Journal of Cheminformatics,
Год журнала:
2025,
Номер
17(1)
Опубликована: Май 5, 2025
Organic
anion
transporting
polypeptides
(OATPs)
are
membrane
transporters
crucial
for
drug
uptake
and
distribution
in
the
human
body.
OATPs
can
mediate
drug-drug
interactions
(DDIs)
which
interaction
of
one
with
an
OATP
impairs
another
drug,
resulting
potentially
fatal
pharmacological
effects.
Predicting
OATP-mediated
DDIs
is
challenging,
due
to
limited
information
on
inhibition
mechanisms
inconsistent
experimental
data
across
different
studies.
This
study
introduces
Heterogeneous
OATP-Ligand
Interaction
Graph
Neural
Network
(HOLIgraph),
a
novel
computational
model
that
integrates
molecular
modeling
graph
neural
network
enhance
prediction
drug-induced
inhibition.
By
combining
ligand
(i.e.,
drug)
features
protein-ligand
from
rigorous
docking
simulations,
HOLIgraph
outperforms
traditional
DDI
models
rely
solely
features.
achieved
median
balanced
accuracy
over
90
percent
when
predicting
inhibitors
OATP1B1,
significantly
outperforming
purely
ligand-based
models.
Beyond
improving
prediction,
used
train
enable
characterization
protein
residues
involved
inhibitory
drug-OATP
interactions.
We
identified
certain
OATP1B1
preferentially
interact
inhibitors,
including
I46
K49.
anticipate
such
will
be
valuable
future
structural
mechanistic
investigations
OATP1B1.
Expert Review of Proteomics,
Год журнала:
2025,
Номер
unknown
Опубликована: Янв. 17, 2025
The
DeepMind's
AlphaFold
(AF)
has
revolutionized
biomedical
research
by
providing
both
experts
and
non-experts
with
an
invaluable
tool
for
predicting
protein
structures.
However,
while
AF
is
highly
effective
structures
of
rigid
globular
proteins,
it
not
able
to
fully
capture
the
dynamics,
conformational
variability,
interactions
proteins
ligands
other
biomacromolecules.
In
this
review,
we
present
a
comprehensive
overview
latest
advancements
in
3D
model
predictions
biomacromolecules
using
AF.
We
also
provide
detailed
analysis
its
strengths
limitations,
explore
more
recent
iterations,
modifications,
practical
applications
strategy.
Moreover,
map
path
forward
expanding
landscape
toward
every
peptide
proteome
most
physiologically
relevant
form.
This
discussion
based
on
extensive
literature
search
performed
PubMed
Google
Scholar.
While
significant
progress
been
made
enhance
AF's
modeling
capabilities,
argue
that
combined
approach
integrating
various
silico
vitro
methods
will
be
beneficial
future
structural
biology,
bridging
gaps
between
static
dynamic
features
their
functions.
bioRxiv (Cold Spring Harbor Laboratory),
Год журнала:
2025,
Номер
unknown
Опубликована: Фев. 21, 2025
Abstract
Advances
in
machine
learning
have
transformed
structural
biology,
enabling
swift
and
accurate
prediction
of
protein
structure
from
sequence.
However,
challenges
persist
capturing
sidechain
packing,
condition-dependent
conformational
dynamics,
biomolecular
interactions,
primarily
due
to
scarcity
high-quality
training
data.
Emerging
techniques,
including
cryo-electron
tomography
(cryo-ET)
high-throughput
crystallography,
promise
vast
new
sources
data,
but
translating
raw
experimental
observations
into
mechanistically
interpretable
atomic
models
remains
a
key
bottleneck.
Here,
we
aim
address
these
by
improving
the
efficiency
analysis
through
combining
measurements
with
landmark
method
–
AlphaFold2.
We
present
an
augmentation
AlphaFold2,
ROCKET,
that
refines
its
predictions
using
cryo-EM,
cryo-ET,
X-ray
crystallography
demonstrate
this
approach
captures
biologically
important
variation
AlphaFold2
does
not.
By
performing
optimization
space
coevolutionary
embeddings,
rather
than
Cartesian
coordinates,
ROCKET
automates
difficult
modeling
tasks,
such
as
flips
functional
loops
domain
rearrangements,
are
beyond
scope
current
state-of-the-art
methods
and,
some
instances,
even
manual
human
modeling.
