its estimation from skinfold thickness: measurements on 481 men and women aged from 16 to 72 years. Br J Nutr. 1974, 32 (01): 77-97. 10.1079/BJN19740060.PubMedView ArticleGoogle Scholar
- Schapira DV, Kumar NB, Lyman GH, Cox CE: Abdominal obesity and breast cancer risk. Ann Intern Med. 1990, 112 (3): 182-186. 10.7326/0003-4819-112-3-182.PubMedView ArticleGoogle Scholar
- Zheng J-S, Arnett DK, Lee Y-C, Shen J, Parnell LD, Smith CE, Richardson K, Li D, Borecki IB, Ordovás JM: Genome-wide contribution of genotype by environment interaction to variation of diabetes-related traits. PLoS One. 2013, 8 (10): e77442-10.1371/journal.pone.0077442.PubMedPubMed CentralView ArticleGoogle Scholar
- Cornes B, Lind P, Medland S, Montgomery G, Nyholt D, Martin N: Replication of the association of common rs9939609 variant of FTO with increased BMI in an Australian adult twin population but no evidence for gene by environment (G×E) interaction. Int J Obesity. 2008, 33 (1): 75-79.View ArticleGoogle Scholar
- Levy D, Ehret GB, Rice K, Verwoert GC, Launer LJ, Dehghan A, Glazer NL, Morrison AC, Johnson AD, Aspelund T: Genome-wide association study of blood pressure and hypertension. Nat Genet. 2009, 41 (6): 677-687. 10.1038/ng.384.PubMedPubMed CentralView ArticleGoogle Scholar
- Huang DW, Sherman BT, Lempicki RA: Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009, 4 (1): 44-57.View ArticleGoogle Scholar
- Albelda SM, Buck CA: Integrins and other cell adhesion molecules. FASEB J. 1990, 4 (11): 2868-2880.PubMedGoogle Scholar
- Raven P, Johnson G: Biology 6th ed. 2002, NY: McGraw-Hill PublishingGoogle Scholar
- Schwartz MW, Woods SC, Porte D, Seeley RJ, Baskin DG: Central nervous system control of food intake. Nat London. 2000, 404 (6778): 661-671.Google Scholar
- Woods S, Seeley R: Understanding the physiology of obesity: review of recent developments in obesity research. Int J Obes Relat Metab Disord. 2002, 26: S8-10.1038/sj.ijo.0802211.PubMedView ArticleGoogle Scholar
- Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT: Gene ontology: tool for the unification of biology. Nat Genet. 2000, 25 (1): 25-10.1038/75556.PubMedPubMed CentralView ArticleGoogle Scholar
- Ebrahim S, Kinra S, Bowen L, Andersen E, Ben-Shlomo Y, Lyngdoh T, Ramakrishnan L, Ahuja R, Joshi P, Das SM: The effect of rural-to-urban migration on obesity and diabetes in India: a cross-sectional study. PLoS Med. 2010, 7 (4): e1000268-10.1371/journal.pmed.1000268.PubMedPubMed CentralView ArticleGoogle Scholar
- Taylor A, Sandeep M, Janipalli C, Giambartolomei C, Evans D, Kranthi Kumar M, Vinay D, Smitha P, Gupta V, Aruna M: Associations of FTO and MC4R variants with obesity traits in Indians and the role of rural/urban environment as a possible effect modifier. J Obes. 2011, 2011: 7-View ArticleGoogle Scholar
- Eichler EE, Flint J, Gibson G, Kong A, Leal SM, Moore JH, Nadeau JH: Missing heritability and strategies for finding the underlying causes of complex disease. Nat Rev Genet. 2010, 11 (6): 446-450. 10.1038/nrg2809.PubMedPubMed CentralView ArticleGoogle Scholar
- Cho YS, Go MJ, Kim YJ, Heo JY, Oh JH, Ban HJ, Yoon D, Lee MH, Kim DJ, Park M: A large-scale genome-wide association study of Asian populations uncovers genetic factors influencing eight quantitative traits. Nat Genet. 2009, 41 (5): 527-534. 10.1038/ng.357.PubMedView ArticleGoogle Scholar
- Perusse L, Bouchard C: Genotype‒environment interaction in human obesity. Nutr Rev. 1999, 57 (5): 31-38.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.