Genetic linkage mapping in complex pedigrees

Genetic linkage analysis is the primary method for the identification of loci contributing to complex disease susceptibility. Linkage analysis techniques can be applied to both disease status (discrete traits) and to quantitative trait measures (quantitative trait loci or QTL mapping). Such techniques will be most effective if they can be applied to all of the available data; in human, ecological and livestock genetics this often means families with complex pedigree structures. The analysis of complex pedigrees is more difficult, both in terms of model formulation and computational ease, than similar studies of small family structures such as affected sibling pairs (ASP). Univariate variance component (VC) techniques suitable for QTL analysis of both quantitative and qualitative (via a threshold model) traits are described. Extensions to the univariate VC methods are proposed, allowing QTL analyses of longitudinal data in complex pedigrees, with polynomial based covariance functions offering a parsimonious description of the covariance structure across measures. Computer simulations are used to show that, under a range of realistic scenarios, the longitudinal QTL method offers more power to detect QTL than univariate or repeated measures methods. The longitudinal method is subsequently applied to 330 extended families from the Framingham Heart Study, allowing the identification of QTL for a number of cardiovascular disease risk factors. The maximum LOD score (3.12) is obtained on chromosome 16 for Body Mass Index (BMI) and subsequent multivariate analyses showed that this QTL is most relevant to BMI at early ages. Threshold model based VC and parametric linkage analyses are applied to a set of Scottish families affected by psychiatric disease. The results from this analysis are in agreement with previous results implicating chromosome 1q42 in psychiatric disease susceptibility. The broad application of the VC techniques is further demonstrated by applying the techniques to a QTL mapping problem in a very large Red Deer (Cervus Elaphus) pedigree. Linkage analysis is commonly used to identify candidate regions for further study. These candidate regions will be the chromosomal segments shared among related individuals with common diagnoses, with recombination events delineating the regions of interest. However in genetically complex traits, the relationship between phenotype and genotype is not one to one. The effect of changing the parameters defining the relationship between phenotype and genotype is investigated, both analytically and by computer simulation. Increasing the rate at which affected individuals without mutations in the disease region of interest occur in the sample (the phenocopy rate) is found to have a large effect on the validity of the inferred region. This has implications in genetic studies of common disease (e.g. schizophrenia), where the phenocopy rate will often be non-zero. The use of extended families for linkage mapping has become a controversial issue, with the field of psychiatric genetics somewhat polarised; a number of groups have collected mainly extended family data whilst others have focused on small ASP family structures. Whilst the advantages of a given study design will vary depending upon the unknown 'true' disease model, it is argued that extended families will often be more useful for locus identification than sib pair based studies. It is shown that the heterogeneity introduced by collecting large numbers of sib pairs from a number of different populations will impact significantly upon the power to detect the effects of any single gene.