• obesity phenotypes;
  • principal component analysis;
  • linkage;
  • QTL


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Traditional whole genome linkage scans for obesity were usually performed for a number of correlated obesity related phenotypes separately without considering their correlations. The purpose of this study was to identify quantitative trait loci (QTLs) underlying variations in multiple correlated obesity phenotypes.

We performed principal component analysis (PCA) for four highly correlated obesity phenotypes (body mass index [BMI], fat mass, percentage of fat mass [PFM], and lean mass) in a sample of 427 pedigrees (comprising 3,273 individuals) and generated two independent principal components (PC1 and PC2). A whole genome linkage scan (WGS) was then conducted for PC1 and PC2.

For PC1, the strongest linkage signal was identified on chromosome 20p12 (LOD = 2.67). For PC2, two suggestive linkages were found on 5q35 (LOD = 2.03) and 7p22 (LOD = 2.18).

This study provided evidence supporting several previously identified linkage regions for obesity (e.g., 1p36, 6p23 and 7q34). In addition, our approach by linear combination of highly correlated obesity phenotypes identified several novel QTLs which were not found in genome linkage scans for individual phenotypes.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Obesity is one of the most prevalent disorders in western societies and is associated with cardiovascular diseases, type 2 diabetes, and several kinds of cancers (Haslam & James, 2005; Kopelman, 2000; Must et al. 1999). Obesity results from a positive energy balance when energy intake exceeds energy expenditure (Kopelman, 2000). As a complex trait, obesity is influenced by multiple genetic and environmental factors (Comuzzie & Allison, 1998; Perusse & Bouchard, 1999).

Extensive efforts have been made to search for obesity genes in the past decade. To date, about 61 whole genome linkage scan (WGS) studies in various populations have reported ∼253 quantitative trait loci (QTLs) for obesity and related phenotypes (Rankinen et al. 2006). A number of phenotypic measures (e.g., body mass index [BMI], fat mass, percentage of fat mass [PFM]) have been used to characterize obese status, each characterizing obesity from different aspects. These obesity phenotypes have respective merits as well as limitations in defining obese status and none of them alone is regarded as the gold standard. For example, BMI is widely used as an index to define overweight and obesity due to its ease of measurement. However, BMI cannot always distinguish obese people from muscular people. Fat mass and PFM are relatively more homogeneous and may reflect obese status more accurately. It should be noted that these obesity related phenotypes, i.e., BMI, fat mass and PFM, are highly correlated with each other and may share common genetic factors (Deng et al. 2002; Kirchengast et al. 2001). Body lean mass is another major component of the human body and it has been used as an independent phenotype in studies of metabolic syndrome, obesity and sarcopenia (Rankinen et al. 2006). So far, most linkage scan studies have been performed for these phenotypes individually. To investigate the genetic factors that may have a common contribution to a clustering of obesity related phenotypes and maximize the chance of identifying obesity genes, it is necessary to adopt an alternative measure that characterizes the common features of these highly correlated phenotypes. Principal component analysis (PCA) is a dimensionality-reduction method for related multivariate analysis. PCA can be a useful way to study multiple correlated phenotypes and has been successfully applied to linkage analysis for correlated traits (Adeyemo et al. 2005; Arya et al. 2002; Karasik et al. 2004).

In this study, we performed PCA for four obesity-related phenotypes (BMI, fat mass, PFM and lean mass) in a large sample of 427 Caucasian pedigrees (comprising 3,273 subjects) and extracted two principal component scores (PC1 and PC2). We then performed WGS for PC1 and PC2 in the total sample to identify obesity QTLs.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References


The study was approved by the Creighton University Institutional Review Board. All the study subjects signed informed-consent documents before entering the project. The subjects came from an expanding database being created for studies to search for genes underlying the risk of osteoporosis and obesity, which is underway in the Osteoporosis Research Center (ORC) at Creighton University. All the study subjects were Caucasians of European origin. Subjects having diseases/conditions that may potentially affect body composition (bone, fat and lean mass) were excluded from the analysis (Deng et al. 2002). Briefly, the exclusion criteria include a history of uncontrolled chronic diseases of the vital organs (e.g., chronic liver disease, lung disease, and heart diseases) and metabolic diseases (e.g., hyperparathyroidism and hyperthyroidism) or musculoskeletal conditions (i.e., osteomalacia and rheumatoid arthritis), which could be non-genetic causes of abnormal body composition. The sample used for this study contained 427 pedigrees comprising 3,273 individuals.


