In the past six months, we had been investigating a novel method for identifying complex genetic traits in mice. This method involves application of Regression and Classification Trees, developed by L. Breiman, in combination with boosting technique (a specific type of resampling). We believe that this method may be able to identify loci involved in complex interaction without strongly pronounced effects on
their own. We have simulated examples of data sets where conventional methods such as MapMaker, say, fail; however, our method is able to find the loci. In addition, we have an example of a mice data set where our
method has succeeded to identify interacting loci whereas other methods have not. |
