Major strengths and weaknesses of model-free methods

This chapter discusses some of the principal advantages and disadvantages inherent in the use of model-free (MF) methods. The principal advantage is that one does not need to specify, a priori, a genetic model for the trait of interest, which often is not known for many complex phenotypes of interest. On the other hand, as with all nonparametric approaches, use of model-free methods results in reduced power for detection of linkage compared with model-based methods when the model is correctly specified. The MF methods also have a potential for computational simplicity and are ideally suited for analysis of specific relative sets such as affected sibpairs. The MF methods are ideally suited to the analysis of quantitative traits for which finding and implementing a suitable genetic model for use in a parametric linkage analysis may be cumbersome. On the other hand, for discrete traits, most model-free methods allow for only a simple definition of "affected," making it difficult to consider such factors as age at onset, diagnostic accuracy of phenotype, or sex-specific disease risks. A factor that can be viewed as both a strength and weakness of MF methods is the large number of statistical approaches and implementation options of model-free methods; while providing a number of choices for the more sophisticated users, such variety also may lead to the risk of overanalysis of the data by selecting the approach that gives the desired result. In the end, the choice between model-free and model-based methods will largely depend on the nature of the phenotype under study and the existing knowledge base about its underlying mode of inheritance.