Pacific Symposium on Biocomputing 11:467-477(2006)

EFFECT OF THE PEROXISOME PROLIFERATORSACTIVATED RECEPTOR (PPAR) GAMMA 3 GENE ON BMI IN 1,210 SCHOOL STUDENTS FROM MORELOS, MEXICO *
LINA CHEN Department of Social Medicine, University of Bristol, Bristol, UK H. EDUARDO VELASCO MONDRAGÓN National Institute of Public Health, Cuernavaca, Morelos, Mexico EDUARDO LAZCANO-PONCE National Institute of Public Health, Cuernavaca, Morelos, Mexico ANDREW COLLINS Human Genetics Research Division, University of Southampton, UK YIN YAO SHUGART Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, USA yyao@jhsph.edu Department of Social Medicine, University of Bristol, Bristol, UK

Little research has been undertaken on risk factors for obesity in young people in Latin America, including Mexico, despite the fact that obesity constitutes the number one public health problem in Mexico. Our objective was to investigate the effect of the Peroxisome proliferators-activated receptor (PPAR)_3 gene on BMI measured among adolescents collected from a cohort study originally designed for epidemiological studies. METHODS: Blood samples and anthropometric measurements were collected from 1,210 out of 13,294 public school students of both sexes, aged 11-24 years in Morelos, Mexico. In this study, we genotyped 7 selected SNPs of the PPAR_ transcript variant 3 (including Pro12Ala) in a group of unrelated 717 males and 493 females (age range 11-24), including 3 SNPs located in the 5' untranslated region. These 7 SNPs were selected by the tagging algorithm implemented in the program haploview to scan the whole gene. We tested each of the 7 SNPs individually for association with the body mass index (BMI), and two SNPs (rs2938392 and rs1175542) revealed significant associations with BMI (p-value=0.008 and 0.029, respectively). The SNP rs2938392 is roughly 41.5 Kb from rs1801282 (Pro12Ala in PPAR_2). Furthermore, we examined the association between haplotypes built from 7 SNPs and BMI using a score statistic implemented in the program haplo.stats. While the permutation based global p-value was 0.544, one individual haplotype with a frequency of 0.279 gave a p-value of 0.089
* 

This work is supported by Glaxo-Smith-Kline Epidemiology This work is partially supported by University of Bristol


Pacific Symposium on Biocomputing 11:467-477(2006)

(permutation based). However, when the analyses were conducted in males only, the permutation based global p-value was 0.055 and one individual haplotype with a frequency of 0.28 gave a significant p-value of 0.013.

1.

Introduction

The goal of this study was to investigate the effect of the Peroxisome proliferators-activated receptor (PPAR)_3 gene on Body Mass Index (BMI) measured among Mexican adolescents. Although it has been established that both PPAR_2 and PPAR_3 are expressed uniquely in colon and adipocyte tissue and their potential role in the metabolic disorders such as obesity and type 2 diabetes has been suggested [1], the reported effects of the Pro12Ala polymorphism on susceptibility for obesity have been inconsistent [2]. The detailed findings from all positive or negative studies are given in Table 1.
Table 1 A summary of Pro12Ala findings in different studies Ref Sample size [3] [3] [4] [4] [5] [6] [6] [7] [7] [8] [9] [9] [10] [10] [11] [11] [12] [12] [12] [13] [13] [14] [15] [16] [17] [18] Ethnicity P-value 0.01 <0.001 0.027 0.015 0.65 >0.05 >0.05 0.008 0.005 0.011 0.8 0.9 >0.05 >0.05 0.67 0.71 >0.05 >0.05 >0.05 0.15 0.10 0.017 >0.05 0.011 0.3 0.64

517 being lean-to-moderately obese Caucasian 169 very obese Caucasian 333 non-diabetic Scandinavian 973 non-diabetic Scandinavian 215 men Asian 296 extremely obese Caucasian 130 underweight Caucasian 752 obese Caucasian 869 non-obese Caucasian 141 obese Scandinavian 131 diabetic Caucasian 312 normoglycemic Caucasian 1025 diabetic Caucasian 310 with normal BMI Caucasian 108 non-diabetic Caucasian 19 overweight Caucasian 295 non-diabetic non-obese Caucasian 372 morbidly obese Caucasian 402 diabetic Caucasian 541 non-diabetic Asian 415 diabetic subjects Asian 165 obese Caucasian 229 Asian 921 Caucasian 476 Scandinavian 675 men Caucasian


