[phenixbb] Problem with low figure of merit
Thomas C. Terwilliger
terwilliger at lanl.gov
Tue Aug 31 09:02:17 PDT 2010
I would say this is a very promising result for a dataset with rather weak
phasing. Yes, phenix.autosol does estimate the solvent fraction for your
crystal. It initially guesses the number of NCS copies that places your
solvent fraction in the range of 0.35 to 0.65. If you know that your
solvent fraction is higher than this, then it is a good idea to specify,
or whatever your solvent fraction is. In recent versions of
phenix.autosol, once NCS is found, this can override the initial guess of
solvent fraction (but does not override a value that you specify).
You won't be able to improve your figure of merit of phasing with a change
in solvent content, but as you suspect, you might indeed be able to
improve the map.
I would recommend going on to use phenix.autobuild at this point,
specifying the solvent content in your inputs. I would expect this to
yield a good model for this structure.
All the best,
>> Hi all,
>> I am new to structure solving. I have a 2.9A dataset for a Seleno crystal
>> which i am trying to use Autosol to build an initial map. I used the
>> settings for a thorough solution, instead of quick. All the other
>> parameters were set to default settings.
>> The solution that i got from autosol, gave me the following result
>> Solution # 1 BAYES-CC: 23.7 +/- 30.9 Dataset #1 FOM: 0.27
>> Residues built: 238
>> Side-chains built: 139
>> Chains: 15
>> Overall model-map correlation: 0.65
>> R/R-free: 0.38/0.42.
>> Score type: SKEW CORR_RMS
>> Raw scores: 0.02 0.77
>> 100x EST OF CC: 16.86 32.11
>> The density modification map matches, quite well with many of my side
>> My question here is , IS THERE A WAY I CAN IMPROVE THE Figure Of Merit ,
>> CHANGING ANY PARAMETERS IN PHENIX-AUTOSOL.
>> I HAVE A HIGH SOLVENT CONTENT IN MY CRYSTAL AND I DID NOT INPUT THAT IN MY
>> INITIAL PARAMETERS. DOES PHENIX AUTOSOL CONSIDER SOLVENT FLATTENING FOR
>> I would appreciate someone's help in this regard
>> phenixbb mailing list
>> phenixbb at phenix-online.org
More information about the phenixbb