Simple controls exceed best deep learning algorithms and reveal foundation model effectiveness for predicting genetic perturbations
bioRxiv (Cold Spring Harbor Laboratory),
Journal Year:
2025,
Volume and Issue:
unknown
Published: Jan. 8, 2025
Abstract
Modeling
genetic
perturbations
and
their
effect
on
the
transcriptome
is
a
key
area
of
pharmaceutical
research.
Due
to
complexity
transcriptome,
there
has
been
much
excitement
development
in
deep
learning
(DL)
because
its
ability
model
complex
relationships.
In
particular,
transformer-based
foundation
paradigm
emerged
as
gold-standard
predicting
post-perturbation
responses.
However,
understanding
these
increasingly
models
evaluating
practical
utility
lacking,
along
with
simple
but
appropriate
benchmarks
compare
predictive
methods.
Here,
we
present
baseline
method
that
outperforms
both
state
art
(SOTA)
DL
other
proposed
simpler
neural
architectures,
setting
necessary
benchmark
evaluate
field
prediction.
We
also
elucidate
for
task
prediction
via
generalizable
fine-tuning
experiments
can
be
translated
different
applications
tasks
interest.
Furthermore,
provide
corrected
version
popular
dataset
used
benchmarking
perturbation
models.
Our
hope
this
work
will
properly
contextualize
further
space
control
procedures.
Language: Английский
Decoding the Molecular Language of Proteins with Evola
bioRxiv (Cold Spring Harbor Laboratory),
Journal Year:
2025,
Volume and Issue:
unknown
Published: Jan. 6, 2025
Abstract
Proteins,
nature’s
intricate
molecular
machines,
are
the
products
of
billions
years
evolution
and
play
fundamental
roles
in
sustaining
life.
Yet,
deciphering
their
language
-
that
is,
understanding
how
protein
sequences
structures
encode
determine
biological
functions
remains
a
cornerstone
challenge
modern
biology.
Here,
we
introduce
Evola,
an
80
billion
frontier
protein-language
generative
model
designed
to
decode
proteins.
By
integrating
information
from
sequences,
structures,
user
queries,
Evola
generates
precise
contextually
nuanced
insights
into
function.
A
key
innovation
lies
its
training
on
unprecedented
AI-generated
dataset:
546
million
question-answer
pairs
150
word
tokens,
reflect
immense
complexity
functional
diversity
Post-pretraining,
integrates
Direct
Preference
Optimization
(DPO)
refine
based
preference
signals
Retrieval-Augmented
Generation
(RAG)
for
external
knowledge
incorporation,
improving
response
quality
relevance.
To
evaluate
performance,
propose
novel
framework,
Instructional
Response
Space
(IRS),
demonstrating
delivers
expert-level
insights,
advancing
research
proteomics
genomics
while
shedding
light
logic
encoded
The
online
demo
is
available
at
http://www.chat-protein.com/
.
Language: Английский
Ai-enabled language models (LMs) to large language models (LLMs) and multimodal large language models (MLLMs) in drug discovery and development
Journal of Advanced Research,
Journal Year:
2025,
Volume and Issue:
unknown
Published: Feb. 1, 2025
Language: Английский
A Review of Recent Artificial Intelligence for Traditional Medicine
Journal of Traditional and Complementary Medicine,
Journal Year:
2025,
Volume and Issue:
unknown
Published: Feb. 1, 2025
Language: Английский
Evaluation of the performance of ChatGPT‐4 and ChatGPT‐4o as a learning tool in endodontics
International Endodontic Journal,
Journal Year:
2025,
Volume and Issue:
unknown
Published: March 2, 2025
Abstract
Aims
The
aim
of
this
study
was
to
evaluate
the
accuracy
and
consistency
responses
given
by
two
different
versions
Chat
Generative
Pre‐trained
Transformer
(ChatGPT),
ChatGPT‐4,
ChatGPT‐4o,
multiple‐choice
questions
prepared
from
undergraduate
endodontic
education
topics
at
times
day
on
days.
Methodology
In
total,
60
multiple‐choice,
text‐based
6
were
prepared.
Each
question
asked
ChatGPT‐4
ChatGPT‐4o
3
a
(morning,
noon,
evening)
for
consecutive
AIs
compared
using
SPSS
R
programs
(
p
<
.05,
95%
confidence
interval).
Results
rate
(92.8%)
significantly
higher
than
that
(81.7%;
.001).
groups
affected
rates
both
which
did
not
affect
either
AI
>
.05).
There
no
statistically
significant
difference
in
between
=
.123).
AI,
too
Conclusions
According
results
study,
better
ChatGPT‐4.
These
findings
demonstrate
chatbots
can
be
used
dental
education.
However,
it
is
also
necessary
consider
limitations
potential
risks
associated
with
AI.
Language: Английский
Prompt-engineering enabled LLM or MLLM and instigative bioinformatics pave the way to identify and characterize the significant SARS-CoV-2 antibody escape mutations
International Journal of Biological Macromolecules,
Journal Year:
2024,
Volume and Issue:
unknown, P. 138547 - 138547
Published: Dec. 1, 2024
Language: Английский