Contents - Index
Remove collinear independent variables in multiple regression
This function becomes activated after multiple regression is completed.
In the biplot, all retained markers can be said to have "independent" effect on the target trait kernel weight; each represents a different QTL. The number of QTL can actually be counted: it is about 7 to 10. These markers explained 69% of the KW variation.
If one uses a log ratio of 3.5, further markers will be deleted:
Now only 5 markers, representing probably 4 QTL are left. They explained 62% of the KW variation, implying that the 6 markers that were deleted explained only 7% of the KW variation.
1) Do the two markers on chromosome 1, VATP57a and MWG626, represent a single QTL or two different QTL since they have independent effects and are closely linked with a distance of about 5 cM? It is likely that they are flanking markers of a single QTL.
2) How log ratio is calculated