The obesity phenotypes used here are BMI, fat mass, PFM, and lean mass. BMI was recorded as body weight (in kilograms) divided by the square of height (in meters). Fat mass and lean mass were measured using Hologic 2000+ or 4500 dual energy X-ray absorptiometry (DXA) scanner (Hologic Corporation, Waltham MA) in the ORC. The PFM was defined as the ratio of fat mass to the total body weight obtained from DXA (i.e., the sum of fat mass, lean mass and bone mass). The precision of BMI, fat mass, PFM, and lean mass, as reflected by coefficient of variation, were 0.2%, 2.2%, 2.2%, and 1.0%, respectively, for measurements on Hologic 2000+, and were 2.0%, 1.2%, 1.1%, and 0.7%, respectively, for measurements on Hologic 4500. Members of the same pedigree were usually measured on the same type of machine. For each study subject, information on age, gender, ethnic background, medical history, family history, physical activity, alcohol use, diet habits, smoking history, etc. were assessed by a questionnaire.


For each subject, DNA was extracted from peripheral blood leukocytes using the Puregene DNA isolation kit (Gentra Systems, Inc., Minneapolis, MN). All subjects were genotyped for 410 microsatellite markers (including 18 markers on chromosome X) from the Marshfield screening set 14 by Marshfield Center for Medical Genetics. These markers have an average population heterozygosity of 0.75 and cover the genome at an average density of 8.9 centimorgans (CM). The detailed genotyping protocol is available at Pedcheck (O'Connell & Weeks, 1998) was used to check the conformity of Mendelian inheritance pattern at all the marker loci and the relationships of family members within pedigrees. The overall genotyping error rate was ∼0.3%.

Statistical Analyses

We first performed PCA for BMI, fat mass, PFM and lean mass. Two principal component scores (PC1 and PC2) which are linear combinations of the original set of variables (i.e., BMI, fat mass, lean mass and PFM) were generated, both following normal distribution.

Using the SOLAR program (Almasy & Blangero, 1998), we calculated genetic and environmental correlations (ρG and ρE, respectively) between pairs of obesity phenotypes by performing bivariate quantitative genetic analysis. Phenotypic correlation (ρP) was calculated by the following equation: ρP=inline image where h21 and h22 are heritabilities of traits 1 and 2. The significant of ρG and ρE between any pair of traits was tested using the likelihood ratio statistic.

We then performed variance component linkage analyses for PC1 and PC2. In linkage analyses, effects of age, sex, age×sex, age2, age2×sex were included as covariates and were evaluated in polygenic models, with only the significant covariates (P≤ 0.05) retained in the models. Multipoint linkage analysis was performed for the 22 autosomes. To assess the significance and robustness of the linkage results, we carried out simulation analysis to adjust LOD scores for multipoint linkage analyses (Hamet et al. 2005). After 10,000 simulations by the procedure “lodadj” implemented in SOLAR, all LOD scores were empirically adjusted. Empirical point-wise p values for adjusted LOD scores were also obtained.

For chromosome X, the current version of SOLAR cannot handle multipoint linkage analysis. Instead, we used software package MERLIN (Abecasis et al. 2002) to perform linkage analysis on chromosome X. However, MERLIN could not process some large pedigrees as used in this study. Hence, we broke down those large pedigrees into small ones using the software MEGA2 (Mukhopadhyay et al. 2005). A total of 638 nuclear families with size ranging from 4 to 12 were generated from the original 427 pedigrees, and multipoint linkage analysis on chromosome X was conducted in the resulting 638 nuclear families using MERLIN.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Table 1 summarizes the basic characteristics of the study subjects. Our sample contains a large number of relative pairs informative for linkage analyses (Table 2). The correlations of between pairs of obesity phenotypes are shown in Table 3. All of the phenotypic, genetic and environmental correlations among four obesity phenotypes are significant (P≤ 0.01).

Table 1.  Basic characteristics of the study subjects
 Total (n = 3,273)Female (n = 1,948)Male (n = 1,325)
  1. For each trait, data are presented as mean ± SD.

