Mapping and verification of susceptibility loci for smoking quantity using permutation linkage analysis

  • A Corrigendum to this article was published on 17 May 2005


Nicotine dependence is the most prevalent form of drug addiction in the US and throughout the world. Epidemiological studies demonstrate that genetics accounts for at least 50% of the liability to nicotine dependence. However, there have been very limited linkage studies providing convincing evidence of susceptibility genomic loci for this disorder. In this study, we conducted genome-wide permutation linkage analyses on the smoking data collected between 1970 and 1972 of the Framingham Heart Study (FHS) to account for the abnormality associated with the smoking quantity (defined as the number of cigarettes smoked per day). We used empirical thresholds obtained from permutation tests to determine the significance of each genomic region. The variance component method implemented in SOLAR was used for the analysis. Under the empirical genome-wide thresholds determined specifically for the FHS smoking data, we found two highly or near-highly significant linkages of nicotine dependence on chromosomes 1 and 4 (P=0.001) and eight significant linkages on chromosomes 3, 7, 8, 9, 11, 16, 17, and 20 (P<0.05). These findings strongly indicate that some of these regions may harbor susceptibility loci for nicotine dependence. Further analysis of these positive regions by fine mapping and/or association analysis is warranted. To our knowledge, this study presents the most convincing linkage evidence for nicotine dependence in the field.


Tobacco is one of the most widely abused substances, killing approximately 435 000 US citizens in 2000.1 Despite increasing public awareness of the health risks inherent in the use of tobacco products and legislation that reduces the availability of cigarettes and prohibits smoking in many public facilities, virtually no further reduction in smoking prevalence has occurred in the US since 1990.2 This phenomenon is largely attributable to nicotine dependence, a genetically complex disorder.3, 4, 5, 6 Many twin studies have revealed that there is a strong genetic component in nicotine dependence.7, 8, 9, 10, 11, 12, 13, 14, 15, 16 Meta-analysis of reported twin studies indicated that genetics accounts for at least 50% of the liability to this disorder.17, 18 Thus, identifying the genes predisposing to nicotine dependence and understanding of its molecular mechanism will be of great interest to effectively prevent or treat it.

Although other alternatives (eg, association study) exist, genome-wide linkage analysis remains an effective approach to searching for initial evidence of genomic regions affecting a complex disease. Very limited efforts, however, have been made for linkage study on smoking behavior.3, 6 So far there were only a few independent studies intended for finding genetic linkage of nicotine dependence. Straub et al19 detected a modest linkage (LOD=2.63) of nicotine dependence near marker D2S1325 on chromosome 2 in the Christchurch sample of New Zealand, but failed to replicate this linkage in their Richmond sample of Virginia. Using data from the Collaborative Study on the Genetics of Alcoholism (COGA), Duggirala et al20 identified a strong linkage (LOD=3.2) of smoking behavior in chromosomal region 5q. Additionally, the authors also reported several other chromosomal regions that showed minor linkage evidence. Similarly, Bergen et al21 analyzed the COGA smoking data using different methods and reported several minor linkages. However, when these positive regions are compared, one will find that the detected linkage regions for nicotine dependence lacked reproducibility.3

Recently, several research groups have performed linkage analyses with the smoking data of the FHS that were made available through the Genetic Analysis Workshop (GAW) 13.22, 23, 24, 25 In aggregate, these studies reported approximately 29 potential linkage regions for the smoking behavior. These potential linkage regions were located on most of the 22 autosomal chromosomes except chromosomes 3, 10, 18, and 19 (no genotypic data were available on the sex chromosomes). However, except for the results from Li et al,25 almost none of these reported linkage regions had reached a significant linkage based on the theoretical thresholds proposed by Lander and Kruglyak.28 Such discrepancy was largely due to the use of different statistical methods/programs and ways of defining the smoking phenotype. It should be pointed out that the P-values or LOD scores derived from all these analyses were based on assumptions of statistical distributions (eg, chi-square distribution for LOD scores) for the related test statistics. Due to the apparent violation of the normality assumption for the FHS smoking data (including transformed data), however, these test statistics presumably would not follow the theoretical distributions well. As a result, the theoretical thresholds (eg, LOD=2.2 for suggestive linkage and LOD=3.6 for significant linkage) derived from such assumptions could become unreliable, which in turn would lead to either excessive false positives or false negatives as expected from the preset significance level. But none of the reported analyses on the FHS smoking data evaluated the influences of the departure from normality, and they all used a subjective threshold to determine significance of linkage.

