I
AE
S
I
n
t
e
r
n
at
ion
al
Jou
r
n
al
of
Ar
t
if
icial
I
n
t
e
ll
ig
e
n
c
e
(
I
J
-
AI
)
Vol.
14
,
No.
4
,
Augus
t
2025
,
pp.
2815
~
2825
I
S
S
N:
2252
-
8938
,
DO
I
:
10
.
11591/i
jai
.
v
14
.i
4
.
pp
28
15
-
2825
2815
Jou
r
n
al
h
omepage
:
ht
tp:
//
ij
ai
.
iaes
c
or
e
.
c
om
N
on
-
sm
al
l
c
e
ll
l
u
n
g
c
an
c
e
r
ac
t
iv
e
c
om
p
ou
n
d
s
d
is
c
ov
e
r
y
h
ol
d
in
g
on
p
r
ot
e
i
n
e
x
p
r
e
ssi
on
u
si
n
g
m
ac
h
in
e
l
e
a
r
n
in
g
m
od
e
ls
Ham
z
a
Han
af
i
1
,
M
’
h
am
e
d
Aï
t
K
b
ir
1
,
B
ad
r
Di
n
e
Ros
s
i
Has
s
an
i
2
1
I
nt
e
ll
ig
e
nt
A
ut
oma
ti
on
a
nd
B
io
M
e
dG
e
nomi
c
s
L
a
bor
a
to
r
y
,
S
T
S
M
D
oc
to
r
a
l
C
e
nt
e
r
,
A
bde
lm
a
le
k E
s
s
a
a
di
U
ni
ve
r
s
it
y,
T
a
ngi
e
r
,
M
or
oc
c
o
2
L
A
B
I
P
H
A
B
E
L
a
bor
a
to
r
y,
S
T
I
D
oc
to
r
a
l
C
e
nt
e
r
,
A
bde
lm
a
le
k E
s
s
a
a
di
U
ni
ve
r
s
it
y, T
a
ngi
e
r
, M
or
oc
c
o
Ar
t
icle
I
n
f
o
AB
S
T
RA
CT
A
r
ti
c
le
h
is
tor
y
:
R
e
c
e
ived
M
a
y
23,
2024
R
e
vis
e
d
F
e
b
2
5
,
2025
Ac
c
e
pted
M
a
r
15,
2025
Co
mp
u
t
a
t
i
o
n
a
l
met
h
o
d
s
h
a
v
e
t
ran
s
fo
rme
d
t
h
e
fi
el
d
o
f
d
r
u
g
d
i
s
co
v
ery
,
w
h
i
ch
s
i
g
n
i
f
i
ca
n
t
l
y
h
e
l
p
e
d
i
n
t
h
e
d
e
v
el
o
p
m
en
t
o
f
n
ew
t
re
at
men
t
s
.
N
o
w
a
d
ay
s
,
res
earch
er
s
are
ex
p
l
o
r
i
n
g
a
w
i
d
e
ran
g
er
o
f
o
p
p
o
r
t
u
n
i
t
i
e
s
t
o
i
d
en
t
i
f
y
n
e
w
co
mp
o
u
n
d
s
u
s
i
n
g
mach
i
n
e
l
ear
n
i
n
g
.
W
e
co
n
d
u
ct
e
d
a
co
mp
arat
i
v
e
s
t
u
d
y
b
et
w
een
mu
l
t
i
p
l
e
mo
d
e
l
s
cap
a
b
l
e
o
f
p
re
d
i
c
t
i
n
g
co
m
p
o
u
n
d
s
t
o
t
ar
g
et
n
o
n
-
s
mal
l
cel
l
l
u
n
g
can
cer
,
w
e
fo
cu
s
ed
o
n
i
n
t
e
g
rat
i
n
g
p
r
o
t
e
i
n
ex
p
res
s
i
o
n
s
t
o
i
d
e
n
t
i
fy
p
o
t
e
n
t
i
al
c
o
mp
o
u
n
d
s
t
h
a
t
ex
h
i
b
i
t
a
h
i
g
h
effi
ca
cy
i
n
t
ar
g
e
t
i
n
g
l
u
n
g
can
cer
cel
l
s
.
A
d
at
as
e
t
w
as
co
n
s
t
r
u
c
t
ed
b
as
e
d
o
n
t
h
e
t
ri
al
s
av
ai
l
a
b
l
e
i
n
t
h
e
Ch
E
MBL
d
at
a
b
as
e
.
T
h
en
,
mo
l
ecu
l
ar
d
e
s
cri
p
t
o
rs
w
ere
cal
cu
l
at
e
d
t
o
e
x
t
rac
t
s
t
ru
ct
u
re
-
act
i
v
i
t
y
rel
a
t
i
o
n
s
h
i
p
s
fr
o
m
t
h
e
s
e
l
ect
e
d
co
mp
o
u
n
d
s
a
n
d
fee
d
i
n
t
o
s
ev
era
l
mach
i
n
e
l
earn
i
n
g
mo
d
el
s
t
o
l
earn
fr
o
m
.
W
e
co
m
p
ared
t
h
e
p
erfo
rma
n
ce
o
f
v
ari
o
u
s
al
g
o
r
i
t
h
ms
.
T
h
e
mu
l
t
i
l
ay
er
p
ercep
t
ro
n
mo
d
e
l
ex
h
i
b
i
t
e
d
t
h
e
h
i
g
h
es
t
F1
s
co
re,
ac
h
i
e
v
i
n
g
an
o
u
t
s
t
an
d
i
n
g
v
a
l
u
e
o
f
0
,
8
6
1
.
Mo
reo
v
er,
w
e
p
res
en
t
a
l
i
s
t
o
f
1
0
d
ru
g
s
p
re
d
i
c
t
ed
as
act
i
v
e
i
n
l
u
n
g
can
cer,
al
l
o
f
w
h
i
ch
are
s
u
p
p
o
r
t
ed
b
y
rel
e
v
an
t
s
ci
e
n
t
i
fi
c
e
v
i
d
e
n
ce
i
n
t
h
e
med
i
cal
l
i
t
erat
u
re.
O
u
r
s
t
u
d
y
s
h
o
w
c
as
e
s
t
h
e
p
o
t
en
t
i
a
l
o
f
co
mb
i
n
i
n
g
p
ro
t
ei
n
ex
p
re
s
s
i
o
n
an
al
y
s
i
s
a
n
d
mac
h
i
n
e
l
ear
n
i
n
g
t
ech
n
i
q
u
e
s
t
o
i
d
e
n
t
i
fy
n
o
v
e
l
d
r
u
g
s
.
O
u
r
an
a
l
y
t
i
ca
l
ap
p
r
o
ach
co
n
t
r
i
b
u
t
e
s
t
o
t
h
e
d
ru
g
d
i
s
c
o
v
er
y
p
i
p
el
i
n
e
,
an
d
o
p
e
n
s
n
e
w
o
p
p
o
rt
u
n
i
t
i
es
t
o
ex
p
l
o
re
an
d
i
d
e
n
t
i
fy
n
e
w
t
ar
g
et
e
d
t
h
era
p
i
e
s
.