Age (years)46.62 ± 15.9345.96 ± 15.6047.59 ± 16.36
BMI (kg/m2)27.21 ± 5.2326.68 ± 5.7327.98 ± 4.29
Fat mass (kg)24.87 ± 9.7226.37 ± 10.3122.67 ± 8.29
PFM (%)31.24 ± 8.4835.59 ± 7.0824.89 ± 6.03
Lean Mass (kg)54.31 ± 12.7045.92 ± 6.7866.65 ± 8.65
Table 2.  The informative relationships for linkage analyses
Number of pedigrees427
Number of genotyped subjects3273
Relative pairs 
 First cousins14274
 Second cousins25969
 First cousins, 1 removed19473
 Second cousins, 1 removed23630
Table 3.  The correlations of between pairs of obesity phenotypes
  1. ρP, ρG and ρE denote phenotypic, genetic and environmental correlation, respectively. *P≤ 0.01; **P≤ 0.001.

BMI ×fat mass0.868*0.876 ± 0.013**0.941 ± 0.005**
BMI ×lean mass0.698*0.659 ± 0.029**0.749 ± 0.020**
BMI ×PFM0.739*0.702 ± 0.031**0.770 ± 0.017**
Fat mass ×lean mass0.611*0.604 ± 0.034**0.631 ± 0.026**
Fat mass ×PFM0.897*0.891 ± 0.133**0.903 ± 0.007**
Lean mass ×PFM0.287*0.213 ± 0.053**0.368 ± 0.037**

PCA for Obesity Phenotypes

PCA analysis on the four correlated obesity phenotypes extracted two uncorrelated components PC1 and PC2 which together explain 96.9% of the total variation (Table 4). PC1 explains 61.8% of the total variation and the loadings of PC1 are 0.90, 0.98, 0.82, and 0.18 for BMI, fat mass, PFM, and lean mass, respectively. PC2 explains 35.0% of the total variation, and the loading is 0.97 for lean mass. PC1 and PC2 were used as independent phenotypes for the following linkage analysis.

Table 4.  Principal component analysis for obesity phenotypes
Obesity phenotypeComponent1Component2
  1. Principal component loadings are shown.

BMI0.90 0.36 
Fat mass0.98−0.04 
Lean mass0.18 0.97 
Eigenvalue2.47 1.40 
Explained variance61.84% 35.03%

Single-Locus WGS for PC1 and PC2

Quantitative genetic analysis showed that the estimated heritabilities (±SE) of PC1 and PC2 were 0.44 (±0.03) and 0.61 (±0.03), respectively, after adjustment for significant covariates (age, sex and age2 for PC1; age, sex, age2 and age2×sex for PC2).

Figure 1 illustrates the multipoint linkage results on autosomes for PC1 and PC2. The major multipoint linkage findings with LOD>1.0 are presented in Table 5. Suggestive linkage signals (including chromosome X) are displayed in Figure 2. We adopted the empirical thresholds of LOD of 1.9 and 3.3 for suggestive and significant linkage respectively in pedigree-based WGS analysis (Lander & Kruglyak, 1995). We found suggestive linkages on 20p12 (LOD = 2.67, P = 0.0003) for PC1, 5q35 (LOD = 2.03, P = 0.0008) and 7p22 (LOD = 2.18, P = 0.0006) for PC2, and a suggestive linkage on Xq28 (LOD = 2.82) for PC2.


Figure 1. Results of multipoint WGS for PC1 and PC2 on autosomes.

Download figure to PowerPoint

Table 5.  Results of multipoint linkage analyses for PCs
PhenotypeLOD scoreChromosomal regionMarkerLocation (cM)P-valueLod score or p valuePrevious studies
  1. LOD, logarithm of odds; MLS, maximum LOD score; NPL, non-parametric linkage.

