Hongxu,

Considering that there are no other sources of variation in your experiment, your model is reasonable.

You can easily use SAS for that. You can sort your data by Marker and then run your analysis by marker. For example:

proc mixed data=you_data;

class QTL Location;

model Phenotype = QTL|Location;

by Marker;

run;

Some ideas:

1) I added "QTL" to the "class" statement. In that way, you are treating your genotypes as classes, and thus, you can write contrasts for the Additive and Dominance effects for your QTL, as well as these effects interacting with Location.

2) If you run by Marker, you have to "close" the output and save it to a data set using "ods output". For example, you can save LSMeans, P-values, Contrasts, Residuals, Variances, etc, to different datasets and then look at them at the end.

These are easy things to add to the SAS code. Please let me know if you need help.

I was at UIUC for my PhD at the ANSC department, but, I worked closely with Dr. Bollero from the CPSC department. With who do you work with?

Thanks, Nick

Also, you can fit more that one QTL in the model... It depends a lot of how large is your population and how many QTL you want to fit.

In addition, there are other methods that are more suitable when you are fitting several methods simultaneously (such as Bayesian Methods used for Genomic Selection).

Regards, Nick

Thanks Nick, you information is really informative. I came to UIUC last Aug and I work with Dr. Sacks. Dr. Bollero is our department head now and he teaches STAT courses in our department.