Figure 1.
A directed graphical model of the ensemble model (on the left) and its interplay with a fine-grained conformational prior distribution (top right) through the reference ratio method, (bottom right).
In the graphical model, the black circles are random variables, and the arrows determine their conditional independencies. The parameter , marked in grey on the left, is fixed and given, and denotes the experimental error in this particular example.
denotes the reference distribution.
Figure 2.
A random backbone conformation of the GB1 hairpin.
The restraints listed in Table 1 are shown as dashed lines. The distance shown in red is used as the reaction coordinate used in Figures 3 and 4. This figure was created using PyMOL (DeLano Scientific LCC).
Figure 3.
Histograms, , of a representative pairwise distance
(between C
-C
, in
) in the ensembles.
The black and blue lines are obtained from the PROFASI and ISD ensembles respectively, while the cyan line represent the prior . Finally, the green and red lines respectively represent ensembles obtained from the power-averaged and linearly averaged data.
Figure 4.
The influence of and
on the ensembles.
The figure shows histograms, , of a representative pairwise distance
(between C
-C
, in Å ) in the ensembles obtained without the reference distribution
or the scale vector
. The black line denotes the PROFASI target ensemble; the red and green lines denote the ensembles obtained using the linearly and the power averaged data, respectively. The blue line denotes the case of the power averaged data without
, but with
.
Table 1.
Synthetic datasets used in this study.