Sci. Aging Knowl. Environ., 28 April 2004
Vol. 2004, Issue 17, p. pe17
[DOI: 10.1126/sageke.2004.17.pe17]


Methods for Genetic Dissection of Complex Traits

Trudy F. C. Mackay

The author is in the Department of Genetics, North Carolina State University, Raleigh, NC 27695, USA. E-mail: trudy_mackay{at}

Key Words: quantitative trait loci • genetics • Drosophila • allelic effects • epistasis • polymorphisms

This Perspective is a response to a previously published SAGE KE Perspective. Most traits of medical, agricultural, and evolutionary importance are genetically complex. These complex, or quantitative, traits are controlled by multiple interacting quantitative trait loci (QTLs) with small individual effects, whose expression is contingent on gender, genetic background, and the environment. One of the greatest challenges facing biologists today is to understand the "genetic architecture" of such traits: What are the QTLs affecting the trait and variation in the trait; what is the distribution of QTL allelic effects; what is the extent of epistasis (interaction) between QTLs; what molecular polymorphisms define QTL alleles; and what molecular mechanisms cause differences in trait phenotypes (1)? The reason why this is challenging is that the combination of subtle and conditional QTL effects means that one cannot infer an individual's genotype for the multiple QTLs that affect the trait from observations of the phenotype. Therefore, QTLs must be mapped by linkage to polymorphic molecular markers whose genotype can be scored unambiguously.

Could one design a single perfect experiment to map all segregating QTLs in a given population to the level of genetic locus and estimate all additive and nonadditive (dominance and epistatic) effects? A possibility is to sample 10,000 or more individuals from a long-established, random mating, outbred population; determine their genotypes for all segregating polymorphisms; and identify polymorphisms associated with differences in the trait phenotype. This linkage disequilibrium (LD) mapping design uses historical recombination to map QTLs with high resolution and, in theory, is very powerful (2) if all of the assumptions underlying its application are met. A sample size of 10,000 at the very least is necessary, because QTL effects are not large and the environment would not be controlled, both of which greatly reduce the power to detect subtle effects. Large sample sizes are also necessary if one is to evaluate pairwise and higher-order epistatic interactions. Currently, such an experiment is not technically feasible because of the cost of genotyping all polymorphisms. Further, population admixture will generate spurious associations, and LD may not have decayed to a point that is useful for mapping QTLs to the level of genes in populations derived from a relatively recent bottleneck in population size; that is, a population that is not at equilibrium, because it was established from a small population not too many generations ago (evolutionarily speaking) and is growing (for example, humans).

Given that we cannot perform an ideal single experiment to solve the genetic architecture of any quantitative trait, does this mean we should abandon all efforts a priori? Or should we proceed with the imperfect tools at our disposal, while remaining fully cognizant of the limitations of each? We have embraced the latter philosophy and argued that much progress toward understanding complex traits can be made by using the powerful genetic and genomic resources in model organisms such as Drosophila (3). Specifically, we have proposed that one can identify candidate QTLs that affect any complex trait in Drosophila by performing the following series of experiments: (i) first localize the QTLs to relatively broad genomic regions by recombination mapping, then (ii) narrow the QTL location by quantitative complementation tests to overlapping deficiencies, and finally, (iii) use complementation tests to mutations of positional candidate genes to nominate, for further study, genes that potentially correspond to the QTL. If linkage disequilibrium mapping reveals that molecular polymorphisms in the candidate QTL are associated with genetic variation in the trait in large samples of alleles derived from a natural population, then we can be fairly confident that the uncertainties encountered at each step in the process have not culminated in false positive results, because any errors in interpretation at any stage will be corrected in the subsequent stage.

In his recent Perspective, Service (4) has espoused the former and more pessimistic viewpoint, pointing out the deficiencies of the tools in our genetic toolkit. He reminds us that QTL mapping typically uses a mapping population derived from two strains, and is therefore a restricted sample of the genetic variation existing for the trait. Further, the size of the mapping population is usually such that QTLs are localized to broad chromosomal regions that contain many genes. All of this is true. Thus, the challenge for quantitative geneticists is to either (i) create and genotype mapping populations large enough to map QTLs to the level of genes, and with multiple parental lines to capture more of the segregating variation; or (ii) to devise strategies that bypass the difficulties of high-resolution recombination mapping. Indeed, these hurdles were the motivation for implementing the quantitative deficiency complementation test (5), because deficiency mapping has been the standard workhorse of the Drosophila community for fine-scale mapping of mutations. Specifically, deficiency mapping is capable of much higher-resolution mapping of QTLs and requires much less effort than recombination mapping. Further, deficiency or mutant complementation tests are not restricted to only two strains and can be used with a larger number of wild-derived alleles of the candidate gene/gene region (5-8), as was done by Geiger-Thornsberry and Mackay (9) to identify candidate genes and gene regions that affect naturally occurring variation in life span.