It was this concern that motivated us to evaluate the FHS smoking data because it represents one of the largest data sets containing smoking-related information. Fully exploiting the genetic information contained in this data set will be of great interest and cost efficient to our genetic study of nicotine dependence. In this study, we conducted a genome-wide linkage analysis using a variance component (VC) method with the FHS smoking data collected between 1970 and 1972. Instead of using a fixed theoretical threshold, we tested linkage significances using permutation-based empirical thresholds that appropriately accounted for genome-wide false-positive rates. Though computationally intensive, the permutation test has been gradually accepted as a standard approach for generating objective thresholds for linkage analysis and other genetic data analyses.26, 27, 28, 29, 30, 31, 32, 33 The most attractive feature of the permutation test is that it is straightforward and does not require strong assumption of statistical distributions for test statistics. Our goals were to define appropriate empirical thresholds for linkage analysis specifically for smoking phenotype in the FHS data and then use them as criteria to identify significant linkage regions.


Description of Clinical and Genetic Data

There are two cohorts (original and offspring) in the FHS data. The original cohort began in 1948 with the enrollment of 5209 men and women, 28–62 years of age at study entry, with subjects undergoing repeat exams every 2 years.34 The offspring cohort, recruited into the study in 1971, includes 5124 of the original participants' adult children and their spouses. Offspring subjects underwent examinations approximately every 4 years.35 In the mid-1990s, the largest 330 pedigrees were selected from the two cohorts for genome-wide scan. The pedigrees consist of 4692 subjects, of whom 2885 have participated in the FHS.

Based on the number of smokers present at each exam, the consistency of the clinical data and interviewing time between the two cohorts, and potential environmental effect on smoking phenotype included in the FHS data, Exams 12 and 13 from 1970 and 1972 for the original cohort and Exam 1 from 1971 for the offspring cohort were selected and used in this study. The smoking data for the original cohort were averaged for the two exams, while the offspring smoking data were used as is. Pedigrees with very few or no smokers during this period of time were excluded, which resulted in a total of 305 pedigrees and 4445 individuals.

Nuclear Family vs Extended Family used for Linkage Analysis

Using extended pedigrees and multipoint linkage analysis methods has been a paradigm to increase statistical power in linkage analysis because they are generally more informative than nuclear families and single-point analysis.36, 37 However, even with statistical methods and computing programs greatly improved over the past decades, it is still difficult to efficiently analyze large pedigrees using multipoint methods. The sizes of the original FHS-extended pedigrees range from 7 to 87, with an average of 14 members per pedigree, and many of the pedigrees are beyond the capability of most existing linkage packages. To handle large pedigrees, we separated the original extended pedigrees into nuclear families. In forming the nuclear family data, we included only families with two or more children, each having both CPD (number of cigarettes smoked per day) and DNA marker data. In this way, the resulting data included a total of 430 nuclear families, and 2119 individuals, among which 898 were smokers (CPD>0).

Smoking Phenotype and DNA Markers

Many measurements have been used to assess nicotine dependence, which include CPD, or smoking quantity,12, 38 current tobacco use,16 and nicotine dependence.13, 39 Although these measurements may not be completely identical, they are all highly correlated with nicotine dependence.17, 18 Based on the availability of smoking information in the FHS data set, smoking quantity is used to measure nicotine dependence in this study. As in our previous analyses,25 we used the original CPD, natural logarithm transformation of CPD (log-CPD), and categorized CPD (cat-CPD) in the permutation test. The cat-CPD consists of six categories, indexed by 0, 1, …, 5, corresponding to CPD ranges of 0, 1–5, 6–15, 16–25, 26–35, and ≥36, respectively.

The FHS genotypic data include 401 DNA markers covering the 22 autosomal chromosomes (an average marker interval of 9 cM). These markers were typed for a total of 1702 individuals (394 original cohort and 1308 offspring cohort). Genotype data cleaning, including verification of family relationships and Mendelian inconsistencies, has been described elsewhere.40

Permutation Procedure and Test Statistics

The VC method implemented in SOLAR41 was used for analyzing the original nuclear family data and random permuted samples. Additionally, we also used the VC method implemented in the GeneHunter42 package in our study, which yielded virtually the same results as the VC method from SOLAR. Thus, only the linkage results from SOLAR are reported in this study. LOD scores were calculated across the whole genome, with adjacent test positions spaced at 2 cM. Multipoint methods were used for estimating IBD sharing distributions needed for calculating LOD scores. Other conditions for running the program followed the relevant program defaults.

