hat tip to NinjaGreg for clarifying what we're supposed to be doing.
I believe the idea is to take the 9 starting positions mentioned above and try to improve on them, or find a different folding theme. The 9 positions are the initial game start plus the 8 positions obtained by going into design mode, then "Show Alignment" then applying "Thread from Active Template" (the happy face) to each of the Reserve Templates (scroll with the <-- arrows --> near the bottom. Apologies for veterans to whom this is obvious.
Here's my results after initial shake / wiggle / shake for the 9 (your mileage may vary). Would like to save one or two folders from wasting the hour I just did ;-)
0 (initial game position) -- 8446
1 -- 9660; 2 -- 9598; 3 -- 9659; 4 -- 9642
5 -- 9675; 6 -- 9661; 7 -- 9623; 8 -- 9662
I looked at the reserve templates and they all looked identical, so I figured it was another puzzle set-up screw-up and didn't even try threading all those identical looking templates.
How would we know they are different without actually trying threading each one?
I am sorry for not making this clear and thank you John McLeod for bring this up.
Since we are giving you 9 different server models for the same CASP target, the sequences of each server model will be the same: namely, the target sequence of that particular protein that has yet to be solved and released.
You can just load them all up randomly (by resetting the
puzzle a bunch of times) but during CASP many players mentioned that they wanted to be able to differentiate between the different multi-start models by name (to describe which one they had started from to other players for example).
For this reason (and also so you can use partial threading between 2 different server models, creating a hybrid model that might be even better!) we have placed these 9 starting models into the Template Reserve.
Although they all have exactly the same sequence and alignment score (since that particular score is based entirely on sequence) when you thread them they will be very different models.
I apologize for the confusion and will make this more clear for future CASP9 Revisited Puzzles!