Hi all, bug or feature? Just performed BE on a conventional 2x2 study, but due to logistic reasons subjects in the study were split into two groups. Fixed effects in the model were: Sequence+Treatment+Period+Group+Group*Treatment (Bolton & Bon, Pharmaceutical Statistics, 2004). Performed the evaluation on parent compound. Fine. Next I copypasted the BE-object in the workflow. When I changed the source to the metabolite (same data structure), all mappings were kept, but the fixed effects switched back to the conventional ones (Sequence+Treatment+Period). No big deal, but may go unnoticed if not checked in the corresponding tab.
Hhm, obviously there’s a time limit for editing one own’s post… OK, replying myself. It’s getting even worse: If I ran the model on PK metrics X and Y, and later on add another metric Z, the variance structure returns to the default. Now I would call it a bug.
Hi Helmut, I think this is an example of a problem that would be better reported through the Support Portal, please can you send an example project illustrating the copied object losing the model settings. Thanks, Simon. PS I will look into how long the edit time is set for users. [color=#FF0000] PPS Tyler just changed the edit window to 24 hours. It was previously set to 30 minutes, hope this helps.[/color]
Hi Simon, yes, I will notify support - but it sooo easy to give it a try: Open wonderful ‘2x2 Data.csv’ > Edit in Excel, add another column of PK metrics (e.g., Cmax, =E2/2, drag down to fill), add another column ‘Group’, 1 to subjects 1-10 and 2 to 20 > return to Phoenix > BE as usual, add ‘Group’ as Classification, Dependent AUClast, Fixed Effects: Sequence+Formulation+Period+Group+Group*Formulation execute the workflow, fine. Keeping the Fixed Effects tab open, switch from the Results tab to Setup again. With one eye watch the FE tab while with the other control mapping Cmax as Dependent. [size=4][color=#FF0000]Ooops.[/color][/size] Edit: I tried to contact support (see screenshot); not sure whether it worked… 
Dear Helmut, Thank you for bringing this up to our attention. It does appear that changing a dependent variable resets the model to the default. We have logged this in our tracking system and will try to address as soon as possible. The tracking number is QC 9541. We will continue to reset the model to the default if a different dataset is remapped as an input as we feel it makes sense that a different dataset should start from the default model. In the meantime Users are advise to review their model specification prior to execution. Sincerely, Ana Henry
Dear Ana, THX for the reply. [quote]It does appear that changing a dependent variable resets the model to the default.[/quote]Caution: Not only changing, but also adding another one. [quote]The tracking number is QC 9541.[/quote]BTW, any ideas why I couldn’t post the issue at the support site myself? [quote]We will continue to reset the model to the default if a different dataset is remapped as an input as we feel it makes sense that a different dataset should start from the default model.[/quote]Would you mind explaining your reasoning a little bit further? I didn’t get the point.