The
ability
efficiently
sample
barrier-crossing
rearrangements
unlocks
horizon
for
scalable
automated
model
building.
Crucially,
not
require
retraining
is
readily
adaptable
multimers,
ligand-cofolding,
other
data
modalities.
Conversely,
our
differentiable
crystallographic
cryo-EM
target
functions
capable
augmenting
methods.
thus
provides
extensible
framework
integration
observables
learning.
Biomolecules,
Год журнала:
2025,
Номер
15(3), С. 423 - 423
Опубликована: Март 17, 2025
Understanding
protein
structures
can
facilitate
the
development
of
therapeutic
drugs.
Traditionally,
have
been
determined
through
experimental
approaches
such
as
X-ray
crystallography,
NMR
spectroscopy,
and
cryo-electron
microscopy.
While
these
methods
are
effective
considered
gold
standard,
they
very
resource-intensive
time-consuming,
ultimately
limiting
their
scalability.
However,
with
recent
developments
in
computational
biology
artificial
intelligence
(AI),
field
prediction
has
revolutionized.
Innovations
like
AlphaFold
RoseTTAFold
enable
structure
predictions
to
be
made
directly
from
amino
acid
sequences
remarkable
speed
accuracy.
Despite
enormous
enthusiasm
associated
newly
developed
AI-approaches,
true
potential
structure-based
drug
discovery
remains
uncertain.
In
fact,
although
algorithms
generally
predict
overall
well,
essential
details
for
ligand
docking,
exact
location
side
chains
within
binding
pocket,
not
predicted
necessary
Additionally,
docking
methodologies
more
a
hypothesis
generator
rather
than
precise
predictor
ligand–target
interactions,
thus,
usually
identify
many
false-positive
hits
among
only
few
correctly
interactions.
this
paper,
we
reviewing
latest
cutting-edge
emphasis
on
GPCR
target
class
assess
role
AI
discovery.
Communications Chemistry,
Год журнала:
2025,
Номер
8(1)
Опубликована: Апрель 7, 2025
Accurate
protein-ligand
binding
affinity
prediction
is
crucial
in
drug
discovery.
Existing
methods
are
predominately
docking-free,
without
explicitly
considering
atom-level
interaction
between
proteins
and
ligands
scenarios
where
crystallized
conformations
unavailable.
Now,
with
breakthroughs
deep
learning
AI-based
protein
folding
conformation
prediction,
can
we
improve
prediction?
This
study
introduces
a
framework,
Folding-Docking-Affinity
(FDA),
which
folds
proteins,
determines
conformations,
predicts
affinities
from
three-dimensional
structures.
Our
experimental
results
indicate
that
FDA
performs
comparably
to
state-of-the-art
docking-free
methods.
We
anticipate
our
proposed
framework
serves
as
starting
point
for
integrating
structures
more
accurate
prediction.
Journal of Cheminformatics,
Год журнала:
2025,
Номер
17(1)
Опубликована: Май 5, 2025
Organic
anion
transporting
polypeptides
(OATPs)
are
membrane
transporters
crucial
for
drug
uptake
and
distribution
in
the
human
body.
OATPs
can
mediate
drug-drug
interactions
(DDIs)
which
interaction
of
one
with
an
OATP
impairs
another
drug,
resulting
potentially
fatal
pharmacological
effects.
Predicting
OATP-mediated
DDIs
is
challenging,
due
to
limited
information
on
inhibition
mechanisms
inconsistent
experimental
data
across
different
studies.
This
study
introduces
Heterogeneous
OATP-Ligand
Interaction
Graph
Neural
Network
(HOLIgraph),
a
novel
computational
model
that
integrates
molecular
modeling
graph
neural
network
enhance
prediction
drug-induced
inhibition.
By
combining
ligand
(i.e.,
drug)
features
protein-ligand
from
rigorous
docking
simulations,
HOLIgraph
outperforms
traditional
DDI
models
rely
solely
features.
achieved
median
balanced
accuracy
over
90
percent
when
predicting
inhibitors
OATP1B1,
significantly
outperforming
purely
ligand-based
models.
Beyond
improving
prediction,
used
train
enable
characterization
protein
residues
involved
inhibitory
drug-OATP
interactions.
We
identified
certain
OATP1B1
preferentially
interact
inhibitors,
including
I46
K49.
anticipate
such
will
be
valuable
future
structural
mechanistic
investigations
OATP1B1.