Permuted samples were generated by randomly shuffling the CPD observations among all children with a valid CPD observation (non-missing) across all nuclear families, while keeping their marker data unchanged. Following a rule of thumb proposed by Churchill and Doerge,31 we generated 1000 such permuted samples, and analyzed each of them using SOLAR under the same settings as for the original data. A genome-wide maximum LOD score was identified and recorded for each permutation; 1000 such maximum LOD scores formed an empirical null distribution of maximum LOD scores, which was used for our linkage test.

Peaks of LOD scores calculated with the original nuclear family data along each chromosome were identified and compared with the corresponding empirical distribution, and an objective estimate of false-positive probability (P-value) for each potential linkage was obtained. We declared significant linkage if the genome-wide P-value is less than or equal to 0.05 or highly significant linkage if the P-value is less than or equal to 0.001, as recommended by Lander and Kruglyak28 for controlling genome-wide false-positive rates. The three forms of the smoking phenotype (ie, CPD, log-CPD, and cat-CPD) each used its own empirical null distribution for the permutation test.


Sample Characteristics

The smoking phenotypic data used in the permutation linkage analysis of this study were selected from Exams 12 and 13 for the original cohort and Exam 1 for the offspring cohort. As shown in Table 1, 2119 subjects representing 430 nuclear families were included in the study. Of them, 898 subjects were smokers with an average CPD of 22.5 and 17.1 cigarettes for men and women smokers, respectively. The average age of smokers was 38.5±13.3 for males and 38.7±13.7 for females, which is very similar to the overall male and female groups.

Table 1 Characteristics of samples used to map susceptibility loci for smoking rate (1970–1972)

Determination of Thresholds Specific for the FHS Smoking Data

As shown in Figure 1, the empirical null distributions of maximum LOD scores for the three forms of smoking quantity (CPD, log-CPD, and cat-CPD) were very similar to each other, suggesting that data transformations had only minor effect on final linkage results. These distributions were exceedingly right-skewed, and had small average LOD values (around 0.12). Based on the empirical distributions specific for FHS smoking data, the LOD threshold for a genome-wide false-positive rate of 0.05 (corresponding to 0.05 false positive expected per genome-wide scan) was 0.68 for CPD, 0.48 for log-CPD, and 0.52 for cat-CPD, respectively, which appears to be much smaller than the corresponding theoretical LOD threshold of 3.6. Under the theoretical threshold, the actual genome-wide false-positive rates were close to zero for all the three phenotypic transformations. Thus, using the theoretical threshold would mask most or all linkage signals that otherwise could be detected. This illustrated the necessity of permutations for linkage analysis with the FHS smoking data.

Figure 1

Empirical null distributions of genome-wide maximum LOD scores for the three forms of the smoking phenotype based on 1 000 permutations.

Linkage Regions for Smoking Quantity

Based on the empirical null distributions, an empirical P-value for each test position was found with the original data, which provides an overall picture of linkage signals on 22 autosomal chromosomes across the whole genome (see Figure 2). Analyses with the three forms of phenotypic transformations yielded very similar results, only with a minor trend of log-CPD>cat-CPD>CPD in terms of linkage signal. Under a genome-wide significance level of 0.05, 10 regions showed significant linkage (Table 2). These regions include three (ND1, ND3, and ND8) on chromosomes 1, 4, and 16 detected with all the phenotypic transformations, three (ND2, ND7, and ND9) on chromosomes 3, 11, and 17 detected with two transformations (cat-CPD and log-CPD), and four regions (ND4, ND5, ND6, and ND10) on chromosomes 7, 8, 9, and 20 detected with either log-CPD or cat-CPD. These detected regions varied in linkage strengths. Of them, two regions (ND1 and ND3) exhibited highly or near-highly significant linkage (P≤0.001) for smoking quantity; the peak positions are flanked by markers GGAA22G10 and ATA4E02 for ND1, and 165 × c11 and GATA5B02 for ND3. Other regions, though all significant, have different extents of significance from P-values of 0.0419 to 0.0171. Compared to the regions identified previously in the FHS cohort, we found that ND2 on chromosome 3 (between markers 059 × a9 and GATA6G12) and ND6 on chromosome 9 (between markers GATA87E02 and GATA12C06) represent newly detected loci for ND in our permutation test with the same dataset. On the other hand, while other regions were reported previously, they never reached the significance levels we obtained in this study for most of the positive regions.

