Predicting Coaxial Stacking by Free Energy Minimization David Mathews Department of Biochemistry &...
-
date post
20-Dec-2015 -
Category
Documents
-
view
219 -
download
1
Transcript of Predicting Coaxial Stacking by Free Energy Minimization David Mathews Department of Biochemistry &...
Predicting Coaxial Stacking by Free
Energy Minimization
David Mathews
Department of Biochemistry & Biophysics
University of Rochester Medical Center
A step towards tertiary structure prediction
Secondary structure representation
1 stacked on 2 mediated by mismatch
2 flush stacked on 3
Flush and Mismatch-Mediated Stacking
Mismatch-mediated stacking
Flush stacking
- Stacking stabilization: Thought to arise from hydrophobic effect, charge interactions and van der Waals interactions.
Predicting Coaxial Stacking
- Find all the non-redundant RNA crystal structures from NDb.
“The stacking configuration with lowest free energy as predicted by Nearest Neighbour Parameters exists in naturally occurring RNAs.”
- Compare predictions with crystal structures.
- Predict the coaxial stacking configuration by finding free energy of all possible configurations in all MBLs.
Hypothesis
Nearest Neighbor Model for Coaxial Stacking
Model based on work by Walter, Kim and others in Turner lab.
Stacking Definition for Verification
Basepair center and basepair plane definition
from Biochemistry 2nd Ed. by Garrett & Grisham
Coaxial Stacking Discovery
Criteria for stacking
a. Basepair plane tilt< 26º for Flush / 32º for MM
N1
N1N2
D1-2
b. Distance between basepair “centers” < 5 Å for Flush / 12 Å for MM
(based on Gabb et al., J. Mol. Graph., 14, 6-11Burkard et al., JMB, 290, 967-982and Gendron at al., JMB, 308, 919-936
Stacking Definition for Verification
c. Basepair shear
angle between inter-center vector
and baseplane normal vectors
< 60º
Capturing Complex Stacks
relaxed tilt and distance criteria:
distance of basepair centers from normal to the other basepair < 10 Å
RNA structure dataset
The ribosome RNA structures provide maximum data.
Data distribution by RNA type
0
10
20
30
40
50
60
70
80
90
100
tRNA ribozymes rRNA others
RNA type
nu
mb
er
of
MB
Ls/
stack
s
MBLs Predicted Stacks Total Stacks
Dependence on MBL Size(no. of branches)
PPV and Sensitivity dependence on number of branches in MBLs
0
10
20
30
40
50
60
70
80
90
100
3 4 5 6 7 8
Number of branches
PP
V/S
en
sitiv
ity
(%
)
0
20
40
60
80
100
120
nu
mb
er o
f MB
Ls
PPV Sensitivity MBLs
Dependence on MBL Size(no. of bases)
PPV and sensitivity dependence on the size of MBL
0
10
20
30
40
50
60
70
80
90
100
8-11 12-15 16-19 20-23 24-27 28-31 32-35 36-39 40-56
number of bases in MBL
PP
V/S
en
sitiv
ity
(%
)
0
5
10
15
20
25
30
35
40
45
50
nu
mb
er o
f MB
Ls
PPV Sensitivity number of MBLs
Suboptimal ConfigurationsThe Problem with Lowest Free Energy
Consider,
K3/2 < K1, K2 < K3
A stack is more probable if it is part of many different configurations of low free energy.
Just 4 out of 51 possible configurations!
Partition Function and Configuration Probabilities
• PT = Σi exp(-ΔGi/RT) where i varies over ALL the possible configurations.
• PR,S = Σj exp(-ΔGj/RT) where j varies over all the possible configurations that have stack S.
• ps = PR,S / PT
Probability Threshold for Prediction
Both plots show a sharp drop at 0.70
So 70% was chosen to be the cut-off value for prediction