PC11.312q36.2–2q36.3GATA23D03N2310.0089LOD = 2.4BMI(Tang et al. 2003)
1.726p23ATA50C05250.0038MLS = 1.94Factor central obesity(Kraja et al. 2005)
LOD = 1.72Obesity under anti-psychotics(Chagnon et al. 2004)
1.196p21.3GATA61E03510.0123P= 0.01Skinfolds (subscapular, triceps)(Wilson et al. 1991)
P= 0.004Body weight(Elosua et al. 2003)
P= 0.002Body fat (%)(Norman et al. 1995)
1.407p22.1–7p22.2GATA24F0390.0070LOD = 2.53BMI(Palmer et al. 2003)
1.768q13GATA41A01820.0034NPL = 1.9BMI≥35(Li et al. 2004)
NPL = 1.9BMI≥35(Dong et al. 2005)
LOD = 2.06Fat mass(Zhao et al. 2007)
1.1110p12–10p13GATA70E11410.0146P= 0.005BMI(Gorlova et al. 2003)
1.4217q11.2GGAA9D03510.0067P= 0.001BMI(Mitchell et al. 1999)
2.6720p12ATTC013280.0003LOD = 4.08BMI(Gorlova et al. 2003)
MLS = 2.46Factor central obesity(Kraja et al. 2005)
PC21.631p36.2GATA27E01300.0023MLS = 1.38BMI(Liu et al. 2004)
LOD = 2.09BMI(Deng et al. 2002)
LOD = 2.2BMI(Stone et al. 2002)
P= 0.03Skinfolds, suprailiac(Wilson et al. 1991)
1.574q28.2ATA26B081310.0028LOD = 1.71BMI(Dong et al. 2005)
2.035q35.2AAT0131840.0008P= 0.0006BMI(Gorlova et al. 2003)
MLS = 1.87Factor central obesity(Kraja et al. 2005)
LOD = 1.8BMI(Feitosa et al. 2002)
P= 0.0039BMI(Platte et al. 2003)
2.187p22.3AFMb035xb910.0006LOD = 2.53BMI(Palmer et al. 2003)
1.517q34GATA1041560.0036P= 0.0001BMI(Platte et al. 2003)
P= 0.0001BMI(Duggirala et al. 1996)

Figure 2. Suggestive linkage regions detected in the single-locus WGS. (A) Linkage results of chromosome 20 for PC1. (B) Linkage results of chromosome 5 for PC2. (C) Linkage results of chromosome 7 for PC2. (D) Two-point linkage signals on chromosome X for PC2.

Download figure to PowerPoint


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

In this study we performed PCA for four highly correlated obesity phenotypes and extracted two factors, PC1 and PC2, which explain the common feature of these phenotypes. We then conducted WGS for PC1 and PC2 scores to search for QTLs underlying multiple correlated obesity phenotypes. The multipoint linkage analyses identified linkages on 20p12 for PC1, and 5q35 and 7p22 for PC2. Additional PCA with varimax rotation was performed to see if this may result in higher LOD scores. The reanalysis generated factors and loadings that were essentially quite similar to the present. There were no significant differences for LOD scores between the original PCA and PCA with varimax rotation.

PCA is a dimensionality-reduction method for multivariate analysis, which can transform a set of correlated variables into a substantially smaller set of uncorrelated variables (principal components) that capture major information of original data. In this study, we found significant phenotypic and genetic correlations between pairs of obesity phenotypes. Our PCA results showed that about 62% of the total variation of obesity phenotypes can be explained by PC1, with relatively high loading for BMI, fat mass and PFM, indicating that PC1 is a major index reflecting body fat mass. PC2 explained 35% of the total variation with the highest loading for lean mass, suggesting that PC2 is an index mainly reflects body lean mass.

In this study, 20p12 showed evidence of linkage to PC1 (LOD = 2.67). Linkage of this region to BMI has also been found in previous studies (Gorlova et al. 2003; Kraja et al. 2005; Zhao et al. 2007). In our earlier WGS study with the same sample for individual obesity phenotypes, Zhao et al. (2007) found the most pronounced linkage on 20p11–12 for fat mass and PFM. Gorlova et al. (2003) also reported linkage to BMI in children on chromosome 20p11.2-pter near the marker D20S851 (LOD = 4.08, p = 0.000046). Within the region 20p12, MKKS (McKusick- Kaufman syndrome or Bardet-Biedl syndrome 6, BBS6) is an important candidate gene, and mutations in the MKKS gene may cause obesity-related Mendelian disorders (Katsanis et al. 2000; Slavotinek et al. 2000; Stone et al. 1998; Stone et al. 2000). Thus, it is likely that there exist important obesity genes on chromosome 20p11–12.

The genomic regions 5q35 and 7p22 reached suggestive linkage for PC2. The importance of 5q35 on obesity phenotypes has also been suggested in our previous WGS study for individual phenotypes (Zhao et al. 2007) and studies of other groups (Feitosa et al. 2002; Gorlova et al. 2003; Platte et al. 2003). Zhao et al. (2007) reported suggestive linkage for BMI, lean mass and male-specific lean mass on 5q35, which is exactly the same linkage region for PC2 in the present study. Gorlova et al. (2003) also found linkage on 5q35 (LOD = 2.48, P = 0.0006) for children with no parent-of-origin effect. 5q35 harbors the pituitary tumor transforming (PTTG) gene, which is indispensable for normal pancreatic beta cell proliferation and mutations in the gene may impair glucose homeostasis and lead to diabetes (Wang et al. 2003).