Pacific Symposium on Biocomputing 11:467-477(2006)

ID [19] [19] [19] [20] [20] [20] [21] [22] [23] [23] [24] [24] [25] [25] [26] [27] [28] [29] [30] [30] [31] [31] [32] [32] [33] [34] [34] [35] [35] [35]

Sample size

Ethnicity

P-value 0.070 0.034 0.080 >0.05 0.32 <0.05 0.035 >0.05 0.881 0.846 0.89 0.47 0.554 0.678 0.31 >0.05 0.3 >0.05 >0.05 >0.05 >0.05 >0.05 0.09 >0.05 0.005 >0.05 p<0.05 0.566 0.875 0.037

228 with normal BMI Scandinavian 217 with dyslipidemia Scandinavian 649 without dyslipidemia Scandinavian 280 with normal BMI Caucasian 95 obese Caucasian 42 young obese Caucasian 619 Caucasian 453 from 10 families Mexican 2201 diabetics Asian 1212 with normal BMI Asian 292 obese Caucasian 371 lean Caucasian 259 men Caucasian 333 women Caucasian 124 non-diabetics Caucasian 2245 non-diabetics Scandinavian 1107 diabetics Caucasian 438 Caucasian 478 men Asian 117 women Asian 145 obese Caucasian 317 non-obese Caucasian 210 monozygotic twins Scandinavian 344 dizygotic twins Scandinavian 720 Caucasian 253 with low physical activity level Caucasian 253 with high physical activity level Caucasian 420 diabetic Asian 538 with impaired glucose tolerance Asian 3080 with normal BMI Asian

Few studies have investigated the association between PPAR_ gene and BMI in the Mexican population. The most recent report was from Hsueh et al. 2001 [22], based on the study of 453 subjects comprising of 10 pedigrees. However, no significant effect was found in their study. While most of previous studies have focused on the effect of Pro12Ala or PPAR_2, we have chosen to study PPAR_3, which contains the region of PPAR_2, but also includes a 5' untranslated region of the PPAR_ gene. Interestingly, the expression of PPAR_3 is directed by an independent promoter, and to date, at least three promoters in the upstream region of PPAR_3 have been identified through molecular studies [1]. We analyzed seven


Pacific Symposium on Biocomputing 11:467-477(2006)

tagging SNPs spanning 89.5kb of PPAR_3 gene region, and four of them were located in the long 5'-end untranslated region. Our motivation is to examine the effect of the haplotypes that represent the PPAR_3 genes, rather than focusing on Pro12Ala itself. 2. Methods

2.1. Study Population This study builds on a parent cohort study carried out by Lazcano-Ponce et al. (2003). The parent study was the 1998-1999 baseline measurement of a cohort study of 13,293 students at public junior high and high schools and a state university in the central Mexican State of Morelos. The Research Ethics committee of the National Institute of Public Health approved the study Protocol for epidemiological studies. Further, The IRB review board of Johns Hopkins Bloomberg School of Public Health approved the current study which uses 1,270 anonymous DNA samples out of 13,293 Mexican students to test for association between BMI and SNPs residing in the PPAR3 gene. Table 2 gives information on age and gender in this study population.
Table 2 Characteristics of the study group subjects Variable No. of Subjects Age BMI Male 717 15.56±2.67 23.76±5.76 Female 493 15.84±2.93 23.45±5.61 Whole population 1210 15.67±2.78 23.63±5.7

2.2. DNA extraction and Genotyping DNA extraction was performed using the Gentra DNA extraction kit following the manufacturer's suggestion. Out of 1,270 buffy coat samples, 1,210 samples gave sufficient DNA for this study. TaqMan, developed by Applied Biosystems (ABI), is an efficient system and genotyping was performed at the core genotyping facility at Johns Hopkins University. The PCRs were conducted with both primers and probes added and only end point products were read. The Hydra and Biomek FX was used to dispense DNA samples and set up PCRs, respectively. The PCR was conducted in two 9700 thermocyclers each equipped with a dual 384-well blocks. Then the end point products were scored using the 7900HT. In each 384-well plate, two reference samples were included for quality control. The primers and probes were designed using Primer Express (ABI). The probes were labeled with two fluorescent dyes, one as an indicator and the other as a quencher. Two probes


