For the specified trait, SRmapqtl will output a small table:
-------------------------------------------------------
-------------------------------------------------------
Chromosome Marker Rank F-Stat DOF
------------------------------------------------------- -start
1 1 2 13.38778 114
2 3 4 10.12742 110
3 1 5 3.55528 108
3 2 3 11.15490 112
4 3 1 28.85236 116 -end
-------------------------------------------------------
-------------------------------------------------------
The first two columns indicate the chromosome and marker. The third column gives the rank of that marker as determined by the stepwise regression mode of choice. Then there will be an F-statistic indicating the difference between having that variable in the model or not. Finally, the DOF (degrees of freedom) for the numerator of that F statistic is given. For forward stepwise or backward elimination, SRmapqtl will try to rank all of the markers no matter how small the F statistic is. For the forward regression with backward elimination, the program proceeds to add variables until the F statistic p-value is less than that specified by the -F option (0.1 by default). Then SRmapqtl rechecks all the variables added and will eliminate any with an F statistic p-value less than the value given with the -B option.
In general, the FB method is probably the best method for picking background markers to be used with model 6 in Zmapqtl and JZmapqtl. To this end, SRmapqtl should be run prior to using either module. Zmapqtl and JZmapqtl will read the results of SRmapqtl and use the markers that are ranked. You can specify an upper bound to the number of background parameters to be used in Zmapqtl. JZmapqtl will use all the markers that are listed for all traits in its analysis: The FB method thus selects only a subset of significant markers.
Be aware that SRmapqtl tries to determine how many markers can be analyzed at once. The number of parameters has to be smaller than the sample size. If you try to use backward regression, and there are more markers than individuals, then SRmapqtl will default to forward stepwise regression and rank as many markers as possible. You should be aware that when dominance can be estimated, each marker will count two towards the total number of parameters and you will need a sample size of at least twice the number of markers to do backward elimination.