Abstract
Introduction
The
diagnosis
of
COVID-19
is
normally
based
on
the
qualitative
detection
viral
nucleic
acid
sequences.
Properties
host
response
are
not
measured
but
key
in
determining
outcome.
Although
metabolic
profiles
well
suited
to
capture
state,
most
metabolomics
studies
either
underpowered,
measure
only
a
restricted
subset
metabolites,
compare
infected
individuals
against
uninfected
control
cohorts
that
suitably
matched,
or
do
provide
compact
predictive
model.
Objectives
Here
we
well-powered,
untargeted
assessment
120
patient
samples
acquired
at
hospital
admission.
study
aims
predict
patient’s
infection
severity
(i.e.,
mild
severe)
and
potential
outcome
discharged
deceased).
Methods
High
resolution
UHPLC-MS/MS
analysis
was
performed
serum
using
both
positive
negative
ionization
modes.
A
20
intermediary
metabolites
were
selected
univariate
statistical
significance
multiple
predictor
Bayesian
logistic
regression
model
created.
Results
predictors
for
their
relevant
biological
function
include
deoxycytidine
ureidopropionate
(indirectly
reflecting
load),
kynurenine
(reflecting
inflammatory
response),
short
chain
acylcarnitines
(energy
metabolism)
among
others.
Currently,
this
approach
predicts
with
Monte
Carlo
cross
validated
area
under
ROC
curve
0.792
(SD
0.09)
0.793
0.08),
respectively.
blind
validation
an
additional
90
patients
predicted
AUC
0.83
(CI
0.74–0.91)
0.76
0.67–0.86).
Conclusion
Prognostic
tests
markers
discussed
paper
could
allow
improvement
planning
treatment.
Metabolites,
Год журнала:
2020,
Номер
10(4), С. 135 - 135
Опубликована: Март 31, 2020
Metabolomics
analysis
generates
vast
arrays
of
data,
necessitating
comprehensive
workflows
involving
expertise
in
analytics,
biochemistry
and
bioinformatics
order
to
provide
coherent
high-quality
data
that
enable
discovery
robust
biologically
significant
metabolic
findings.
In
this
protocol
article,
we
introduce
notame,
an
analytical
workflow
for
non-targeted
profiling
approaches,
utilizing
liquid
chromatography–mass
spectrometry
analysis.
We
overview
lab
protocols
statistical
methods
commonly
practice
the
nutritional
metabolomics
data.
The
paper
is
divided
into
three
main
sections:
first
second
sections
introducing
background
study
designs
available
research
third
section
describing
detail
steps
used
produce,
preprocess
statistically
analyze
and,
finally,
identify
interpret
compounds
have
emerged
as
interesting.
Frontiers in Chemistry,
Год журнала:
2019,
Номер
7
Опубликована: Май 10, 2019
Drug
of
abuse
(DOA)
consumption
is
a
growing
problem
worldwide,
particularly
with
increasing
numbers
new
psychoactive
substances
(NPS)
entering
the
drug
market.
Generally,
little
information
on
their
adverse
effects
and
toxicity
are
available.
The
direct
detection
identification
NPS
an
analytical
challenge
due
to
ephemerality
scene.
An
approach
that
does
not
directly
focus
structural
analyte
or
its
metabolites,
would
be
beneficial
for
this
complex
scenario
development
alternative
screening
methods
could
help
provide
fast
response
suspected
consumption.
A
metabolomics
might
represent
such
strategy
biomarkers
different
questions
in
DOA
testing.
Metabolomics
monitoring
changes
small
(endogenous)
molecules
(<1,000
Da)
certain
stimulus,
e.g.,
For
review,
literature
search
targeting
"metabolomics"
DOAs
was
conducted.
Thereby,
applications
metabolomic
strategies
biomarker
research
were
identified:
(a)
as
additional
tool
metabolism
studies
bearing
major
advantage
priori
unknown
unexpected
metabolites
can
identified;
(b)
endogenous
metabolite
patterns,
synthetic
cannabinoids
also
indirectly
detect
urine
manipulation
attempts
by
chemical
adulteration
replacement
artificial
samples.
majority
currently
available
field,
however,
deals
better
assess
acute
chronic
find
addiction
tolerance.