Pacific Symposium on Biocomputing 11:467-477(2006)

were synthesized for each locus, each labeled with two different dyes as indicators, respectively. 2.3. SNP tagging We used the haploview program, http://www.broad.mit.edu/mpg/haploview/ [37] to compute the pair-wise LD measure D'. Haploview estimates the maximum-likelihood values of the 4 gamete frequencies, from which D' can be calculated. For these calculations we included members of our melanoma families as well as 350 individuals with BMI less than 25. The SNP haplotype tagging strategy (htSNP) allowed us to identify regions of strong LD using the Gabriel et al block definition [38], and kept every inter-block SNP plus the single SNP within each block with the highest minor allele frequency (MAF). The LD structure of these 7 selected SNPs is presented in Figure 1. The marker rs1801282 represents Pro12Ala in PPAR_2.

Figure 1 LD structure of the selected SNPs in PPAR_3. The numbers within each block indicate the pair ­wise D' as a percentage. The higher LD linkages are shown in the darker color, while the first three SNPs constitute an LD block spanning 9kb of gene region. Moreover, all of these three SNPs are located in the 5'-untranslated region of PPAR_3.


Pacific Symposium on Biocomputing 11:467-477(2006)

2.4. Statistical analysis Hardy-Weinberg proportions were tested among individuals with BMI less than 25 using the Pearson chi-square test. Pair-wise linkage disequilibrium was evaluated by two linkage disequilibrium parameters, Lewontin's D [39] and r2 [40], which were both calculated in haploview (data not shown). 2.4.1. Single SNP analysis Single SNP analyses were conducted using the linear regression procedure implemented in STATA version 9. The SNP effects were analyzed using three models which were, respectively: additive, dominant and recessive models. In the additive model, the wild type genotype, the homozygote consisting of the most common allele in the population, was treated as the baseline and coded as 1, while the heterozygote was coded as 2 and the other homozygote was coded 3. In the dominant model, the wild type homozygote was coded by 1 as baseline and the heterozygote and the other homozygote were both coded 2. Moreover, in the recessive model, both of the wild type homozygote and heterozygote were treated as baseline with the rare homozygote coded 2. 2.4.2. Haplotype analysis All haplotype frequencies were estimated using the expectation maximization ( E M ) algorithm in haplo.stats in the R programming language (http://mayoresearch.mayo.edu/mayo/research/biostat/schaid.cfm). Schaid et al [41] developed a score statistic that can be used to test the statistical association between haplotypes and different types of human traits, including binary and quantitative traits. This method also allows adjustment for non-genetic factors. In this analysis, we used haplo.stats to compute the global score statistic (which allows us to test the significance of association considering all haplotypes) and obtain a global p-value and the haplotype specific statistic (which allows us to compare each haplotype with a selected common haplotype). All haplotypes with a frequency less than 1% were dropped from the score test in order to reduce the degrees of freedom. Another advantage of Schaid's method is to compute empirical p-values using simulation when the haplotype data is sparse. The empirical p-values are computed by repeatedly first permuting the genotypes among the subjects and then computing the score statistics. 3. Result Hardy-Weinberg proportions were tested among individuals with BMI less than 25 using the Pearson chi-square test and no SNPs showed deviation from Hardy Weinberg equilibrium. The minor allele frequencies are 0.479, 0.130, 0.134,


Pacific Symposium on Biocomputing 11:467-477(2006)

0.143, 0.445, 0.151 and 0.441 for SNPs rs17036314, rs12490265, rs4684847, rs1801282, rs2938392, rs2959273, rs1175542, respectively. Linear regression analysis of single SNP revealed two SNPs associated with BMI with significant p-values reported in Table 3 (non-adjusted for multiple testing).
Table 3. Significant associations revealed by single SNP analysis and the body mass index (BMI) SNP ID rs2938392 rs2938392 rs1175542 rs1175542 Location Second intron Second intron Fifth intron Fifth intron Population Whole Male Whole Male Model Dominant Dominant Recessive Recessive Coefficient 1.0866 1.3818 0.898 1.171 P-value 0.008 0.009 0.029 0.029