Figure 2

Genome-wide linkage analysis for cigarette smoking with the FHS smoking data. The three curves (for CPD, log-CPD, and cat-CPD) along each chromosome were drawn for −log10 (empirical P-value) against test positions (cM), where P-values were obtained using the empirical null distributions for genome-wide maximum LODs. The centromere of each chromosome is indicated as an open circle. The vertical dashed lines indicate the threshold for genome-wide alpha=0.05.

Table 2 Linkage results for nicotine dependence using empirical thresholds


In this study, we conducted a genome-wide linkage analysis using the VC approach implemented in SOLAR for the FHS smoking data collected between 1970 and 1972. Unlike previous analyses with the same data set, we carried out linkage tests using empirical thresholds derived from permutations and obtained an appropriate estimate of genome-wide false-positive probability for each potential linkage. With the empirical thresholds, a total of 10 genomic regions were detected to have highly significant or significant linkages with nicotine dependence, which was assessed by number of cigarettes smoked per day. Two of the regions (ND1 and ND3) on chromosomes 1 and 4 showed highly or near-highly significant linkages. Their peak positions are flanked by markers GGAA22G10 and ATA4E02 for ND1, and 165xc11 and GATA5B02 for ND3. Additionally, we identified eight other regions that showed significant linkage of nicotine dependence to chromosomes 3, 7, 8, 9, 11, 16, 17, and 20.

Of these 10 significant or highly significant linkages for smoking quantity, eight of them located on chromosomes 1, 4, 7, 8, 11, 16, 17, and 20 were also reported by one or more research groups22, 23, 24, 25 previously with the FHS smoking data at a nominal significance level, with a P-value of around 0.05 or LOD score between 1.0 and 2.0, except for the results reported by our group25 in which we found significant evidence for linkage on chromosome 11 (pointwise P-value=10−6) and suggestive evidence on chromosomes 4, 7, and 17 with a pointwise P-value of less than 0.0017. Nevertheless, our study does not represent a simple replication of those early works. This is because one would find that most regions reported in the early studies were probably no more than a random peak instead of a significant linkage according to the accepted thresholds for suggestive or significant linkage,28 which of course would easily be ignored by genetic researchers in the nicotine field. However, in our analysis with control of genome-wide false-positive rate, these regions stood out with significant linkage signals, and thus their importance has been emphasized. Moreover, regions (ND2 and ND6) on chromosomes 3 and 9 represent two newly detected linkages for smoking behavior in this study.

A low reproducibility of identified positive regions across multiple studies has been a major concern in linkage studies of almost all complex disorders. To increase our confidence on the regions identified in this study, we compared our results with those reported on other independent populations. We found that the four regions on chromosomes 1 (ND1), 3 (ND2), 8 (ND5), and 17 (ND9) were also reported previously on other independent populations but at a nominal significance level.19, 20, 21 Specifically, Bergen et al21 detected markers D1S534 (P=0.0098) and D8S1136 (P=0.0048) as being of nominal significance by using the affected sib-pair and HE methods on the COGA data. These two markers were located within the genomic regions of ND1 and ND5, respectively, detected in our analyses (Table 2). In addition, Bergen et al21 reported another marker, D3S1744 (P=0.0006) on chromosome 3, which was outside but close to the ND2 region we detected in this study. By ‘close to’, we mean a distance of approximately 10 cM from either boundary of the region in our analysis. A region around marker GATA193 (LOD=2.88), detected by Duggirala et al20 using a multipoint VC method on the COGA data, was outside, but close to, the region on chromosome 17 (ND9) in our study. Although these reported significant regions were only at a nominal significance level in the previous studies, these repeatedly detected regions strongly demonstrate that they are more likely to harbor susceptibility loci for nicotine dependence and deserve more attention in future studies.