K
e
y
w
o
r
d
s
:
Dr
ug
dis
c
ove
r
y
L
ung
c
a
nc
e
r
M
a
c
hine
lea
r
ning
models
P
r
e
c
is
ion
medic
ine
P
r
otein
e
xpr
e
s
s
ions
Th
i
s
i
s
a
n
o
p
en
a
c
ces
s
a
r
t
i
c
l
e
u
n
d
e
r
t
h
e
CC
B
Y
-
SA
l
i
ce
n
s
e.
C
or
r
e
s
pon
din
g
A
u
th
or
:
Ha
mza
Ha
na
f
i
I
ntelli
ge
nt
Automation
a
nd
B
ioM
e
dGe
nomi
c
s
L
a
b
or
a
tor
y,
S
T
S
M
Doc
to
r
a
l
C
e
nter
Abde
lm
a
lek
E
s
s
a
a
di
Unive
r
s
it
y
T
a
ngier
,
M
or
oc
c
o
E
mail:
ha
mza
.
ha
na
f
i
@e
tu
.
ua
e
.
a
c
.
ma
1.
I
NT
RODU
C
T
I
ON
Dr
ug
dis
c
ove
r
y
p
lays
a
f
unda
menta
l
r
ole
in
the
he
a
lt
hc
a
r
e
s
e
c
tor
,
a
s
de
ve
lopi
ng
ne
w
c
ompounds
de
mands
a
mul
ti
dis
c
ipl
inar
y
a
ppr
oa
c
h
to
pr
ovide
nove
l
ther
a
pe
uti
c
int
e
r
ve
nti
ons
.
De
s
pit
e
thi
s
,
the
p
r
oc
e
s
s
is
of
ten
c
ompl
e
x,
ti
me
-
c
ons
umi
ng,
a
nd
r
e
qui
r
e
s
a
n
e
nor
m
ous
e
f
f
or
t
to
va
li
da
te
ne
w
tr
e
a
tm
e
nts
.
M
or
e
ove
r
,
tr
a
dit
ional
methods
of
dr
ug
dis
c
ove
r
y
a
r
e
not
only
r
e
s
our
c
e
-
int
e
ns
iv
e
but
a
ls
o
li
mi
ted
in
their
s
c
ope
[
1]
.
R
e
c
e
nt
a
dva
nc
e
ments
in
c
omput
a
ti
ona
l
biol
ogy
ha
ve
c
ompl
e
tely
tr
a
ns
f
or
med
dr
ug
dis
c
ove
r
y
pipelines
.
T
he
c
ombi
na
ti
on
o
f
biol
ogy
with
c
om
putational
methods
of
f
e
r
s
ne
w
ins
ight
s
to
a
c
c
e
ler
a
te
the
identif
ica
ti
on
a
nd
e
va
luation
o
f
nove
l
c
ompound
s
.
T
he
r
e
f
o
r
e
,
c
omput
a
ti
ona
l
tec
hniques
ha
ve
e
m
e
r
ge
d
a
s
powe
r
f
ul
tool
s
in
the
f
ield
of
pha
r
mac
ologi
c
a
l
medic
ine
[
2]
,
a
nd
r
e
ve
a
led
gr
e
a
t
s
uc
c
e
s
s
c
ompar
e
d
to
tr
a
dit
ional
methods
.
B
e
s
ides
,
thes
e
tec
hniques
h
a
ve
f
ound
wide
s
pr
e
a
d
a
ppli
c
a
ti
on
in
va
r
ious
he
a
lt
hc
a
r
e
domains
,
including
dis
e
a
s
e
c
las
s
if
ica
ti
on
[
3]
a
nd
s
ur
gica
l
e
nha
nc
e
ments
[
4]
.
Now
a
da
ys
,
a
lar
ge
a
mount
of
biol
ogica
l
da
ta
is
s
tor
e
d
in
publi
c
da
taba
s
e
s
a
nd
e
na
bles
r
e
s
e
a
r
c
he
r
s
to
e
xplor
e
a
wide
r
a
nge
o
f
methodologi
e
s
.
F
ur
ther
m
or
e
,
the
in
tegr
a
ti
on
a
nd
a
na
lys
is
o
f
thi
s
biol
ogica
l
da
ta
e
a
s
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14
,
No.
4
,
Augus
t
2025
:
281
5
-
2825
2816
the
s
tudy
of
ne
w
hypothes
e
s
[
5]
,
f
or
e
xa
mpl
e
,
p
r
e
dictive
modeling
us
ing
mac
hine
lea
r
n
ing
(
M
L
)
te
c
hniques
is
one
of
the
mos
t
e
xplo
r
e
d
methodologi
e
s
a
nd
h
a
s
ga
ined
pr
omi
ne
nc
e
.
M
L
models
c
a
n
e
f
f
e
c
ti
ve
l
y
c
las
s
if
y
dr
ugs
int
o
r
e
leva
nt
ther
a
pe
uti
c
c
a
tegor
ies
,
a
c
c
ur
a
tely
de
tec
t
a
nd
c
las
s
if
y
tum
or
s
tage
s
[
6]
,
a
nd
de
s
ign
ne
w
dr
ugs
ba
s
e
d
on
c
he
mi
c
a
l
pr
ope
r
ti
e
s
[
7]
.
C
ons
e
que
ntl
y,
M
L
-
ba
s
e
d
methods
a
r
e
c
a
pa
ble
in
de
tec
ti
ng
pa
tt
e
r
ns
a
nd
identif
ying
c
o
r
r
e
lations
withi
n
lar
ge
a
nd
c
omp
lex
da
tas
e
ts
with
numer
ous
va
r
iable
s
.
F
ur
ther
mor
e
,
bioi
n
f
or
matic
methods
ha
ve
be
e
n
c
r
uc
ial
in
the
d
r
ug
dis
c
ove
r
y
pipelines
,
a
ll
owing
r
e
s
e
a
r
c
he
r
s
to
s
tudy
mol
e
c
ules
f
r
om
a
s
ys
tem
-
leve
l
pe
r
s
pe
c
ti
ve
.
B
y
int
e
gr
a
ti
ng
knowle
dge
f
r
om
va
r
ious
domains
s
uc
h
a
s
ge
nomi
c
s
,
p
r
oteomics
,
t
r
a
ns
c
r
ipt
omi
c
s
,
pop
ulation
ge
ne
ti
c
s
,
a
nd
mol
e
c
ular
phylo
ge
ne
ti
c
s
,
bioi
nf
or
matic
a
na
lys
is
e
a
s
e
s
dr
ug
tar
ge
t
identif
ica
ti
on,
dr
ug
c
a
ndidate
s
c
r
e
e
ning,
pr
e
diction
of
dr
ug
r
e
s
is
tanc
e
,
a
nd
mi
nim
iza
ti
on
of
s
ide
e
f
f
e
c
ts
.