The disadvantage of deficiency and mutant complementation mapping is that failure to complement--whether qualitative or quantitative--can be attributable to allelism or epistasis. This was elaborated in the original paper utilizing the technique (5), by Geiger-Thornsberry and Mackay (9), and by Service (4). The utility of the method thus depends on the pervasiveness and magnitude of epistatic interactions between QTLs that affect life span and other complex traits. Some controls can be done to exclude obvious epistatic interactions, such as failure to complement arising from interactions with genes on the balancer chromosome, and controlling the genetic backgrounds of strains used in the tests, such that the only natural genetic variation arises from QTLs on the chromosome that contains the candidate gene (9); or, better still, by introgressing the genomic fragment that contains the natural alleles of the candidate gene as well as the tested mutant into a standard background (6-8). Using these more sophisticated genetic manipulations of the naturally occurring candidate gene alleles, in combination with the new set of transposon insertions (10) and small deletions with molecularly defined endpoints (11) that have been generated in an isogenic background, should greatly ameliorate the possibility of epistatic interactions that confound the interpretation of these tests in the future. The use of the parental isogenic strain as a control, rather than a balancer chromosome that itself contains a host of mutations, will be a further improvement.

The concern about failure to complement being attributable to epistatic interactions is valid. We know that epistatic interactions do occur between QTLs that affect variation in many quantitative traits in Drosophila (12-16), including life span (17, 18). Therefore, any candidate gene or gene regions implicated by deficiency or mutant complementation tests remains a "candidate gene" until verified by independent methods. The question is whether epistasis is so pervasive that the method is not useful [the position endorsed by Service (4)] or whether epistasis typically does not confound the interpretation, given appropriate genetic and statistical controls (our viewpoint). This issue can only be addressed by experimentation. To our knowledge, the paradigm of QTL mapping, followed by complementation tests to deficiencies and/or mutations, has had 100% success in identifying bona fide candidate genes that affect naturally occurring variation in quantitative traits, as confirmed by subsequent LD mapping studies (6, 8, 19, 20). Nevertheless, it is important to realize that these methods were introduced as a screening method, to whittle down the size of the genomic region embraced by the QTL that affects the trait of interest. The alternative would be to conduct association studies of all genes in the QTL intervals, which would arguably generate far more false positives than complementation tests.

In summary, we agree with Service's conclusion (4) that neither recombination mapping nor complementation tests are sufficient by themselves to identify a relatively complete set of genes that account for the natural variation of life span (or any other complex trait) in Drosophila. We disagree with the supposition that the genetic architecture of life span is likely to be simple. (In his Perspective, Service states that our finding, that single QTLs affecting variation in life span between the Oregon and 2b strains fractionate into multiple closely linked QTLs, must be wrong and attributable to epistasis, because it's unlikely that many genes with complicated properties could cause variation between just two strains.) Our observation that single QTLs fractionate into multiple linked QTLs (often with opposite effects) upon further scrutiny (5, 20) is as likely to reflect reality as a problem with the complementation mapping approach. This issue will be resolved by further experimentation, not argumentation.

