I am trying to process a Titan K3 data set of an Icosahedral virus. I am trying to get a decent ab initio model. I submit it for 3 ab initio model generation. From the results it appears that it is close but not quite there. I am wondering if I change the following parameters to improve the ab initio model generation.
Starting and target resolution. Currently set at 35 and 12 A. I am thinking of changing it to 40 and 15. Another one is to vary the number of initial models from 3 to say 5. If the number of models are increased, does that improve the chances of getting a correct initial model?
Additionally, should I also increase the class similarity parameter from 0.1 (default) to higher value.
Also, there are two other points that I am not really clear about. Initial and final iterations before annealing starts and ends and the other one being class similarity annealing.