T
hus
,
M
L
a
lgor
it
hms
a
r
e
e
mpl
oye
d
a
longs
ide
b
ioi
nf
or
matics
to
pr
e
dict
int
e
r
a
c
ti
ons
a
mong
biol
ogica
l
e
nti
ti
e
s
[
8
]
a
nd
de
s
ign
c
us
tom
ize
d
dr
ugs
f
or
s
pe
c
if
ic
tr
e
a
tm
e
nts
,
ult
im
a
tely
a
dva
nc
ing
pr
e
c
is
ion
medic
ine.
How
e
ve
r
,
r
e
s
e
a
r
c
he
r
s
ha
ve
to
f
a
c
e
s
e
ve
r
a
l
c
ha
ll
e
nge
s
to
buil
d
M
L
models
in
dr
ug
dis
c
ove
r
y.
B
iol
ogica
l
da
ta
of
ten
va
r
ies
in
qua
li
ty
a
nd
a
lwa
ys
ne
e
ds
pr
e
pr
oc
e
s
s
ing
be
f
or
e
it
c
a
n
be
us
e
d
f
or
lea
r
ning
pur
pos
e
s
[
9]
.
Additi
ona
ll
y,
c
a
nc
e
r
c
las
s
if
ica
t
ion
pr
oblems
typi
c
a
ll
y
invol
ve
im
ba
lanc
e
d
da
tas
e
ts
,
c
ha
r
a
c
ter
ize
d
by
both
e
xc
e
s
s
ive
nois
e
a
nd
a
de
f
icie
nc
y
of
labe
led
da
ta
,
whic
h
s
igni
f
ica
ntl
y
a
f
f
e
c
ts
the
lea
r
ning
pr
oc
e
s
s
.
E
va
luating
the
e
f
f
ica
c
y
o
f
M
L
models
in
s
uc
h
s
c
e
na
r
ios
be
c
omes
c
ompl
e
x,
pa
r
ti
c
ular
ly
whe
n
c
onf
r
onted
with
li
mi
ted
or
b
ias
e
d
da
ta
[
10]
.
Our
c
ontr
ibut
ion
a
im
s
to
de
ve
lop
a
M
L
-
ba
s
e
d
c
las
s
if
ier
c
a
pa
ble
of
pr
e
dicting
a
c
ti
ve
c
ompounds
that
c
a
n
tar
ge
t
non
-
s
mall
c
e
ll
lung
c
a
nc
e
r
(
N
S
C
L
C
)
.
F
i
r
s
t,
we
c
ur
a
te
d
a
da
tas
e
t
by
e
xtr
a
c
ti
ng
b
ioac
ti
vit
y
d
a
ta
f
r
om
C
hE
M
B
L
[
11]
da
taba
s
e
ba
s
e
d
on
p
r
oteins
e
xpr
e
s
s
e
d
in
NSC
L
C
.
S
e
c
ond,
mol
e
c
ular
de
s
c
r
ipt
or
s
of
the
s
e
lec
ted
c
ompounds
we
r
e
c
a
lcula
ted
a
nd
us
e
d
a
s
input
f
e
a
tur
e
f
or
s
e
ve
r
a
l
models
.
T
he
n,
numer
ous
M
L
models
we
r
e
f
e
d
with
thi
s
da
ta
a
nd
tr
a
ined
to
le
a
r
n
f
r
om
the
s
tr
uc
tur
e
a
nd
c
he
mi
c
a
l
c
ha
r
a
c
ter
is
ti
c
s
of
the
mol
e
c
ules
.
F
inally,
we
pe
r
f
o
r
med
a
c
ompar
a
ti
ve
a
na
lys
is
to
identif
y
the
opti
mal
model
.
T
he
r
e
s
t
of
the
pa
pe
r
is
o
r
ga
nize
d
a
s
f
oll
ows
:
s
e
c
ti
on
2
pr
ovides
a
n
ove
r
view
o
f
r
e
late
d
wo
r
ks
i
n
the
f
ield.
S
e
c
ti
on
3
p
r
e
s
e
nts
our
a
pp
r
oa
c
h,
including
da
ta
c
oll
e
c
ti
on
a
nd
the
methodology
e
mpl
oye
d.
S
e
c
ti
on
4
dis
c
us
s
e
s
the
r
e
s
ult
s
obtaine
d
f
r
om
ou
r
a
na
lys
is
,
f
o
ll
owe
d
by
the
c
onc
lus
ion
in
s
e
c
ti
on
5
.
2.
RE
L
AT
E
D
WORK
Now
a
da
ys
,
many
s
tudi
e
s
ha
ve
be
e
n
c
onduc
ted
to
e
xplor
e
a
nd
unde
r
s
tand
the
biol
ogica
l
a
s
pe
c
ts
of
c
a
nc
e
r
c
e
ll
s
us
ing
M
L
models
.
I
n
pa
r
ti
c
ular
,
thes
e
s
tudi
e
s
a
im
to
be
tt
e
r
e
xplain
the
mec
ha
nis
ms
of
dif
f
e
r
e
nt
s
ignaling
pa
thwa
ys
that
tr
a
ns
mi
t
s
ignal
s
withi
n
c
e
l
l
s
a
nd
a
f
f
e
c
t
ge
ne
s
r
e
gulation.
P
r
oteins
s
uc
h
a
s
R
a
s
play
a
n
im
por
tant
r
ole
in
r
e
gulating
va
r
ious
biom
olec
ular
int
e
r
a
c
ti
ons
in
the
c
e
ll
’
s
li
f
e
c
yc
le
[
12]
.
T
he
R
a
s
pa
thwa
ys
tr
a
ns
mi
t
s
ignals
to
a
c
ti
va
te
ge
ne
s
that
pr
omot
e
c
e
l
l
gr
owth
a
nd
divi
s
ion
.
M
ut
a
ti
ons
in
ge
ne
s
a
s
s
oc
ia
ted
with
thes
e
pa
thwa
ys
c
a
n
lea
d
to
dif
f
e
r
e
nt
types
of
c
a
nc
e
r
s
[
13]
,
[
14]
.
T
he
r
e
f
or
e
,
ther
e
is
a
gr
owing
i
nter
e
s
t
in
identif
ying
ne
w
a
nti
-
R
a
s
ther
a
pe
uti
c
s
tr
a
tegie
s
.
I
n
a
s
tudy
c
onduc
ted
by
W
a
y
e
t
a
l
.
[
15
]
,
th
r
e
e
types
of
b
iol
ogica
l
da
ta
we
r
e
e
xplor
e
d:
ge
ne
e
xpr
e
s
s
ions
,
mut
a
ti
on
c
ounts
,
a
nd
mut
a
ti
on
c
opie
s
f
ound
in
va
r
ious
types
o
f
c
a
nc
e
r
s
us
ing
M
L
me
thods
to
pr
e
dict
the
a
c
ti
va
ti
on
of
the
R
a
s
pa
thwa
ys
.
T
he
a
uthor
s
of
thi
s
s
tudy
we
r
e
a
ble
to
de
s
ign
a
model
c
a
pa
ble
of
pr
e
dicting
R
NA
s
e
que
nc
e
s
that
a
c
ti
va
te
the
R
a
s
pa
thwa
ys
.
S
im
il
a
r
ly,
Knijnenbur
g
e
t
al
.