Certain
compounds
detected
all
studied
DOAs,
but
often
similar
compounds/pathways
influenced.
When
evaluating
these
regard
possible
consumption,
observed
appear,
albeit
statistically
significant,
too
reliably
work
Further,
drugs
shown
affect
same
pathways.
In
conclusion,
approaches
possess
potential
indicating
More
studies,
including
more
sensitive
targeted
analyses,
multi-variant
statistical
models
deep-learning
needed
fully
explore
omics
science
Briefings in Bioinformatics,
Год журнала:
2020,
Номер
22(2), С. 1531 - 1542
Опубликована: Авг. 11, 2020
Deep
learning
(DL),
an
emerging
area
of
investigation
in
the
fields
machine
and
artificial
intelligence,
has
markedly
advanced
over
past
years.
DL
techniques
are
being
applied
to
assist
medical
professionals
researchers
improving
clinical
diagnosis,
disease
prediction
drug
discovery.
It
is
expected
that
will
help
provide
actionable
knowledge
from
a
variety
'big
data',
including
metabolomics
data.
In
this
review,
we
discuss
applicability
metabolomics,
while
presenting
discussing
several
examples
recent
research.
We
emphasize
use
tackling
bottlenecks
data
acquisition,
processing,
metabolite
identification,
as
well
metabolic
phenotyping
biomarker
Finally,
how
used
genome-scale
modelling
interpretation
The
DL-based
approaches
discussed
here
may
computational
biologists
with
integration,
drawing
statistical
inference
about
biological
outcomes,
based
on
European Journal of Mass Spectrometry,
Год журнала:
2020,
Номер
26(3), С. 165 - 174
Опубликована: Апрель 10, 2020
Data
normalization
is
a
big
challenge
in
quantitative
metabolomics
approaches,
whether
targeted
or
untargeted.
Without
proper
normalization,
the
mass-spectrometry
and
spectroscopy
data
can
provide
erroneous,
sub-optimal
data,
which
lead
to
misleading
confusing
biological
results
thereby
result
failed
application
human
healthcare,
clinical,
other
research
avenues.
To
address
this
issue,
number
of
statistical
approaches
software
tools
have
been
proposed
literature
implemented
over
years,
providing
multitude
choose
from
–
either
sample-based
data-based
strategies.
In
recent
new
dedicated
for
surfaced
as
well.
account
article,
I
summarize
existing
discoveries
findings
area
introduce
some
that
aid
normalization.
Abstract
Introduction
The
diagnosis
of
COVID-19
is
normally
based
on
the
qualitative
detection
viral
nucleic
acid
sequences.
Properties
host
response
are
not
measured
but
key
in
determining
outcome.
Although
metabolic
profiles
well
suited
to
capture
state,
most
metabolomics
studies
either
underpowered,
measure
only
a
restricted
subset
metabolites,
compare
infected
individuals
against
uninfected
control
cohorts
that
suitably
matched,
or
do
provide
compact
predictive
model.
Objectives
Here
we
well-powered,
untargeted
assessment
120
patient
samples
acquired
at
hospital
admission.
study
aims
predict
patient’s
infection
severity
(i.e.,
mild
severe)
and
potential
outcome
discharged
deceased).
Methods
High
resolution
UHPLC-MS/MS
analysis
was
performed
serum
using
both
positive
negative
ionization
modes.
A
20
intermediary
metabolites
were
selected
univariate
statistical
significance
multiple
predictor
Bayesian
logistic
regression
model
created.
Results
predictors
for
their
relevant
biological
function
include
deoxycytidine
ureidopropionate
(indirectly
reflecting
load),
kynurenine
(reflecting
inflammatory
response),
short
chain
acylcarnitines
(energy
metabolism)
among
others.
Currently,
this
approach
predicts
with
Monte
Carlo
cross
validated
area
under
ROC
curve
0.792
(SD
0.09)
0.793
0.08),
respectively.
blind
validation
an
additional
90
patients
predicted
AUC
0.83
(CI
0.74–0.91)
0.76
0.67–0.86).
Conclusion
Prognostic
tests
markers
discussed
paper
could
allow
improvement
planning
treatment.