Technical capabilities and limitations

The algorithm that has been implemented in TreeLD is very computationally intensive. Therefore the available computational resources pose a limit on the size of the dataset that can be analyzed. The largest dataset we have worked with consists of 250 individuals, with 130 markers typed in each individual. A thorough analysis of this dataset took 48 hours on 10 processors in a Linux-cluster. The parts of the analysis that are computationally most intensive are the treebuilding step (see 6.2 and the generation of p-values by permutation (see 6.5).

The major determinant of the total time an analysis takes depends strongly on the initial burn-in of the MCMC. Section 8.1 provides an overview for these times. In section 8.4, we provide some concepts on how to apply TreeLD to large datasets and receive satisfying results.