Hi everyone,
I have an urgent need of a suggestion..have any of you ever used the metaserver Procksi for structural comparison?I am using that for the first time to perform structural comparison between viral proteins showing no significant sequence similarity. For those who don' t know this server, briefly, it allows to compare proteins by using a variety of methods (e.g. Dali, TM, Vorolign, USM etc..) . The similarity matrices obtained by each method are then combined as to obtain a consensus similarity . In turn, this consensus similarity matrix can be used as input for a clustering process thus obtaining a consensus hierarchical clustering tree. (I send to you the link to the paper including the figure 4 with the clusters obtained)
http://www.biomedcentral.com/content/pdf/1471-2105-8-416.pdf
In my case, I produced clusters and I have hierarchical trees for my proteins, following the protocol described in the paper, but I do not know now how to analyze and validate the results of the clusters. I cannot compare the classification given by clusters with the structural classification provided by SCOP , because my proteins are not present in SCOP. In your opinion, if I just superpose the proteins (with some molecular visualization program) belonging to the same cluster and see if these are really similar (in terms of RMSD) could be a method to proceed? Which threshold of RMSD is better to consider in order to say that two proteins are significantly similar or not? And how can I distinguish among different groups of the hierarchical tree? From the first branching, the second or which one?
I hope to have clarified the problem I am completely new in this field.
Any help would be really appreciated.
Thank you very much.
Silvia