[
16
]
e
mpl
oye
d
ge
nomi
c
a
nd
mol
e
c
ular
da
ta
to
p
r
e
dict
the
a
c
ti
va
ti
on
of
p53
pa
thwa
ys
.
T
he
ge
ne
T
P
53
c
ontains
ins
tr
uc
ti
ons
f
or
r
e
gulating
a
p
r
otein
c
a
ll
e
d
p53
,
whic
h
f
unc
ti
o
ns
a
s
a
tum
or
s
uppr
e
s
s
or
a
nd
int
e
r
a
c
ts
with
the
a
poptos
is
mec
ha
nis
m
[
17]
.
C
ons
e
que
ntl
y,
mut
a
ti
ons
in
the
g
e
ne
T
P
53
c
a
n
lea
d
to
meta
s
tatic
c
a
nc
e
r
[
18
]
.
S
ome
meta
s
tatic
c
a
nc
e
r
s
a
r
e
a
s
s
oc
iate
d
with
the
los
s
of
phe
notypi
c
tr
a
it
s
e
xpr
e
s
s
e
d
by
s
tem
c
e
ll
s
[
19]
.
I
n
th
is
c
ontext,
to
e
lucida
te
the
r
e
lations
hip
be
twe
e
n
tu
mor
dif
f
e
r
e
nti
a
ti
on
phe
notype
a
n
d
tum
or
pr
opa
ga
ti
on
or
ge
ne
ti
c
a
lt
e
r
a
ti
ons
,
M
a
lt
a
e
t
al
.
[
20]
int
r
oduc
e
d
a
n
M
L
model
a
im
e
d
a
t
pr
e
dictin
g
c
a
nc
e
r
de
ve
lopm
e
nt
withi
n
s
pe
c
if
ic
c
e
ll
ula
r
ti
s
s
ue
s
.
T
he
y
r
e
li
e
d
on
da
ta
f
r
om
s
tem
c
e
ll
s
a
nd
their
pr
oge
nit
o
r
c
e
ll
s
to
c
ons
tr
uc
t
a
c
las
s
if
ier
f
or
ge
ne
ti
c
e
xpr
e
s
s
ion
tr
a
it
s
.
S
ubs
e
que
ntl
y,
they
a
ppli
e
d
thi
s
c
las
s
if
ier
to
a
c
e
ll
s
a
mpl
e
to
pr
e
dict
the
e
xpr
e
s
s
e
d
tr
a
it
s
.
T
he
y
we
r
e
a
ble
to
i
de
nti
f
y
c
a
nc
e
r
c
e
ll
s
withi
n
the
s
a
mpl
e
,
but
they
did
not
pr
ovide
de
tailed
inf
o
r
mation
a
bout
the
lea
r
ning
me
thodol
ogy
us
e
d
to
buil
d
the
c
las
s
if
ier
.
M
utations
in
the
e
pider
mal
gr
owth
f
a
c
tor
r
e
c
e
ptor
(
E
GFR
)
ha
ve
be
e
n
known
to
c
a
us
e
unc
ontr
oll
e
d
c
e
ll
pr
oli
f
e
r
a
ti
on
[
21]
.
Nume
r
ous
s
tud
ies
a
im
to
identif
y
s
mall
inhi
bit
or
y
mol
e
c
ules
that
tar
ge
t
th
e
E
GFR
ge
ne
.
Qur
e
s
hi
e
t
al
.
[
22
]
p
r
opos
e
s
a
pe
r
s
ona
li
z
e
d
model
f
or
pr
e
dicting
dr
ug
r
e
s
pons
e
in
lung
c
a
nc
e
r
pa
ti
e
nts
.
S
pe
c
if
ica
ll
y,
thi
s
model
wa
s
te
s
ted
to
p
r
e
dict
th
e
r
e
s
pons
e
to
US
F
ood
a
nd
Dr
ug
Admini
s
tr
a
ti
o
n
(
F
DA
)
-
a
ppr
ove
d
s
mall
mol
e
c
ules
,
s
uc
h
a
s
E
r
lot
ini
b
a
nd
Ge
f
it
ini
b.
T
o
c
ons
tr
uc
t
their
model,
the
a
uthor
s
a
s
s
e
mbl
e
d
va
r
ious
types
of
da
ta:
E
GFR
mut
a
ti
ons
f
ound
in
l
ung
c
a
nc
e
r
pa
ti
e
nts
,
c
li
nica
l
d
a
ta
including
pa
ti
e
nt
s
ur
vival
a
nd
c
li
nica
l
r
e
s
pons
e
to
dr
ugs
,
de
mogr
a
phic
da
ta
s
uc
h
a
s
a
ge
,
s
e
x,
a
nd
s
moki
ng
his
tor
y,
a
nd
the
3D
s
tr
uc
tur
e
of
E
GFR
ge
ne
mut
a
ti
ons
f
ound
in
pa
ti
e
nts
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
N
on
-
s
mall
c
e
ll
lung
c
anc
e
r
ac
ti
v
e
c
ompounds
dis
c
ov
e
r
y
holdi
ng
on
pr
otein
e
x
pr
e
s
s
ion
…
(
Ham
z
a
Hanafi)
2817
A
de
c
is
ion
tr
e
e
-
ba
s
e
d
c
las
s
if
ier
wa
s
tr
a
ined
us
in
g
thi
s
da
ta
to
pr
e
dict
the
leve
l
of
dr
ug
r
e
s
pons
e
a
mong
f
our
c
a
tegor
ies
:
no
r
e
s
pons
e
,
pa
r
ti
a
l
r
e
s
po
ns
e
,
moder
a
te
r
e
s
pons
e
,
a
nd
s
tr
ong
r
e
s
pons
e
.
T
he
a
uthor
s
f
ound
that
de
mog
r
a
phic
da
ta
ha
d
a
we
a
k
im
pa
c
t
on
th
e
lea
r
ning
outcome
of
the
model
.
Only
E
GFR
m
utations
a
nd
s
tr
uc
tur
e
s
s
howe
d
a
good
pr
e
dictive
r
e
s
pons
e
leve
l
of
dr
ug
r
e
s
pons
e
.
T
he
a
uthor
s
did
no
t
f
ur
ther
us
e
thi
s
model
to
tes
t
the
r
e
s
pons
e
leve
l
of
mol
e
c
ules
that
we
r
e
not
us
e
d
dur
ing
the
lea
r
ning
pha
s
e
.
Y
a
ng
e
t
al.
[
23]
a
im
ing
to
de
ter
mi
ne
the
da
ta
that
c
a
n
e
s
tablis
h
a
n
M
L
model
to
p
r
e
dict
E
GFR
mut
a
ti
ons
in
lung
c
a
nc
e
r
,
the
a
uthor
s
c
ompar
e
d
the
pe
r
f
o
r
manc
e
of
s
e
ve
r
a
l
lea
r
n
ing
a
lgor
it
hms
r
a
ndom
f
or
e
s
t
(
R
F
)
,
li
ght
g
r
a
dient
boos
ti
ng
mac
hine
(
L
ight
GB
M
)
,
s
uppor
t
ve
c
tor
mac
hine
(
S
VM
)
,
mul
ti
laye
r
pe
r
c
e
ptr
on
(
M
L
P
)
,
a
nd
e
xtr
e
me
gr
a
dient
boos
ti
n
g
(
XG
B
)
us
ing
mul
ti
p
le
c
li
nica
l
a
nd
de
mogr
a
phic
da
ta.