Importantly, our current study identified a novel linkage on 7p22 for PC2, which was not found by Zhao et al. (2007) for individual phenotypes. Interestingly, the linkage on 7p22 was reported by Palmer et al. (2003) in which a peak multipoint LOD score of 2.53 was achieved for BMI, flanking the marker D7S2477 (the same one as in our study). This suggests that, under certain situations, WGS using PCA scores may offer higher power in linkage mapping of complex diseases/traits. Other potentially interesting linkage findings of this study were 4q35.2 (LOD = 1.70) and 6q14.1 (LOD = 1.25) which were not found in any other WGS studies. The importance of the two regions awaits further replication/confirmation studies. 6q14.1 contains the serotonin 5-HT1B receptor gene (HTR1B) and polymorphisms of the HTR1B gene were associated with minimum lifetime BMI in women with bulimia nervosa (Levitan et al. 2001). Our study reported here suggests that PCA may provide additional information which may not be captured by individual phenotypes. Thus, WGS on PCA factors could be useful in yielding additional information to those of univariate WGS.

Although a host of QTLs were identified for obesity, only a few have been replicated in two or more studies (Rankinen et al. 2006). Our results here provide evidence of replication for some earlier identified linkages (e.g., 1p36, 2q36, 4q28.2, 6p23, 6p21.3, 7p22, 7q34, 8q13.1 and 17q11.2) (summarized in Table 5) (Chagnon et al. 2004; Deng et al. 2002; Dong et al. 2005; Duggirala et al. 1996; Elosua et al. 2003; Heilbronn et al. 2000; Li et al. 2004; Liu et al. 2004; Mitchell et al. 1999; Norman et al. 1995; Stone et al. 2002; Tang et al. 2003; Wilson et al. 1991). For example, in our previous studies in subsamples, both Deng et al. (2002) and Liu et al. (2004) found that 1p36 is an important region that may harbor QTLs for obesity. This is consistent with our results of this study for PC2. It is also interesting that some linkage regions for PC2 in this study (e.g. 1p36.2, 4q28.2 and 5q35.2) coincide with those for BMI in some other WGS studies. For example, studies of BMI (Gorlova et al. 2003; Kraja et al. 2005; Zhao et al. 2007) showed linkage to 5q35, where linkage was found for PC2 in this study and for lean mass in our earlier univariate analyses (Zhao et al. 2007). This suggests that 5q35 might be a locus influencing lean mass rather than adiposity.

This study has several strengths. First, our sample contains more than 158,000 informative relationships and is powerful for linkage analysis. Second, we employed PCA to extract useful information from multiple obesity phenotypes which are highly correlated. PCA is a dimension reducing method to decrease the likelihood of type I error rate by avoiding multiple testing due to separate analysis for correlated obesity phenotypes (Holberg et al. 2001). Third, compared with the original variables, PCA can extract PCs which are more likely to follow normal distribution (Boomsma & Dolan, 1998). This is important for linkage analyses because the assumption of normal distribution of phenotypes is required by variance component analysis to achieve robust and reliable results (Allison et al. 1998). Fourth, the analysis of combined phenotypes may detect genomic regions that analyses of individual phenotypes may miss.

The choice of the traits for PCA analysis may influence the linkage results. In this study, the principal components were derived from BMI, fat mass, PFM, and lean mass. It may be interesting to conduct PCA on the more directly measured traits such as weight, fat mass, lean mass and bone mass to investigate if a different constellation of body size traits can be identified and how the linkage results can be influenced. Based on this consideration, we performed further PCA on weight, fat mass, lean mass and bone mass (with height included as a covariate) and generated one principal component, for which we performed linkage analyses. Two suggestive linkages were found on 20p12 (LOD = 2.05) and 4q28 (LOD = 1.90), which are acutally linked to PC1 and PC2, respectively. This is not surprising, because the principle component derived from direct measures may capture the major features of PC1 and PC2, and thus the identified linkage regions may partially overlap with those for PC1 and/or PC2.