In addition, we used a sliding window technique to fine map any potential signals within the larger haplotype using haplo.stats [41]. The size of the window varied from 3-4 SNPs in order to provide a comprehensive assessment of the haplotype subsets within the gene. Permutation based p-values were summarized in Table 4 for significant SNP combinations.
Table 4. Haplotypes revealing significant associations with BMI using 1,210 Mexican samples, p-value(sim) indicated the p-values based on permutation. Haplotypes rs17036314*1/rs12490265*2/rs1801282*1 rs17036314*1/rs4684847*1/rs1801282*1 rs17036314*2/rs2938392*1/rs1175542*2 rs12490265*2 /rs1801282*1/ rs2938392*2 rs12490265*2 /rs1801282*1/ rs1175542*1 rs17036314*1/rs12490265*2/rs4684847*1/rs1801282*1 rs17036314*2/rs4684847*1/rs2938392*1/rs1175542*2 rs12490265*2/rs4684847*1/rs1801282*1/ rs2938392*2 rs12490265*2/rs4684847*1/rs1801282*1/ rs1175542*1 Frequency 0.38952 0.38808 0.28722 0.42052 0.42607 0.38607 0.2869 0.42129 0.42648 P-value 0.05224 0.05637 0.05607 0.05121 0.05451 0.03466 0.05644 0.0388 0.04124 P-value (sim) 0.0495 0.058 0.0555 0.0512 0.0558 0.034 0.0564 0.0387 0.043

Furthermore, we examined the association between haplotypes built from 7 SNPs and BMI using a global score statistic. While the permutation based global p-value was 0.544 (degree of freedom=11), one individual haplotype gave a p-value of 0.086 (permutation based) in all samples for a haplotype frequency of 0.279. Interestingly, when the analyses were conducted in males only, the permutation based global p-value was 0.055, and one relatively frequent haplotype (frequency =0.28) gave a significant signal of 0.013.


Pacific Symposium on Biocomputing 11:467-477(2006)

3.

Discussion

To summarize, we genotyped the 7 selected SNPs of the PPAR_ transcript variant 3 (including pro12Ala) in a group of unrelated 717 males and 493 females (ages ranging from 11-24), among which 3 SNPs are located in the 5' untranslated region. These 7 SNPs were selected by the tagging algorithm implemented in haploview to cover the whole gene and all SNPs conformed to Hardy-Weinberg equilibrium. We tested each of the 7 SNPs individually for association with body mass index (BMI). Although no association between Pro12Ala and BMI was observed, two SNPs rs2938392 and rs1175542 revealed significant associations with BMI (p-values=0.008 and 0.029, respectively). It is worth noting that the SNP rs2938392 is roughly 41.5 Kb away from Pro12Ala (rs1801282), a polymorphism known to be associated with BMI and diabetes type II. Furthermore, when all 7 SNPs were analyzed together, the permutation based global p-value was 0.544, and one individual haplotype (haplotype frequency =0.279) gave a significant signal of 0.086 (permutation based). Moreover, when the analyses were conducted in males only, the permutation based global p-value was 0.055, and one relatively frequent haplotype (frequency =0.28) gave a significant signal of 0.013. Several speculations may explain our findings. 1 . There is no association between Pro12Ala polymorphism and BMI in the 1,210 school students from Morelos, Mexico. The positive signals we detected are false positives, which may be due to the high-level admixture structure in the Mexican population. 2. There may not be a direct association between Pro12Ala and BMI, however, SNPs which are in high LD with Pro12Ala are associated with BMI in M e x i c a n population, but this hypothesis will need to be further investigated. 3. Environmental factors may play a role to increase the genetic effect and the lack of data on dietary factors and physical activities in this data set may have limited the power to detect an association. T o summarize, our results indicate that SNPs in high linkage disequilibrium with Pro12Ala are associated with BMI in 1,210 students in Morelos, Mexico. However, due to limited access to the epidemiological factors such as dietary factors and physical activities, the results we reported here need to be interpreted with caution. Future study will explore causal associations between other genetic and non-genetic risk factors and obesity in this population. Further studies will consider increased sample sizes, and we will genotype 40 random SNPs to assess the level of population admixture this population.