T
he
y
f
ound
that
tobac
c
o
c
ons
umpt
ion,
s
e
x,
c
holes
ter
ol,
a
ge
,
a
nd
the
a
lbum
in/
globul
in
r
a
ti
o
we
r
e
a
mong
the
top
f
ive
va
r
iable
s
r
e
late
d
t
o
E
GFR
mut
a
ti
on,
whic
h
dif
f
e
r
e
d
s
li
ghtl
y
f
r
om
the
r
e
s
ult
s
obtaine
d
in
the
s
tudy
[
22]
,
whe
r
e
the
i
mpac
t
of
de
mogr
a
phic
da
ta
wa
s
we
a
k.
W
idyana
nda
e
t
al
.
[
24
]
inves
ti
ga
te
the
potential
of
Que
r
c
e
ti
n,
a
na
tur
a
l
c
ompound
f
ound
in
f
r
uit
s
a
nd
ve
ge
table
s
,
to
c
ombat
gli
ob
las
tom
a
mul
ti
f
or
me.
B
y
e
xa
mi
ning
da
taba
s
e
s
l
ike
na
ti
ona
l
c
e
nter
f
or
biot
e
c
hnology
inf
or
mation
(
NC
B
I
)
,
s
upe
r
-
e
nha
nc
e
r
a
r
c
hive
(
S
E
A)
,
c
ompar
a
ti
ve
toxi
c
oge
nomi
c
s
da
taba
s
e
(
C
T
D)
,
a
nd
s
e
a
r
c
h
tool
f
or
the
r
e
tr
ieva
l
o
f
int
e
r
a
c
t
ing
ge
ne
s
/pr
oteins
(
S
T
R
I
NG
)
,
the
s
tudy
identif
ies
f
our
ke
y
pr
oteins
s
e
r
ine/thr
e
onine
kinas
e
1
(
AK
T
1
)
,
mat
r
ix
meta
ll
ope
pti
da
s
e
9
(
M
M
P
9)
,
AT
P
bindi
ng
c
a
s
s
e
tt
e
s
ubf
a
mi
ly
B
membe
r
1
(
AB
C
B
1
)
,
a
nd
va
s
c
ular
e
ndothelial
gr
owth
f
a
c
tor
A
(
VE
GF
A
)
,
that
Que
r
c
e
ti
n
dir
e
c
tl
y
a
f
f
e
c
ts
.
Us
ing
S
T
I
T
C
H
,
S
E
A,
a
nd
S
T
R
I
NG
,
th
e
s
tudy
c
ons
tr
uc
ts
pr
otein
-
pr
otein
int
e
r
a
c
ti
on
n
e
twor
ks
,
highl
ight
ing
c
onne
c
ti
ons
be
twe
e
n
thes
e
pr
oteins
.
F
unc
ti
ona
l
a
nnotation
a
na
lys
is
thr
ough
the
DA
VI
D
we
b
s
e
r
ve
r
c
lar
if
ies
the
biol
ogica
l
pr
oc
e
s
s
e
s
inf
luenc
e
d
by
thes
e
pr
oteins
.
M
olec
ular
doc
king
s
im
ulations
with
AutoDoc
k
Vina
[
25
]
pr
ovide
ins
ight
s
int
o
how
Que
r
c
e
ti
n
int
e
r
a
c
ts
with
thes
e
p
r
oteins
,
e
xtend
ing
our
unde
r
s
tanding
of
it
s
potential
a
s
a
gli
oblas
tom
a
mul
ti
f
o
r
me
t
r
e
a
tm
e
nt.
T
he
s
tudy
not
only
u
nc
ove
r
s
Que
r
c
e
ti
n’
s
im
pa
c
t
on
c
r
uc
ial
g
li
oblas
tom
a
mul
ti
f
or
me
r
e
late
d
pr
oteins
but
a
ls
o
e
mphas
ize
s
it
s
pote
nti
a
l
a
s
a
tar
ge
ted
ther
a
pe
uti
c
opt
ion
a
ga
ins
t
gli
oblas
tom
a
m
ult
if
or
me.
3.
M
E
T
HO
DOL
OG
Y
Qua
nti
tative
s
tr
uc
tur
e
-
a
c
ti
vit
y
r
e
lations
hip
(
QSA
R
)
modeling
leve
r
a
ge
s
the
r
e
lations
hips
be
twe
e
n
the
c
he
mi
c
a
l
s
tr
uc
tur
e
a
nd
the
biol
ogica
l
a
c
ti
vi
ty
of
mol
e
c
ules
[
26]
.
QSAR
models
e
mpl
oy
m
olec
ular
de
s
c
r
ipt
or
s
,
whic
h
c
a
ptur
e
the
phys
ica
l
a
nd
c
he
mi
c
a
l
pr
ope
r
t
ies
dis
ti
nguis
hing
one
mol
e
c
ule
f
r
o
m
a
nother
[
27]
.
T
he
s
e
models
p
r
ovide
va
luable
ins
ight
s
int
o
t
he
c
he
mi
c
a
l
pr
ope
r
ti
e
s
that
a
r
e
c
r
uc
ial
f
or
the
inhi
bit
ion
of
s
pe
c
if
ic
biol
ogica
l
pr
oc
e
s
s
e
s
.
T
hus
,
a
idi
ng
biol
og
is
ts
a
nd
c
he
mi
s
ts
in
the
de
s
ign
of
r
obus
t
mol
e
c
u
les
with
opti
mi
z
e
d
pr
ope
r
ti
e
s
.
Util
izing
M
L
-
ba
s
e
d
QSAR
a
na
lys
is
a
nd
mol
e
c
ular
doc
king,
I
r
e
s
ha
e
t
al
.
[
28]
e
xplor
e
s
medic
inal
plant
c
ompounds
a
s
inhi
bit
or
s
f
or
H
I
V
-
1
r
e
ve
r
s
e
tr
a
ns
c
r
ipt
a
s
e
,
a
ddr
e
s
s
ing
r
e
s
i
s
tanc
e
is
s
ue
s
.
S
im
il
a
r
ly,
our
s
tudy
a
im
s
to
us
e
M
L
m
ode
ls
to
pr
e
dict
a
c
ti
ve
c
ompounds
ba
s
e
d
on
the
ge
ne
s
e
xpr
e
s
s
e
d
in
NSC
L
C
thr
ough
QSAR
a
na
lys
is
.
T
o
c
ons
tr
uc
t
ou
r
da
tas
e
t,
F
ir
s
t,
we
s
e
lec
ted
a
s
e
t
of
ge
ne
s
that
ha
ve
be
e
n
e
xtens
ively
a
s
s
oc
iate
d
with
NSC
L
C
in
va
r
ious
s
tudi
e
s
[
6]
,
[
7]
.
