Hi Randy,
since I'm a big fan of using ensembles for MR as well, one question:
I was under the impression that it wasn't necessary to trim the
deviating loops and other bits that don't match because Phaser's ML
approach would take care of that. Is that not so?
Thanks and all best.
Andreas
On 29/07/2016 5:46, Randy Read wrote:
> Dear Almu,
>
> First, if there is tNCS then there should be a large off-origin peak in
> the native Patterson map, which Phaser (as well as pointless or xtriage)
> should detect. If you look at the section of the log file about tNCS,
> has it found a big Patterson peak? If not, you must have another problem.
>
> Second, a solution with a TFZ of 5.8 and an LLG of 44 could be correct,
> but it might also be incorrect. We like to see a TFZ above 7 or 8 and
> an LLG above 60.
>
> In a difficult case, it can be very useful to modify your model with
> programs like Sculptor or Chainsaw. Sculptor will even try a number of
> different protocols, and it may be that one of these protocols will give
> you a model that works better than others. An even better option, in
> many cases, is to make an ensemble of alternative models, as long as
> they superimpose reasonably well on each other, and turning on the
> option in Ensembler to trim off deviating loops can help a great deal.
>
> Best wishes,
>
> Randy
>
|