In summary, we performed a WGS for two components extracted from PCA for four obesity-related phenotypes. Linkages were identified on genomic regions of 20p12, 7p22 and 5q35. Together with the findings from previous studies, the current study has provided additional insight into the genetic basis of obesity.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Investigators of this work were partially supported by grants from NIH ((R01 AR050496-01, R21 AG027110, R01 AG026564, and P50 AR055081) and an LB595 grant from the State of Nebraska. The study was also benefited from grant 30570875 from National Science Foundation of China, Xi'an Jiaotong University, and the Ministry of Education of China Huo Ying Dong Education Foundation, Hunan Province and Hunan Normal University. The genotyping experiment was performed by Marshfield Center for Medical Genetics and supported by NHLBI Mammalian Genotyping Service (Contract Number HV48141).


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Abecasis, G. R., Cherny, S. S., Cookson, W. O. & Cardon, L. R. (2002) Merlin–rapid analysis of dense genetic maps using sparse gene flow trees. Nat Genet 30, 97101.
  • Adeyemo, A. A., Johnson, T., Acheampong, J., Oli, J., Okafor, G., Amoah, A., Owusu, S., Agyenim-Boateng, K., Eghan, B. A., Jr., Abbiyesuku, F., Fasanmade, O., Rufus, T., Doumatey, A., Chen, G., Zhou, J., Chen, Y., Furbert-Harris, P., Dunston, G., Collins, F. & Rotimi, C. (2005) A genome wide quantitative trait linkage analysis for serum lipids in type 2 diabetes in an African population. Atherosclerosis 181, 389397.
  • Allison, D. B., Thiel, B., St Jean, P., Elston, R. C., Infante, M. C. & Schork, N. J. (1998) Multiple phenotype modeling in gene-mapping studies of quantitative traits: power advantages. Am J Hum Genet 63, 11901201.
  • Almasy, L. & Blangero, J. (1998) Multipoint quantitative-trait linkage analysis in general pedigrees. Am J Hum Genet 62, 11981211.
  • Arya, R., Blangero, J., Williams, K., Almasy, L., Dyer, T. D., Leach, R. J., O'Connell, P., Stern, M. P. & Duggirala, R. (2002) Factors of insulin resistance syndrome–related phenotypes are linked to genetic locations on chromosomes 6 and 7 in nondiabetic mexican-americans. Diabetes 51, 841847.
  • Boomsma, D. I. & Dolan, C. V. (1998) A comparison of power to detect a QTL in sib-pair data using multivariate phenotypes, mean phenotypes, and factor scores. Behav Genet 28, 329340.
  • Chagnon, Y. C., Merette, C., Bouchard, R. H., Emond, C., Roy, M. A. & Maziade, M. (2004) A genome wide linkage study of obesity as secondary effect of antipsychotics in multigenerational families of eastern Quebec affected by psychoses. MolPsychiatry 9, 10671074.
  • Comuzzie, A. G. & Allison, D. B. (1998) The search for human obesity genes. Science 280, 13741377.
  • Deng, H. W., Deng, H., Liu, Y. J., Liu, Y. Z., Xu, F. H., Shen, H., Conway, T., Li, J. L., Huang, Q. Y., Davies, K. M. & Recker, R. R. (2002) A genomewide linkage scan for quantitative-trait loci for obesity phenotypes. Am J Hum Genet 70, 11381151.
  • Dong, C., Li, W. D., Li, D. & Price, R. A. (2005) Interaction between obesity-susceptibility loci in chromosome regions 2p25–p24 and 13q13–q21. Eur J Hum Genet 13, 102108.
  • Duggirala, R., Stern, M. P., Mitchell, B. D., Reinhart, L. J., Shipman, P. A., Uresandi, O. C., Chung, W. K., Leibel, R. L., Hales, C. N., O'Connell, P. & Blangero, J. (1996) Quantitative variation in obesity-related traits and insulin precursors linked to the OB gene region on human chromosome 7. Am J Hum Genet 59, 694703.
  • Elosua, R., Demissie, S., Cupples, L. A., Meigs, J. B., Wilson, P. W., Schaefer, E. J., Corella, D. & Ordovas, J. M. (2003) Obesity modulates the association among APOE genotype, insulin, and glucose in men. Obes Res 11, 15021508.