Af
ter
ga
ther
ing
the
t
a
r
ge
t
pr
oteins
r
e
late
d
to
thes
e
ge
ne
s
f
r
om
the
C
hE
M
B
L
da
taba
s
e
,
we
s
e
lec
ted
their
bioac
ti
vit
ies
a
nd
c
omput
e
d
their
mol
e
c
ular
de
s
c
r
ipt
or
s
to
a
na
lyze
the
c
he
mi
c
a
l
s
tr
uc
tur
e
a
nd
identif
y
pa
tt
e
r
ns
in
a
c
ti
ve
c
ompounds
.
Af
ter
wa
r
ds
,
we
tr
a
ined
s
e
ve
r
a
l
models
us
i
ng
the
c
ons
tr
uc
ted
de
s
c
r
ipt
or
s
a
nd
e
va
luate
d
their
pe
r
f
or
manc
e
ba
s
e
d
on
the
c
onf
us
ion
ma
tr
ix
a
nd
the
a
c
hieve
d
F
1
s
c
or
e
.
T
he
s
e
e
va
luation
metr
ics
pr
ovide
a
c
ompr
e
he
ns
ive
a
s
s
e
s
s
ment
of
the
models
’
pr
e
dictive
a
bil
it
ies
.
F
igur
e
1
il
lus
tr
a
tes
ou
r
p
r
opos
e
d
meth
odology,
a
nd
the
e
xpe
r
im
e
ntal
pr
oc
e
dur
e
e
s
tablis
he
d.
T
he
ter
m
“
tar
ge
ts
”
in
the
C
h
E
M
B
L
da
taba
s
e
r
e
f
e
r
s
to
p
r
oteins
o
r
or
ga
nis
ms
that
c
ompounds
a
c
t
upon.
B
iol
ogica
ll
y,
thes
e
c
ompounds
e
nga
ge
in
int
e
r
a
c
t
ions
with
the
tar
ge
ted
pr
o
teins
,
r
e
s
ult
ing
in
a
mo
dulato
r
y
a
c
ti
vit
y.
S
uc
h
a
c
ti
vi
ty
may
e
nc
ompas
s
the
a
c
ti
va
ti
on
or
inh
ibi
ti
on
of
the
tar
ge
ted
pr
otein
.
T
h
e
ove
r
a
ll
a
ppr
oa
c
h
is
f
oll
owe
d
to
pr
e
dict
c
ompounds
’
a
c
ti
vit
y
to
tar
ge
t
NSC
L
C
.
I
n
s
tep
1
the
e
xpr
e
s
s
e
d
ge
ne
s
a
r
e
identif
ied
f
r
om
the
medic
a
l
li
ter
a
tur
e
[
29]
,
thi
s
include
s
8
e
xpr
e
s
s
e
d
ge
ne
s
:
b
-
r
a
f
pr
oto
-
onc
oge
ne
,
s
e
r
ine/thr
e
onine
kinas
e
(
B
R
AF)
,
E
GFR
,
kir
s
ten
r
a
t
s
a
r
c
om
a
vir
us
–
pr
oto
-
onc
oge
ne
,
GT
P
a
s
e
(
KR
AS)
,
phos
pha
tas
e
a
nd
tens
in
homol
og
(
P
T
E
N)
,
r
e
c
e
ptor
ty
r
os
ine
kinas
e
(
R
OS1)
,
v
-
e
r
b
-
b2
a
vian
e
r
ythr
oblas
ti
c
leuke
mi
a
vir
a
l
onc
oge
ne
homol
og
2,
a
ls
o
known
a
s
HE
R
2
a
nd
ne
u
(
E
R
B
B
2)
,
M
E
T
pr
o
to
-
onc
oge
ne
,
r
e
c
e
ptor
tyr
o
s
ine
kinas
e
(
M
E
T
)
,
a
nd
a
na
plas
ti
c
lym
phoma
kinas
e
(
AL
K)
.
I
n
s
tep
2,
b
ioac
ti
vit
y
da
ta
of
the
tar
ge
t
pr
o
tein
is
e
xtr
a
c
ted
f
r
o
m
C
hE
M
B
L
da
taba
s
e
.
I
n
s
tep
3
,
m
olec
ular
de
s
c
r
ipt
or
s
of
the
bioac
ti
vit
y
da
ta
a
r
e
c
a
lcula
ted.
I
n
s
tep
4,
s
e
ve
r
a
l
M
L
models
a
r
e
tr
a
ined
on
thes
e
mol
e
c
ular
de
s
c
r
ipt
or
s
.
F
inally,
in
s
tep
5,
the
models
a
r
e
e
va
luate
d
to
a
s
s
e
s
s
their
pr
e
dictive
a
c
c
ur
a
c
y
f
or
de
te
r
mi
ning
c
ompound
a
c
ti
vit
y
to
tar
ge
t
NSC
L
C
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14
,
No.
4
,
Augus
t
2025
:
281
5
-
2825
2818
F
igur
e
1.
P
r
opos
e
d
methodology
f
o
r
p
r
e
dicting
c
o
mpound
a
c
ti
vit
y
to
tar
ge
t
N
S
C
L
C
us
ing
b
ioac
ti
vit
y
da
ta,
mol
e
c
ular
de
s
c
r
ipt
or
s
,
a
nd
ML
models
3.
1.
M
olec
u
lar
d
e
s
c
r
ip
t
or
s
M
olec
ular
de
s
c
r
ipt
or
s
a
r
e
numer
ica
l
r
e
pr
e
s
e
ntat
ions
that
c
a
ptur
e
va
r
ious
phys
icoc
he
mi
c
a
l
a
nd
topol
ogica
l
pr
ope
r
ti
e
s
of
mol
e
c
ules
.
T
he
s
e
de
s
c
r
i
ptor
s
pr
ovide
va
luable
qua
nti
t
a
ti
ve
inf
or
mation
a
bout
the
c
ha
r
a
c
ter
is
ti
c
s
a
nd
be
ha
vior
of
c
he
mi
c
a
l
c
ompounds
.
B
y
a
na
lyzing
a
nd
c
ompar
ing
mol
e
c
ular
de
s
c
r
ipt
or
s
,
r
e
s
e
a
r
c
he
r
s
c
a
n
ga
in
ins
ight
s
int
o
the
s
tr
uc
tur
e
-
a
c
ti
vit
y
r
e
lations
hips
of
mol
e
c
ules
a
nd
make
pr
e
dictions
a
bout
their
pr
ope
r
t
i
e
s
,
r
e
a
c
ti
vit
y
,
a
nd
po
tential
biol
ogica
l
a
c
ti
vit
ies
.
One
c
omm
only
us
e
d
tool
f
o
r
c
omput
ing
mo
lec
ular
de
s
c
r
ipt
or
s
is
P
a
DE
L
-
de
s
c
r
ipt
or
[
30]
.
I
t
is
a
s
of
twa
r
e
pr
ogr
a
m
that
c
a
lcula
tes
a
c
ompr
e
he
ns
ive
s
e
t
of
mol
e
c
ular
de
s
c
r
i
ptor
s
ba
s
e
d
on
s
im
pli
f
ied
mol
e
c
ular
input
li
ne
e
ntr
y
s
ys
tem
(
S
M
I
L
E
S
)
notat
ions
[
31]
.