Our approach of using permutation tests detected more significant linkage regions than the conventional one-round linkage analyses performed previously with the FHS data.22, 23, 24, 25 Such a difference in statistical power was attributable to the use of different thresholds adopted among these studies. Conventional one-round linkage analysis relied on theoretical approximate thresholds derived for certain given conditions,28 including statistical distribution (eg, normality) and the characteristics of data under investigation (eg, genome size, marker density, sample size, etc), while those conditions are rarely satisfied completely in practice. The FHS smoking quantity is not a typical, continuous, quantitative trait and greatly deviates from normality. Multiple genome-wide linkage tests, which are not all independent of each other, further complicate the determination of appropriate thresholds. The commonly used approximate genome-wide thresholds (eg, LOD=2.2 for suggestive linkage and 3.6 for significant linkage28) were not appropriate for linkage analysis with this type of data. Therefore, using the theoretical threshold would decrease statistical power for detection of significant linkage. In contrast, empirical thresholds from permutations are the best suited for linkage testing of the data since they reflect the particular characteristics of the data under investigation.

In summary, we identified 10 genomic regions that may harbor susceptibility loci for nicotine dependence. Of these loci, two showed highly or near-highly significant linkages and the remaining eight showed significant linkage. To our knowledge, this represents the most convincing evidence for linkage studies of nicotine dependence in this field. Interestingly, eight of the 10 positive regions identified in this study were also reported previously by several research groups, including ours with the same data set but at a nominal P-value level for most of them. Additionally, four regions (including a new ND2 locus on chromosome 3) identified in the FHS cohort had been reported for linkage to smoking behavior in the COGA cohort. Therefore, they represent good candidate regions to harbor allelic variants influencing vulnerability to nicotine dependence. However, we should caution that the linkage regions detected in our analysis are still subject to sampling errors. Further investigations are needed to make confirmation before more costly experiments for isolating the responsible genes are initiated.


  1. 1

    Mokdad AH, Marks JS, Stroup DF, Gerberding JL . Actual causes of death in the United States, 2000. JAMA 2004; 291 (10): 1238–1245.

    Article  Google Scholar 

  2. 2

    Husten CG, Shelton DM, Chrismon JH, Lin YC, Mowery P, Powell FA . Cigarette smoking and smoking cessation among older adults: United States, 1965–94. Tob Control 1997; 6: 175–180.

    CAS  Article  Google Scholar 

  3. 3

    Li MD, Ma JZ, Beuten J . Progress in searching for susceptibility loci and genes for smoking-related behavior. Clin Genet 2004; 66: 382–392.

    CAS  Article  Google Scholar 

  4. 4

    Tyndale RF . Genetics of alcohol and tobacco use in humans. Ann Med 2003; 35: 94–121.

    CAS  Article  Google Scholar 

  5. 5

    Lerman C, Berrettini W . Elucidating the role of genetic factors in smoking behavior and nicotine dependence. Am J Med Genet 2003; 118B: 48–54.

    Article  Google Scholar 

  6. 6

    Uhl GR, Liu QR, Naiman D . Substance abuse vulnerability loci: converging genome scanning data. Trends Genet 2002; 18: 420–425.

    CAS  Article  Google Scholar 

  7. 7

    Fisher RA . Lung cancer and smoking. Nature 1958; 182: 108.

    CAS  Article  Google Scholar 

  8. 8

    Fisher RA . Cancer and smoking. Nature 1958; 182: 596.

    CAS  Article  Google Scholar 

  9. 9

    Carmelli D, Swan GE, Robinette D, Fabsitz R . Genetic influence on smoking—a study of male twins. N Engl J Med 1992; 327: 829–833.

    CAS  Article  Google Scholar 

  10. 10

    Kaprio J, Hammar N, Koskenvuo M, Floderus-Myrhed B, Langinvainio H, Sarna S . Cigarette smoking and alcohol use in Finland and Sweden: a cross-national twin study. Int J Epidemiol 1982; 11: 378–386.

    CAS  Article  Google Scholar 

  11. 11

    Swan GE, Carmelli D, Reed T, Harshfield GA, Fabsitz RR, Eslinger PJ . Heritability of cognitive performance in aging twins. The National Heart, Lung, and Blood Institute Twin Study. Arch Neurol 1990; 47: 259–262.

    CAS  Article  Google Scholar 

  12. 12

    Swan GE, Carmelli D, Rosenman RH, Fabsitz RR, Christian JC . Smoking and alcohol consumption in adult male twins: genetic heritability and shared environmental influences. J Subst Abuse 1990; 2: 39–50.