April 28, 2004
  1. T. F. C. Mackay, The genetic architecture of quantitative traits. Annu. Rev. Genet. 35, 303-339 (2001).[CrossRef][Medline]
  2. N. Risch, K. Merikangas, The future of genetic studies of complex human diseases. Science 273, 1516-1517 (1996).[Abstract/Free Full Text]
  3. T. F. C. Mackay, Quantitative trait loci in Drosophila. Nat. Rev. Genet. 2, 11-20 (2001).[Medline]
  4. P. M. Service, How good are quantitative complementation tests? Sci. Aging Knowl. Environ. 2004 (12), pe13 (2004).
  5. E. G. Pasyukova, C. Vieira, T. F. C. Mackay, Deficiency mapping of quantitative trait loci affecting longevity in Drosophila melanogaster. Genetics 156, 1129-1146 (2000).[Abstract/Free Full Text]
  6. A. D. Long, R. F. Lyman, C. H. Langley, T. F. C. Mackay, Two sites in the Delta gene region contribute to naturally occurring variation in bristle number in Drosophila melanogaster. Genetics 149, 999-1017 (1998). [Abstract/Free Full Text]
  7. R. F. Lyman, C. Lai, T. F. C. Mackay, Linkage disequilibrium mapping of molecular polymorphisms at the scabrous locus associated with naturally occurring variation in bristle number in Drosophila melanogaster. Genet. Res. 74, 303-311 (1999).[CrossRef][Medline]
  8. C. Robin, R. F. Lyman, A. D. Long, C. H. Langley, T. F. C. Mackay, hairy: A quantitative trait locus for Drosophila bristle number. Genetics 162, 155-164 (2002).[Abstract/Free Full Text]
  9. G. L. Geiger-Thornsberry, T. F. C. Mackay, Quantitative trait loci affecting natural variation in Drosophila longevity. Mech. Ageing Dev. 125, 179-189 (2004).[CrossRef][Medline]
  10. S. T. Thibault, M. A. Singer, W. Y. Miyazaki, B. Milash, N. A. Dompe, C. M. Singh, R. Buchholz, M. Demsky, R. Fawcett, H. L. Francis-Lang et al., A complementary transposon tool kit for Drosophila melanogaster using P and piggy Bac. Nat. Genet. 36, 283-287 (2004).[CrossRef][Medline]
  11. A. L. Parks, K. R. Cook, M. Belvin, N. A. Dompe, R. Fawcett, K. Huppert, L. R. Tan, C. G. Winter, K. P. Bogart, J. E. Deal et al., Systematic generation of high-resolution deletion coverage of the Drosophila melanogaster genome. Nat. Genet. 36, 288-292 (2004).[CrossRef][Medline]
  12. A. D. Long, S. L. Mullaney, L. A. Reid, J. D. Fry, C. H. Langley, T. F. C. Mackay, High resolution mapping of genetic factors affecting abdominal bristle number in Drosophila melanogaster. Genetics 139, 1273-1291 (1995).[Abstract/Free Full Text]
  13. M. C. Gurganus, S. V. Nuzhdin, J. W. Leips, T. F. C. Mackay, High resolution mapping of quantitative trait loci for sternopleural bristle number in Drosophila melanogaster. Genetics 152, 1585-1604 (1999).[Abstract/Free Full Text]
  14. C. L. Dilda, T. F. C. Mackay, The genetic architecture of Drosophila sensory bristle number. Genetics 162, 1655-1674 (2002).[Abstract/Free Full Text]
  15. K. Weber, R. Eisman, L. Morey, A. Patty, J. Sparks, M. Tausek, Z.-B. Zeng, An analysis of polygenes affecting wing shape on chromosome 3 in Drosophila melanogaster. Genetics 153, 773-786 (1999). [Abstract/Free Full Text]
  16. K. Weber, R. Eisman, S. Higgins, L. Morey, A. Patty, M. Tausek, Z.-B. Zeng, An analysis of polygenes affecting wing shape on chromosome 2 in Drosophila melanogaster. Genetics 159, 1045-1057 (2001).[Abstract/Free Full Text]
  17. J. Leips, T. F. C. Mackay, Quantitative trait loci for life span in Drosophila melanogaster: interactions with genetic background and larval density. Genetics 155, 1773-1788 (2000).[Abstract/Free Full Text]
  18. J. Leips, T. F. C. Mackay, The complex genetic architecture of Drosophila life span. Exp. Aging Res. 28, 361-390 (2002).[CrossRef][Medline]
  19. A. D. Long, R. F. Lyman, A. H. Morgan, C. H. Langley, T. F. C. Mackay, Both naturally occurring insertions of transposable elements and intermediate frequency polymorphisms at the achaete-scute complex are associated with variation in bristle number in Drosophila melanogaster. Genetics 154, 1255-1269 (2000).[Abstract/Free Full Text]
  20. M. De Luca, N. V. Roshina, G. L. Geiger-Thornsberry, R. F. Lyman, E. G. Pasyukova, T. F. C. Mackay, Dopa decarboxylase (Ddc) affects variation in Drosophila longevity. Nat. Genet. 34, 429-433 (2003).[CrossRef][Medline]
Citation: T. F. C. Mackay, Methods for Genetic Dissection of Complex Traits. Sci. Aging Knowl. Environ. 2004 (17), pe17 (2004).

Science of Aging Knowledge Environment. ISSN 1539-6150