Reply To: all-atom modelling and clustering

[Anonymous]

Please check the following answers from Dr Bradley.

> Dear all
>
> I would like implement the most recent/accurate all-atom Rosetta protocols. In other words, 2 rounds, in the first of which quick backbone models are generated and clustered (the aa protocol). In round 2, members of the largest clusters are all-atom modelled (the cc relax protocol), reclustered and the lowest energy model of the largest cluster taken as the final model. I have a few queries
>
> – can the modelling of homologs of the target (Bradley et al 2005) be done automatically?
Not that I’m aware of. I did this using a collection of python scripts which have fallen into disuse.

> – is there a consensus view on whether single or multiple models should be generated from each starting point in round 2? (this was the subject of a previous thread but no clear opinion seemed to emerge)
My opinion would be that multiple models, say 3-5 should be generated due to the stochastic nature of the refinement protocol.

> – can Rosetta cluster the all-atom models at the end of round 2? the rosetta_scripts/abinitio/bin/cluster.pl script, as far as I can tell, works with a silentfile which would be the result of a single run, and would only contain backbone in any case. If Rosetta can’t, what tools would people recommend?
I’m not aware of all-atom clustering functionality in Rosetta.

> Many thanks in advance
>
> Dan