    CAS  Article  Google Scholar 

  13. 13

    Kendler KS, Neale MC, Sullivan P, Corey LA, Gardner CO, Prescott CA . A population-based twin study in women of smoking initiation and nicotine dependence. Psychol Med 1999; 29: 299–308.

    CAS  Article  Google Scholar 

  14. 14

    Hannah MC, Hopper JL, Mathews JD . Twin concordance for a binary trait. II. Nested analysis of ever-smoking and ex-smoking traits and unnested analysis of a ‘committed-smoking’ trait. Am J Hum Genet 1985; 37: 153–165.

    CAS  PubMed  PubMed Central  Google Scholar 

  15. 15

    True WR, Heath AC, Scherrer JF, Waterman B, Goldberg J, Lin N et al. Genetic and environmental contributions to smoking. Addiction 1997; 92: 1277–1287.

    CAS  Article  Google Scholar 

  16. 16

    Maes HH, Woodard CE, Murrelle L, Meyer JM, Silberg JL, Hewitt JK et al. Tobacco, alcohol and drug use in eight- to sixteen-year-old twins: the Virginia Twin Study of Adolescent Behavioral Development. J Stud Alcohol 1999; 60: 293–305.

    CAS  Article  Google Scholar 

  17. 17

    Li MD, Cheng R, Ma JZ, Swan GE . A meta-analysis of estimated genetic and environmental effects on smoking behavior in male and female adult twins. Addiction 2003; 98: 23–31.

    Article  Google Scholar 

  18. 18

    Sullivan PF, Kendler KS . The genetic epidemiology of smoking. Nicotine Tob Res 1999; 1 (Suppl 2): S51–S57, discussion S69–S70.

    Article  Google Scholar 

  19. 19

    Straub RE, Sullivan PF, Ma Y, Myakishev MV, Harris-Kerr C, Wormley B et al. Susceptibility genes for nicotine dependence: a genome scan and followup in an independent sample suggest that regions on chromosomes 2, 4, 10, 16, 17 and 18 merit further study. Mol Psychiatry 1999; 4: 129–144.

    CAS  Article  Google Scholar 

  20. 20

    Duggirala R, Almasy L, Blangero J . Smoking behavior is under the influence of a major quantitative trait locus on human chromosome 5q. Genet Epidemiol 1999; 17 (Suppl 1): S139–S144.

    Article  Google Scholar 

  21. 21

    Bergen AW, Korczak JF, Weissbecker KA, Goldstein AM . A genome-wide search for loci contributing to smoking and alcoholism. Genet Epidemiol 1999; 17 (Suppl 1): S55–S60.

    Article  Google Scholar 

  22. 22

    Bergen A, Yang X, Bai Y, Beerman M, Goldstein A, Goldin L . Genomic regions linked to alcohol consumption in the Framingham Heart Study. BMC Genet 2003; 4 (Suppl 1): S101.

    Article  Google Scholar 

  23. 23

    Goode E, Badzioch M, Kim H, Gagnon F, Rozek LS, Edwards KL et al. Multiple genome-wide analyses of smoking behavior in the Framingham Heart Study. BMC Genetics 2003; 4 (Suppl 1): S102.

    Article  Google Scholar 

  24. 24

    Saccone N, Neuman R, Saccone S, Rice J . Genetic analysis of maximum cigarette-use phenotypes. BMC Genet 2003; 4 (Suppl 1): S105.

    Article  Google Scholar 

  25. 25

    Li MD, Ma JZ, Cheng R, Dupont R, Williams KC, Payne TJ et al. A genome wide scan to identify loci for smoking quantity in the Framingham Heart Study populations. BMC Genet 2003; 4 (Suppl I): S103.

    Article  Google Scholar 

  26. 26

    Nettleton D, Doerge RW . Accounting for variability in the use of permutation testing to detect quantitative trait loci. Biometrics 2000; 56: 52–58.

    CAS  Article  Google Scholar 

  27. 27

    Wan Y, Cohen J, Guerra R . A permutation test for the robust sib-pair linkage method. Ann Hum Genet 1997; 61 (Part 1): 79–87.

    CAS  PubMed  Google Scholar 

  28. 28

    Lander E, Kruglyak L . Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat Genet 1995; 11: 241–247.

