[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [modeller_usage] modeller ga341 score



Hi

Okay I have calculated the DOPE scores and I will use these. Just out of interest, is there are way of utilising the Z scores or compactness scores from GA341? From the example below I am assuming that the lower the Z score the better, would this be correct? Also when it comes to the compactness value is a higher or lower score better?

Many thanks
Jo

model1
% sequence identity      :    30.174000
Sequence length          :          531
Compactness              :     0.270804
Native energy (pair)     :  -225.273799
Native energy (surface)  :   -13.946407
Native energy (combined) :    -6.357012
Z score (pair)           :    -9.813731
Z score (surface)        :    -7.147611
Z score (combined)       :   -11.464915
GA341 score              :     1.000000

model2
% sequence identity      :    30.174000
Sequence length          :          531
Compactness              :     0.256617
Native energy (pair)     :  -268.422282
Native energy (surface)  :   -16.727187
Native energy (combined) :    -7.618686
Z score (pair)           :    -9.969647
Z score (surface)        :    -7.755271
Z score (combined)       :   -12.156111
GA341 score              :     1.000000
On 2/21/06, Modeller Caretaker <">> wrote:
jo hanna wrote:
> After having built my models, i have decided to assess them using the
> ga341 score.

> the ouput of this script for two different models is:
>
> model1
> GA341 score              :     1.000000
>
> model2
> GA341 score              :     1.000000
>
> how do i interpret these outputs? should both scores be 1?

GA341 is not sufficiently sensitive to distinguish between two 'very
good' models. Any good model will give a score very close to 1 ( e.g.
running GA341 on most PDB structures will give a 1.0 score). You should
use it only to discard 'bad' models (e.g. those with score less than
0.6). DOPE may be more sensitive in your case.

        Ben Webb, Modeller Caretaker
--
">             http://www.salilab.org/modeller/
Modeller mail list: http://salilab.org/mailman/listinfo/modeller_usage