  • Feitosa, M. F., Borecki, I. B., Rich, S. S., Arnett, D. K., Sholinsky, P., Myers, R. H., Leppert, M. & Province, M. A. (2002) Quantitative-trait loci influencing body-mass index reside on chromosomes 7 and 13: the National Heart, Lung, and Blood Institute Family Heart Study. Am J Hum Genet 70, 7282.
  • Gorlova, O. Y., Amos, C. I., Wang, N. W., Shete, S., Turner, S. T. & Boerwinkle, E. (2003) Genetic linkage and imprinting effects on body mass index in children and young adults. Eur J Hum Genet 11, 425432.
  • Hamet, P., Merlo, E., Seda, O., Broeckel, U., Tremblay, J., Kaldunski, M., Gaudet, D., Bouchard, G., Deslauriers, B., Gagnon, F., Antoniol, G., Pausova, Z., Labuda, M., Jomphe, M., Gossard, F., Tremblay, G., Kirova, R., Tonellato, P., Orlov, S. N., Pintos, J., Platko, J., Hudson, T. J., Rioux, J. D., Kotchen, T. A. & Cowley, A. W., Jr. (2005) Quantitative founder-effect analysis of French Canadian families identifies specific loci contributing to metabolic phenotypes of hypertension. Am J Hum Genet 76, 815832.
  • Haslam, D. W. & James, W. P. (2005) Obesity. Lancet 366, 11971209.
  • Heilbronn, L. K., Kind, K. L., Pancewicz, E., Morris, A. M., Noakes, M. & Clifton, P. M. (2000) Association of −3826 G variant in uncoupling protein-1 with increased BMI in overweight Australian women. Diabetologia 43, 242244.
  • Holberg, C. J., Halonen, M., Solomon, S., Graves, P. E., Baldini, M., Erickson, R. P. & Martinez, F. D. (2001) Factor analysis of asthma and atopy traits shows 2 major components, one of which is linked to markers on chromosome 5q. J Allergy Clin Immunol 108, 772780.
  • Karasik, D., Cupples, L. A., Hannan, M. T. & Kiel, D. P. (2004) Genome screen for a combined bone phenotype using principal component analysis: the Framingham study. Bone 34, 547556.
  • Katsanis, N., Beales, P. L., Woods, M. O., Lewis, R. A., Green, J. S., Parfrey, P. S., Ansley, S. J., Davidson, W. S. & Lupski, J. R. (2000) Mutations in MKKS cause obesity, retinal dystrophy and renal malformations associated with Bardet-Biedl syndrome. Nat Genet 26, 6770.
  • Kirchengast, S., Peterson, B., Hauser, G. & Knogler, W. (2001) Body composition characteristics are associated with the bone density of the proximal femur end in middle- and old-aged women and men. Maturitas 39, 133145.
  • Kopelman, P. G. (2000) Obesity as a medical problem. Nature 404, 635643.
  • Kraja, A. T., Rao, D. C., Weder, A. B., Cooper, R., Curb, J. D., Hanis, C. L., Turner, S. T., De Andrade, M., Hsiung, C. A., Quertermous, T., Zhu, X. & Province, M. A. (2005) Two major QTLs and several others relate to factors of metabolic syndrome in the family blood pressure program. Hypertension 46, 751757.
  • Lander, E. & Kruglyak, L. (1995) Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat Genet 11, 241247.
  • Levitan, R. D., Kaplan, A. S., Masellis, M., Basile, V. S., Walker, M. L., Lipson, N., Siegel, G. I., Woodside, D. B., Macciardi, F. M., Kennedy, S. H. & Kennedy, J. L. (2001) Polymorphism of the serotonin 5-HT1B receptor gene (HTR1B) associated with minimum lifetime body mass index in women with bulimia nervosa. Biol Psychiatry 50, 640643.
  • Li, W. D., Dong, C., Li, D., Zhao, H. & Price, R. A. (2004) An obesity-related locus in chromosome region 12q23–24. Diabetes 53, 812820.
  • Liu, Y. J., Xu, F. H., Shen, H., Liu, Y. Z., Deng, H. Y., Zhao, L. J., Huang, Q. Y., Dvornyk, V., Conway, T., Davies, K. M., Li, J. L., Recker, R. R. & Deng, H. W. (2004) A follow-up linkage study for quantitative trait loci contributing to obesity-related phenotypes. J Clin Endocrinol Metab 89, 875882.