    CAS  Article  Google Scholar 

  29. 29

    Lander ES, Schork NJ . Genetic dissection of complex traits. Science 1994; 265: 2037–2048.

    CAS  Article  Google Scholar 

  30. 30

    Doerge RW, Churchill GA . Permutation tests for multiple loci affecting a quantitative character. Genetics 1996; 142: 285–294.

    CAS  PubMed  PubMed Central  Google Scholar 

  31. 31

    Churchill GA, Doerge RW . Empirical threshold values for quantitative trait mapping. Genetics 1994; 138: 963–971.

    CAS  PubMed  PubMed Central  Google Scholar 

  32. 32

    Abecasis GR, Cardon LR, Cookson WO . A general test of association for quantitative traits in nuclear families. Am J Hum Genet 2000; 66: 279–292.

    CAS  Article  Google Scholar 

  33. 33

    Allison DB, Neale MC, Zannolli R, Schork NJ, Amos CI, Blangero J . Testing the robustness of the likelihood-ratio test in a variance-component quantitative-trait loci-mapping procedure. Am J Hum Genet 1999; 65: 531–544.

    CAS  Article  Google Scholar 

  34. 34

    Dawber TR, Kannel WB, Lyell LP . An approach to longitudinal studies in a community: the Framingham Study. Ann NY Acad Sci 1963; 107: 539–556.

    CAS  Article  Google Scholar 

  35. 35

    Feinleib M, Kannel WB, Garrison RJ, McNamara PM, Castelli WP . The Framingham Offspring Study. Design and preliminary data. Prev Med 1975; 4: 518–525.

    CAS  Article  Google Scholar 

  36. 36

    Elston RC, Stewart J . A general model for the genetic analysis of pedigree data. Hum Hered 1971; 21: 523–542.

    CAS  Article  Google Scholar 

  37. 37

    Kruglyak L, Lander ES . Complete multipoint sib-pair analysis of qualitative and quantitative traits. Am J Hum Genet 1995; 57: 439–454.

    CAS  PubMed  PubMed Central  Google Scholar 

  38. 38

    Swan GE, Carmelli D, Cardon LR . The consumption of tobacco, alcohol, and coffee in Caucasian male twins: a multivariate genetic analysis. J Subst Abuse 1996; 8: 19–31.

    CAS  Article  Google Scholar 

  39. 39

    True WR, Xian H, Scherrer JF, Madden PA, Bucholz KK, Heath AC et al. Common genetic vulnerability for nicotine and alcohol dependence in men. Arch Gen Psychiatry 1999; 56: 655–661.

    CAS  Article  Google Scholar 

  40. 40

    Levy D, DeStefano AL, Larson MG, O'Donnell CJ, Lifton RP, Gavras H et al. Evidence for a gene influencing blood pressure on chromosome 17. Genome scan linkage results for longitudinal blood pressure phenotypes in subjects from the framingham heart study. Hypertension 2000; 36: 477–483.

    CAS  Article  Google Scholar 

  41. 41

    Almasy L, Blangero J . Multipoint quantitative-trait linkage analysis in general pedigrees. Am J Hum Genet 1998; 62: 1198–1211.

    CAS  Article  Google Scholar 

  42. 42

    Kruglyak L, Daly MJ, Reeve-Daly MP, Lander ES . Parametric and nonparametric linkage analysis: a unified multipoint approach. Am J Hum Genet 1996; 58: 1347–1363.

    CAS  PubMed  PubMed Central  Google Scholar 

Download references


We thank Dr David Bronson for helpful comments and suggestions on the manuscript. We also thank Mr Dong Zhang, MS, for assisting in setting up the Linux cluster and keeping it running smoothly. This study was supported by NIH grant DA-12844 to MDL. The Framingham Heart Study is supported by the National Heart, Lung, and Blood Institute (NHLBI) in collaboration with Boston University. This manuscript was not prepared in collaboration with investigators of the FHS and does not necessarily reflect the opinions or reviews of the FHS, Boston University, or NHLBI.

Author information



Corresponding author

Correspondence to M D Li.

Additional information


None declared.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Wang, D., Ma, J. & Li, M. Mapping and verification of susceptibility loci for smoking quantity using permutation linkage analysis. Pharmacogenomics J 5, 166–172 (2005).

Download citation


  • tobacco smoking
  • linkage analysis
  • FHS
  • susceptibility loci
  • QTL
  • nicotine dependence

Further reading