bioRxiv (Cold Spring Harbor Laboratory),
Год журнала:
2024,
Номер
unknown
Опубликована: Июль 24, 2024
Several
peptide-based
drugs
fail
in
clinical
trials
due
to
their
toxicity
or
hemolytic
activity
against
red
blood
cells
(RBCs).
Existing
methods
predict
peptides
but
not
the
concentration
(HC50)
required
lyse
50%
of
RBCs.
In
this
study,
we
developed
a
classification
model
and
regression
identify
quantify
peptides.
Our
models
were
trained
validated
on
1924
with
experimentally
determined
HC50
mammalian
Analysis
indicates
that
hydrophobic
positively
charged
residues
associated
higher
activity.
achieved
maximum
AUC
0.909
using
hybrid
ESM-2
motif-based
approach.
Regression
compositional
features
R
0.739
R2
0.543.
outperform
existing
are
implemented
web-based
platform
HemoPI2
standalone
software
for
designing
desired
values
(http://webs.iiitd.edu.in/raghava/hemopi2/).
Briefings in Bioinformatics,
Год журнала:
2024,
Номер
26(1)
Опубликована: Ноя. 22, 2024
Cytometry
is
an
advanced
technique
for
simultaneously
identifying
and
quantifying
many
cell
surface
intracellular
proteins
at
a
single-cell
resolution.
Analyzing
high-dimensional
cytometry
data
involves
populations
based
on
their
marker
expressions.
This
study
provided
quantitative
review
comparison
of
various
ways
to
phenotype
cellular
within
the
data,
including
manual
gating,
unsupervised
clustering,
supervised
auto-gating.
Six
datasets
from
diverse
species
sample
types
were
included
in
study,
gating
with
two
hierarchical
layers
was
used
as
truth
evaluation.
For
results
five
researchers
compared
illustrate
consistency
among
different
raters.
23
tools
quantitatively
terms
accuracy
computing
cost.
While
no
method
outperformed
all
others,
several
tools,
PAC-MAN,
CCAST,
FlowSOM,
flowClust,
DEPECHE,
generally
demonstrated
strong
performance.
auto-gating
methods,
four
algorithms
evaluated,
where
DeepCyTOF
CyTOF
Linear
Classifier
performed
best.
We
further
practical
recommendations
prioritizing
methods
application
scenarios.
offers
comprehensive
insights
biologists
understand
choose
best-suited
ones
applications.
bioRxiv (Cold Spring Harbor Laboratory),
Год журнала:
2024,
Номер
unknown
Опубликована: Июль 24, 2024
Several
peptide-based
drugs
fail
in
clinical
trials
due
to
their
toxicity
or
hemolytic
activity
against
red
blood
cells
(RBCs).
Existing
methods
predict
peptides
but
not
the
concentration
(HC50)
required
lyse
50%
of
RBCs.
In
this
study,
we
developed
a
classification
model
and
regression
identify
quantify
peptides.
Our
models
were
trained
validated
on
1924
with
experimentally
determined
HC50
mammalian
Analysis
indicates
that
hydrophobic
positively
charged
residues
associated
higher
activity.
achieved
maximum
AUC
0.909
using
hybrid
ESM-2
motif-based
approach.
Regression
compositional
features
R
0.739
R2
0.543.
outperform
existing
are
implemented
web-based
platform
HemoPI2
standalone
software
for
designing
desired
values
(http://webs.iiitd.edu.in/raghava/hemopi2/).