Pacific Symposium on Biocomputing 11:467-477(2006)

Acknowledgments Professor Andrew Collins would like to acknowledge funding from the Biotechnology and Biological Sciences Research Council (UK). References 1. 2. 3. L. Fajas, J-C. Fruchart and J. Auwerx, FEBS Lett. 438(1-2), 55 (1998). M. Stumvoll and H.Haring, Diabetes. 51(8), 2341 (2002). B.A.Beamer, C.J.Yen, R.E.Andersen, D.Muller, D.Elahi, L.J.Cheskin, R.Andres, J.Roth, and A.R.Shuldiner, Diabetes 47(11), 1806 (1998). 4 . S.S.Deeb, L.Fajas, M.Nemoto, J.Pihlajamaki, L.Mykkanen, J.Kuusisto, M.Laakso, W.Fujimoto, and J.Auwerx, Nat. Gen. 20(3), 284 (1998). 5 . Y.Mori, H.Kim-Motoyama, T.Katakura, K.Yasuda, H.Kadowaki, B.A.Beamer, A.R.Shuldiner, Y.Akanuma, Y.Yazaki, and T.Kadowaki, Biochem. Biophys. Res. Commun. 251(1), 195 (1998). 6 . A.Hamann, H.Munzberg, P.Buttron, B.Busing, A.Hinney, H.Mayer, W.Siegfried, J.Hebebrand, and H.Greten, Eur J Endocrinol 141(1), 90 (1999). 7 . J.Ek, S.A.Urhammer, T.I.A.Sorensen, T.Andersen, J.Auwerx, and O.Pedersen, Diabetologia 42(7), 892 (1999). 8 . R.Valve, K.Sivenius, R.Miettinen, J.Pihlajamaki, A.Rissanen, S.S.Deeb, J.Auwerx, M.Uusitupa, and M.Laakso, J. Clin. Endocrinol. Metab. 84(10), 3708 (1999). 9 . F.P.Mancini, O.Vaccaro, L.Sabatino, A.Tufano, A.A.Rivellese, G.Riccardi, and V.Colantuoni, Diabetes 48(7), 1466 (1999). 1 0 . J.Ringel, S.Engeli, A.Distler, and A.M.Sharma, Biochem. Biophys. Res. Commun. 254(2), 450 (1999). 1 1 . M.Koch, K.Rett, E.Maerker, A.Volk, K.Haist, M.Deninger, W.Renn, and H.U.Haring, Diabetologia 42(6), 758 (1999). 1 2 .K.Clement, S.Hercberg, B.Passinge, P.Galan, M.Varroud-Vial, A.R.Shuldiner, B.A.Beamer, G.Charpentier, B.Guy-Grand, P.Froguel, and C.Vaisse, Int. J. Obes. 24(3), 391 (2000). 1 3 . K.Hara, T.Okada, K.Tobe, K.Yasuda, Y.Mori, H.Kadowaki, R.Hagura, Y.Akanuma, S.Kimura, C.Ito, and T.Kadowaki, Biochem. Biophys. Res. Commun. 271(1), 212 (2000). 14. W.D.Li, J.H.Lee, and R.A.Price, Mol. Genet. Metab. 70(2), 159 (2000). 1 5 . E.Y.Oh, K.M.Min, J.H.Chung, Y.K.Min, M.S.Lee, K.W.Kim, and M.K.Lee, J. Clin. Endocrinol. Metab. 85(5), 1801 (2000). 1 6 .S.A.Cole, B.D.Mitchell, W.C.Hsueh, P.Pineda, B.A.Beamer, A.R.Shuldiner, A.G.Comuzzie, J.Blangero, and J.E.Hixson, Int. J. Obes. 24(4), 522 (2000). 1 7 . J.G.Eriksson, V.Lindi, M.Uusitupa, T.J.Forsen, M.Laakso, C.Osmond, and D.J.P.Barker, Diabetes 51(7), 2321 (2002). 1 8 . O.Poirier, V.Nicaud, F.Cambien, and L.Tiret, J. Mol. Med. 78(6), 346 (2000).


Pacific Symposium on Biocomputing 11:467-477(2006)

