Simulation-based inference of single-molecule experiments
Current Opinion in Structural Biology,
Год журнала:
2025,
Номер
91, С. 102988 - 102988
Опубликована: Фев. 7, 2025
Язык: Английский
MDRefine: A Python package for refining molecular dynamics trajectories with experimental data
The Journal of Chemical Physics,
Год журнала:
2025,
Номер
162(19)
Опубликована: Май 15, 2025
Molecular
dynamics
(MD)
simulations
play
a
crucial
role
in
resolving
the
underlying
conformational
of
molecular
systems.
However,
their
capability
to
correctly
reproduce
and
predict
agreement
with
experiments
is
limited
by
accuracy
force-field
model.
This
can
be
improved
refining
structural
ensembles
or
parameters.
Furthermore,
discrepancies
experimental
data
due
imprecise
forward
models,
namely,
functions
mapping
simulated
structures
observables.
Here,
we
introduce
MDRefine,
Python
package
aimed
at
implementing
refinement
ensemble,
force
field,
and/or
model
comparing
MD-generated
trajectories
data.
The
software
consists
several
tools
that
employed
separately
from
each
other
combined
together
different
ways,
providing
seamless
interpolation
between
these
three
types
refinement.
We
use
some
benchmark
cases
show
approach
superior
applied
refinements.
MDRefine
has
been
released
as
an
open-source
under
LGPLv2+
license.
Source
code,
documentation,
examples
are
available
https://pypi.org/project/MDRefine
https://github.com/bussilab/MDRefine.
Язык: Английский
Model Selection Using Replica Averaging with Bayesian Inference of Conformational Populations
Journal of Chemical Theory and Computation,
Год журнала:
2025,
Номер
unknown
Опубликована: Июнь 2, 2025
Bayesian
Inference
of
Conformational
Populations
(BICePs)
is
a
reweighting
algorithm
that
reconciles
simulated
ensembles
with
sparse
and/or
noisy
observables
by
sampling
the
full
posterior
distribution
conformational
populations
in
presence
experimental
restraints.
By
modifying
BICePs
to
use
replica-averaging
its
forward
model,
becomes
similar
other
MaxEnt
approaches,
but
significant
advantages
(1)
being
able
sample
over
uncertainties
due
random
and
systematic
error,
improved
likelihoods
deal
outliers,
(2)
having
an
objective
score
for
model
selection,
free
energy-like
quantity
called
score.
To
demonstrate
power
our
approach,
we
used
reweight
mini-protein
chignolin
nine
different
force
fields
TIP3P
water,
using
set
158
measurements
(139
NOE
distances,
13
chemical
shifts,
6
vicinal
J-coupling
constants
HN
Hα).
In
all
cases,
reweighted
favor
correctly
folded
conformation.
The
score,
which
reports
energy
"turning
on"
along
restraints,
provides
metric
evaluate
each
field.
For
tested
(A14SB,
A99SB-ildn,
A99,
A99SBnmr1-ildn,
A99SB,
C22star,
C27,
C36,
OPLS-aa),
obtain
results
consistent
previous
work
conventional
χ2
selection
small
polypeptides
ubiquitin
(J.
Chem.
Theory
Comput.,
2012,
8,
1409-1414).
These
suggest
powerful
role
future
applications
requiring
ensemble
selection.
Язык: Английский
Towards accurate, force field independent conformational ensembles of intrinsically disordered proteins
bioRxiv (Cold Spring Harbor Laboratory),
Год журнала:
2024,
Номер
unknown
Опубликована: Окт. 5, 2024
Abstract
Determining
accurate
atomic
resolution
conformational
ensembles
of
intrinsically
disordered
proteins
(IDPs)
is
extremely
challenging.
Molecular
dynamics
(MD)
simulations
provide
atomistic
IDPs,
but
their
accuracy
highly
dependent
on
the
quality
physical
models,
or
force
fields,
used.
Here,
we
demonstrate
how
to
determine
IDPs
by
integrating
all-atom
MD
with
experimental
data
from
nuclear
magnetic
resonance
(NMR)
spectroscopy
and
small-angle
x-ray
scattering
(SAXS)
a
simple,
robust
fully
automated
maximum
entropy
reweighting
procedure.
We
that
when
this
approach
applied
sufficient
data,
IDP
derived
different
fields
converge
similar
distributions.
The
procedure
presented
here
facilitates
integration
extensive
datasets
enables
calculation
accurate,
force-field
independent
IDPs.
Язык: Английский