  • Mitchell, B. D., Cole, S. A., Comuzzie, A. G., Almasy, L., Blangero, J., MacCluer, J. W. & Hixson, J. E. (1999) A quantitative trait locus influencing BMI maps to the region of the beta-3 adrenergic receptor. Diabetes 48, 18631867.
  • Mukhopadhyay, N., Almasy, L., Schroeder, M., Mulvihill, W. P. & Weeks, D. E. (2005) Mega2: data-handling for facilitating genetic linkage and association analyses. Bioinformatics 21, 25562557.
  • Must, A., Spadano, J., Coakley, E. H., Field, A. E., Colditz, G. & Dietz, W. H. (1999) The disease burden associated with overweight and obesity. JAMA 282, 15231529.
  • Norman, R. A., Bogardus, C. & Ravussin, E. (1995) Linkage between obesity and a marker near the tumor necrosis factor-alpha locus in Pima Indians. J Clin Invest 96, 158162.
  • O'Connell, J. R. & Weeks, D. E. (1998) PedCheck: a program for identification of genotype incompatibilities in linkage analysis. Am J Hum Genet 63, 259266.
  • Palmer, L. J., Buxbaum, S. G., Larkin, E., Patel, S. R., Elston, R. C., Tishler, P. V. & Redline, S. (2003) A whole-genome scan for obstructive sleep apnea and obesity. Am J Hum Genet 72, 340350.
  • Perusse, L. & Bouchard, C. (1999) Genotype-environment interaction in human obesity. Nutr Rev 57, S31S37.
  • Platte, P., Papanicolaou, G. J., Johnston, J., Klein, C. M., Doheny, K. F., Pugh, E. W., Roy-Gagnon, M. H., Stunkard, A. J., Francomano, C. A. & Wilson, A. F. (2003) A study of linkage and association of body mass index in the Old Order Amish. Am J Med Genet C Semin Med Genet 121, 7180.
  • Rankinen, T., Zuberi, A., Chagnon, Y. C., Weisnagel, S. J., Argyropoulos, G., Walts, B., Perusse, L. & Bouchard, C. (2006) The human obesity gene map: the 2005 update. Obesity (Silver Spring) 14, 529644.
  • Slavotinek, A. M., Stone, E. M., Mykytyn, K., Heckenlively, J. R., Green, J. S., Heon, E., Musarella, M. A., Parfrey, P. S., Sheffield, V. C. & Biesecker, L. G. (2000) Mutations in MKKS cause Bardet-Biedl syndrome. Nat Genet 26, 1516.
  • Stone, D. L., Agarwala, R., Schaffer, A. A., Weber, J. L., Vaske, D., Oda, T., Chandrasekharappa, S. C., Francomano, C. A. & Biesecker, L. G. (1998) Genetic and physical mapping of the McKusick-Kaufman syndrome. Hum Mol Genet 7, 475481.
  • Stone, D. L., Slavotinek, A., Bouffard, G. G., Banerjee-Basu, S., Baxevanis, A. D., Barr, M. & Biesecker, L. G. (2000) Mutation of a gene encoding a putative chaperonin causes McKusick-Kaufman syndrome. Nat Genet 25, 7982.
  • Stone, S., Abkevich, V., Hunt, S. C., Gutin, A., Russell, D. L., Neff, C. D., Riley, R., Frech, G. C., Hensel, C. H., Jammulapati, S., Potter, J., Sexton, D., Tran, T., Gibbs, D., Iliev, D., Gress, R., Bloomquist, B., Amatruda, J., Rae, P. M., Adams, T. D., Skolnick, M. H. & Shattuck, D. (2002) A major predisposition locus for severe obesity, at 4p15–p14. Am J Hum Genet 70, 14591468.
  • Tang, W., Miller, M. B., Rich, S. S., North, K. E., Pankow, J. S., Borecki, I. B., Myers, R. H., Hopkins, P. N., Leppert, M. & Arnett, D. K. (2003) Linkage analysis of a composite factor for the multiple metabolic syndrome: the National Heart, Lung, and Blood Institute Family Heart Study. Diabetes 52, 28402847.
  • Wang, Z., Moro, E., Kovacs, K., Yu, R. & Melmed, S. (2003) Pituitary tumor transforming gene-null male mice exhibit impaired pancreatic beta cell proliferation and diabetes. Proc Natl Acad Sci USA 100, 34283432.
  • Wilson, A. F., Elston, R. C., Tran, L. D. & Siervogel, R. M. (1991) Use of the robust sib-pair method to screen for single-locus, multiple-locus, and pleiotropic effects: application to traits related to hypertension. Am J Hum Genet 48, 862872.
  • Zhao, L. J., Xiao, P., Liu, Y. J., Xiong, D. H., Shen, H., Recker, R. R. & Deng, H. W. (2007) A genome-wide linkage scan for quantitative trait loci underlying obesity related phenotypes in 434 Caucasian families. Hum Genet 121(1), 1458.