S
M
I
L
E
S
is
a
c
ompac
t
s
tr
ing
r
e
pr
e
s
e
ntation
of
a
mol
e
c
ule’
s
s
tr
uc
tur
e
,
whic
h
e
nc
ode
s
ke
y
s
tr
uc
tur
a
l
f
e
a
tur
e
s
including
a
tom
types
,
bond
c
onne
c
ti
ons
,
a
nd
their
s
pa
ti
a
l
a
r
r
a
nge
ment
wi
thi
n
the
mol
e
c
ule.
T
he
P
a
DE
L
-
de
s
c
r
ipt
or
uti
l
ize
s
a
lgor
it
hms
a
nd
mathe
matica
l
f
or
mul
a
s
to
ge
ne
r
a
te
a
wide
r
a
nge
of
de
s
c
r
ipt
or
s
,
including
c
ons
ti
tut
ional,
topol
ogica
l
,
a
nd
phys
ico
c
he
mi
c
a
l
de
s
c
r
ipt
or
s
.
C
ons
ti
tut
ional
de
s
c
r
ipt
or
s
c
a
ptur
e
ba
s
ic
mol
e
c
ula
r
f
e
a
tur
e
s
,
s
uc
h
a
s
the
number
of
a
tom
s
,
bonds
,
a
nd
f
unc
ti
ona
l
gr
oups
.
T
opologi
c
a
l
de
s
c
r
ipt
or
s
a
s
s
e
s
s
mol
e
c
ular
c
onne
c
ti
vit
y
a
nd
s
ha
pe
,
p
r
ovidi
ng
inf
or
mation
a
bout
the
a
r
r
a
nge
ment
of
a
tom
s
a
nd
t
he
pr
e
s
e
nc
e
of
s
pe
c
if
ic
s
tr
uc
tur
a
l
mot
if
s
.
P
hys
icoc
he
mi
c
a
l
de
s
c
r
ipt
or
s
qua
nti
f
y
pr
ope
r
ti
e
s
s
uc
h
a
s
mol
e
c
ular
we
ight
,
s
olubi
li
ty
,
li
pophil
icity
,
hydr
oge
n
bonding
potential,
a
nd
e
lec
tr
onic
pr
ope
r
ti
e
s
.
I
n
our
s
tud
ying,
we
us
e
d
the
de
s
c
r
ipt
or
s
de
f
ined
by
the
P
ubC
he
m
da
taba
s
e
[
32]
,
whic
h
pr
im
a
r
il
y
f
oc
us
on
the
s
t
r
uc
tur
a
l
a
nd
phys
icoc
he
mi
c
a
l
pr
ope
r
ti
e
s
of
c
ompounds
.
T
he
s
e
de
s
c
r
ipt
or
s
a
r
e
typi
c
a
ll
y
e
nc
ode
d
in
a
byt
e
a
r
r
a
y.
T
a
ble
1
p
r
ovides
a
de
s
c
r
ipt
ion
o
f
the
P
ubC
he
m
de
s
c
r
ipt
or
s
bytes
.
3.
2.
L
e
ar
n
in
g
t
as
k
s
T
o
c
o
ns
t
r
uc
t
o
u
r
d
a
ta
s
e
t
,
8
4
.
07
8
mo
l
e
c
u
l
e
s
w
e
r
e
s
e
le
c
t
e
d
f
r
o
m
C
h
E
M
B
L
da
t
a
b
a
s
e
a
n
d
c
la
s
s
i
f
ie
d
t
he
m
i
n
t
o
a
c
t
i
v
e
a
n
d
i
na
c
t
iv
e
s
e
ts
b
a
s
e
d
o
n
t
h
e
i
r
i
nh
i
b
i
t
io
n
c
o
nc
e
n
t
r
a
t
i
on
v
a
l
ue
a
t
5
0
%
(
I
C
5
0
)
.
I
n
i
t
i
a
l
ly
,
we
u
s
e
d
a
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
N
on
-
s
mall
c
e
ll
lung
c
anc
e
r
ac
ti
v
e
c
ompounds
dis
c
ov
e
r
y
holdi
ng
on
pr
otein
e
x
pr
e
s
s
ion
…
(
Ham
z
a
Hanafi)
2819
t
h
r
e
s
h
o
l
d
o
f
10
0
n
M
,
a
n
d
a
f
te
r
i
t
e
r
a
ti
v
e
t
r
a
i
n
i
ng
o
f
t
h
e
M
L
m
o
de
ls
w
i
t
h
va
r
y
i
n
g
th
r
e
s
h
o
l
d
s
,
we
o
b
s
e
r
v
e
d
t
ha
t
l
o
w
e
r
i
ng
t
h
e
I
C
50
t
h
r
e
s
h
o
ld
i
m
p
r
o
v
e
d
t
h
e
m
o
de
ls
’
p
r
e
d
i
c
t
iv
e
p
e
r
f
o
r
ma
nc
e
.
F
u
r
t
h
e
r
e
x
p
e
r
i
me
n
ta
t
i
o
n
r
e
v
e
a
l
e
d
t
ha
t
s
e
t
ti
n
g
t
h
e
t
h
r
e
s
h
o
l
d
a
t
7
7
nM
y
i
e
ld
e
d
t
he
h
i
g
h
e
s
t
F
1
s
c
o
r
e
,
i
n
d
ic
a
t
in
g
o
p
t
i
ma
l
m
o
de
l
p
e
r
f
o
r
m
a
n
c
e
t
h
a
t
a
ll
o
w
s
t
he
m
o
d
e
l
t
o
e
f
f
e
c
t
iv
e
l
y
e
x
p
l
o
r
e
t
h
e
s
t
r
u
c
tu
r
e
o
f
c
o
m
pou
n
d
s
t
o
de
t
e
r
mi
n
e
i
ts
a
c
ti
v
i
t
y
.
T
h
us
,
c
h
o
os
i
ng
t
h
i
s
th
r
e
s
h
o
l
d
w
a
s
c
r
u
c
i
a
l
t
o
e
n
ha
n
c
e
t
h
e
p
r
e
d
ic
t
i
ve
c
a
p
a
bi
l
i
t
y
o
f
t
he
M
L
a
l
go
r
i
t
h
ms
us
e
d
i
n
t
h
is
s
t
u
d
y
.
C
o
ns
e
qu
e
nt
l
y
,
m
o
l
e
c
u
l
e
s
w
i
t
h
a
n
I
C
5
0
v
a
lu
e
l
o
we
r
t
h
a
n
o
r
e
q
u
a
l
t
o
7
7
n
M
w
e
r
e
c
o
ns
id
e
r
e
d
a
c
t
i
ve
,
d
e
n
o
te
d
b
y
a
n
a
s
s
i
gn
e
d
a
c
t
iv
i
t
y
l
e
v
e
l
o
f
1
,
w
h
i
l
e
t
h
os
e
g
r
e
a
te
r
t
h
a
n
7
7
nM
w
e
r
e
c
o
ns
i
de
r
e
d
i
n
a
c
ti
v
e
,
r
e
p
r
e
s
e
n
te
d
b
y
a
n
a
c
ti
v
i
t
y
le
v
e
l
o
f
0
.