1 9 .J.Pihlajamaki, R.Miettinen, R.Valve, L.Karjalainen, L.Mykkanen, J.Kuusisto, S.Deeb, J.Auwerx, and M.Laakso, Atherosclerosis 151(2), 567 (2000). 2 0 . O.Vaccaro, F.P.Mancini, G.Ruffa, L.Sabatino, V.Colantuoni, and G.Riccardi, Int. J. Obes. 24(9), 1195 (2000). 2 1 . S.J.Hasstedt, Q.F.Ren, K.Teng, and S.C.Elbein, J. Clin. Endocrinol. Metab. 86(2), 536 (2001). 2 2 .W.C.Hsueh, S.A.Cole, A.R.Shuldiner, B.A.Beamer, J.Blangero, J.E.Hixson, J.W.MacCluer, and B.D.Mitchell, Diabetes Care 24(4), 672 (2001). 23. H.Mori, H.Ikegami, Y.Kawaguchi, S.Seino, N.Yokoi, J.Takeda, I.Inoue, Y.Seino, K.Yasuda, T.Hanafusa, K.Yamagata, T.Awata, T.Kadowaki, K.Hara, N.Yamada, T.Gotoda, N.Iwasaki, Y.Iwamoto, T.Sanke, K.Nanjo, Y.Oka, A.Matsutani, E.Maeda, and M.Kasuga, Diabetes 50(4), 891 (2001). 2 4 M.M.Swarbrick, C.M.L.Chapman, B.M.McQuillan, J.Hung, . P.L.Thompson, and J.P.Beilby, Eur J Endocrinol 144(3), 277 (2001). 2 5 J.Luan, P.O.Browne, A.H.Harding, D.J.Halsall, S.O'Rahilly, . V.K.K.Chatterjee, and N.J.Wareham, Diabetes 50(3), 686 (2001). 2 6 .M.Hara, S.Y.Alcoser, A.Qaadir, K.K.Beiswenger, N.J.Cox, and D.A.Ehrmann, J. Clin. Endocrinol. Metab. 87(2), 772 (2002). 2 7 . L.Frederiksen, K.Brodbaek, M.Fenger, T.Jorgensen, K.Borch-Johnsen, S.Madsbad, and S.A.Urhammer, J. Clin. Endocrinol. Metab. 87(8), 3989 (2002). 2 8 . A.Doney, B.Fischer, D.Frew, A.Cumming, D.M.Flavell, M.World, H.E.Montgomery, D.Boyle, A.Morris, and a.Palmer et, BMC Genet 3(1), 21 (2002). 2 9 . O.Vaccaro, F.P.Mancini, G.Ruffa, L.Sabatino, C.Iovine, M.Masulli, V.Colantuoni, and G.Riccardi, Clinical Endocrinology 57(4), 481 (2002). 30. Y.Yamamoto, H.Hirose, K.Miyashita, K.Nishikai, I.Saito, M.Taniyama, M.Tomita, and T.Saruta, Metabolism 51(11), 1407 (2002). 31. J.L.Gonzalez Sanchez, M.Serrano Rios, C.Fernandez Perez, M.Laakso, and M.T.Martinez Larrad, Eur. J. Endocrinol. 147(4), 495 (2002). 3 2 . P.Poulsen, G.Andersen, M.Fenger, T.Hansen, S.M.Echwald, A.Volund, H.Beck-Nielsen, O.Pedersen, and A.Vaag, Diabetes 52(1), 194 (2003). 3 3 . J.Robitaille, J.P.Despres, L.Perusse, and M.C.Vohl, Clin. Genet. 63(2), 109 (2003). 3 4 .P.W.Franks, J.Luan, P.O.Browne, A.-H.Harding, S.O'Rahilly, V.K.K.Chatterjee, and N.J.Wareham, Metabolism 53(1), 11 (2004). 3 5 . E.S.Tai, D.Corella, M.urenberg-Yap, X.Adiconis, S.K.Chew, C.E.Tan, and J.M.Ordovas, J.Lipid Res. 45(4), 674 (2004). 36. E. C. Lazcano-Ponce , B. Hernandez, A. Cruz-Valdez, B. Allen, R. Diaz, C. Hernandez, R. Anaya and M. Hernández-Avila, Arch. Med. Res. 34(3), 222 (2003). 37. J. C. Barrett, B. Fry, J. Maller and M. J. Daly, Bioinformatics, 21(2), 263 (2005). 3 8 . S. B. Gabriel, S. F. Schaffner, H. Nguyen, J. M. Moore, J. Roy, B. Blumenstiel, J. Higgins, M. DeFelice, A. Lochner, M. Faggart, S. N. Liu-


Pacific Symposium on Biocomputing 11:467-477(2006)

Cordero, C. Rotimi, A. Adeyemo, R. Cooper, R. Ward, E. S. Lander, M. J. Daly and D. Altshuler, Science 296(5576) 2225 (2002). 39. R. C. Lewontin and M. W. Feldman, Theor Popul Biol 34(2) 177 (1988). 40. B. Devlin and N. A. Risch, Genomics 29(2), 311 (1995). 4 1 . D. J. Schaid, C. M. Rowland, D. E. Tines, R. M. Jacobson and G. A. Poland Am. J. Hum. Genet. 70(2), 425 (2002).