On Power Estimation and Score Predictions in Affected Sib-Pair Linkage Analysis
This thesis consists of three papers in statistical genetics, with the emphasis on affected sib-pair linkage analysis.
In Paper A, a method to derive the power of linkage analysis in genome scans, using a stepwise approach, is presented. A simulation method to generate marker data is described and power functions are obtained combining the marker simulations with calculations of appropriate linkage statistic moments and a multivariate normal approximation. The approach is tested and illustrated in an example regarding multiple sclerosis.
Paper B treats the situation when the family reproduction processes are interfered by the births of affected offspring. The emphasis is on negative> interference, i.e. when the birth of an affected offspring reduces the probability of further reproduction, denoted limitation. The effects of limitation, in connection to the estimation of the sibling prevalence and the power of affected sib-pair analysis, are discussed. Some results are also given for the opposite case of positive interference, denoted compensation.
In Paper C, a prediction score graph is derived to compensate for incomplete information in linkage analysis. Contrary to the conservative nonparametric linkage (NPL) score, where the calculations are conditional on the null hypothesis of no linkage, the prediction score is derived conditional on an alternative hypothesis, estimated from the existing data. Performing a joint analysis, using both the NPL- and the prediction graph, new potential gene harbouring regions may be identified. Furthermore, regions originally displaying large peaks, despite low information content, can be reanalysed so that the disease loci can be located with an increased accuracy. The value of the prediction score is illustrated in an example regarding the celiac disease.
MSC 2000 subject classification: 62P10, 92D10
nonparametric linkage score