T
a
ble
1.
S
umm
a
r
y
de
s
c
r
ipt
ion
of
P
ubC
he
m
de
s
c
r
i
ptor
s
P
ubC
he
m bi
t
pos
it
io
n r
a
nge
D
e
s
c
r
ip
ti
on
F
r
om 0 t
o 114
T
he
s
e
bi
na
r
y
uni
ts
e
xa
mi
ne
th
e
pr
e
s
e
n
c
e
or
a
bunda
nc
e
of
s
pe
c
if
ic
c
he
mi
c
a
l
a
to
ms
.
F
r
om 115 t
o 262
T
he
s
e
bi
n
a
r
y unit
s
a
s
s
e
s
s
t
he
pr
e
s
e
n
c
e
of
c
yc
li
c
s
tr
uc
tu
r
e
s
.
F
r
om 263 t
o 326
T
h
e
s
e
b
i
n
a
r
y
u
n
i
t
s
e
x
a
m
i
n
e
t
h
e
p
r
e
s
e
n
c
e
o
f
c
o
n
n
e
c
t
e
d
p
a
i
r
s
o
f
a
t
o
m
s
,
d
i
s
r
e
g
a
r
d
i
n
g
t
h
e
i
r
q
u
a
n
t
i
t
y
a
n
d
a
r
r
a
n
g
e
m
e
n
t
.
F
r
om 327 t
o 448
T
he
s
e
bi
na
r
y
uni
ts
a
s
s
e
s
s
th
e
pr
e
s
e
nc
e
of
a
to
m
n
e
a
r
e
s
t
n
e
ig
hbor
pa
tt
e
r
ns
,
c
ons
id
e
r
in
g t
he
r
e
le
va
nc
e
of
a
r
oma
ti
c
it
y a
nd
s
ig
ni
f
ic
a
nt
bonding.
F
r
om 445 t
o 459
T
he
s
e
bi
na
r
y
uni
ts
e
xa
mi
n
e
c
ompl
e
x
a
to
m
n
e
ig
hbor
hood
pa
tt
e
r
ns
,
ir
r
e
s
pe
c
ti
ve
of
t
he
ir
qua
nt
it
y, w
it
h s
pe
c
if
ic
c
ons
id
e
r
a
ti
on give
n t
o bond or
de
r
s
.
F
r
om 460 t
o 712
T
he
s
e
bi
n
a
r
y
uni
ts
e
va
lu
a
te
th
e
pr
e
s
e
nc
e
of
s
tr
a
ig
ht
f
or
w
a
r
d
S
M
I
L
E
S
a
r
bi
t
r
a
r
y
ta
r
ge
t
s
pe
c
if
ic
a
ti
on
(
S
M
A
R
T
S
)
pa
tt
e
r
ns
,
w
it
hout
c
ons
id
e
r
in
g
t
he
ir
qua
nt
it
y,
bu
t
w
it
h
s
pe
c
if
ic
a
tt
e
nt
io
n
gi
ve
n
to
bond
or
de
r
s
a
nd
th
e
c
omp
a
ti
bi
li
ty
of
bond
a
r
oma
ti
c
it
y w
it
h both s
in
gl
e
a
nd double
bonds
.
F
r
om 713 t
o 880
T
he
s
e
bi
n
a
r
y
uni
ts
e
xa
mi
ne
th
e
pr
e
s
e
nc
e
of
c
ompl
e
x
S
M
A
R
T
S
pa
tt
e
r
n
s
,
ir
r
e
s
pe
c
ti
ve
of
th
e
ir
qua
nt
it
y,
w
it
h
pa
r
ti
c
ul
a
r
e
mpha
s
is
on
s
pe
c
if
ic
bond
or
de
r
s
a
nd bond a
r
oma
ti
c
it
y.
T
o
tr
a
in
our
models
,
90
%
o
f
the
da
ta
wa
s
a
ll
oc
a
ted
to
the
tr
a
ini
ng
s
e
t,
while
the
r
e
maining
10%
wa
s
us
e
d
f
or
the
tes
t
s
e
t.
M
ol
e
c
ular
de
s
c
r
ipt
or
s
a
r
e
goi
ng
to
s
e
r
ve
a
s
input
f
e
a
tu
r
e
s
,
while
the
tar
ge
t
f
e
a
tu
r
e
is
the
a
c
ti
vit
y
leve
l
in
NSC
L
C
.
Give
n
the
c
ompl
e
xit
y
int
r
oduc
e
d
by
thi
s
e
xtens
ive
a
r
r
a
y
of
input
f
e
a
t
ur
e
s
,
we
im
pleme
nted
a
pr
e
pr
oc
e
s
s
ing
s
tep
to
r
e
f
ine
the
da
tas
e
t,
to
make
the
M
L
models
mor
e
p
r
e
c
is
e
a
nd
to
de
pict
the
pa
tt
e
r
ns
of
the
mos
t
im
por
tant
mol
e
c
ular
de
s
c
r
ipt
or
s
.
I
n
thi
s
r
e
ga
r
d
,
be
f
o
r
e
f
e
e
ding
the
da
ta
int
o
the
M
L
models
,
we
pe
r
f
o
r
med
a
n
ini
ti
a
l
s
tep
to
r
e
duc
e
the
number
of
input
f
e
a
tur
e
s
.
I
ni
ti
a
ll
y,
ther
e
we
r
e
881
f
e
a
tur
e
s
;
by
a
pplyi
ng
a
va
r
ianc
e
thr
e
s
hold
of
0
.
16,
we
r
e
moved
mol
e
c
ular
de
s
c
r
ipt
or
s
with
low
va
r
ianc
e
,
whic
h
s
howe
d
84%
s
im
il
a
r
it
y
in
their
va
lues
.
T
his
r
e
s
ult
e
d
in
a
f
inal
s
e
t
o
f
160
f
e
a
tur
e
s
with
higher
va
r
ianc
e
,
e
na
bli
ng
the
model
to
de
tec
t
mea
ningf
ul
pa
tt
e
r
ns
withi
n
the
da
tas
e
t.
W
e
tr
a
ined
s
e
ve
r
a
l
M
L
models
on
the
c
omput
e
d
mol
e
c
ular
de
s
c
r
ipt
or
s
,
including
M
L
P
,
XG
B
,
R
F
,
S
VM
,
a
nd
n
a
ive
B
a
ye
s
(
NB
)
.
F
ir
s
tl
y
,
the
M
L
P
ne
ur
a
l
ne
twor
k
model
is
us
e
d
with
a
n
input
laye
r
of
100
unit
s
a
nd
us
e
s
the
r
e
c
ti
f
ied
li
ne
a
r
unit
(
R
e
L
U)
a
s
a
n
a
c
ti
va
ti
on
f
unc
ti
on.
T
his
is
f
oll
owe
d
by
s
ome
hidden
laye
r
s
with
50,
20
,
a
nd
5
unit
s
,
r
e
s
pe
c
ti
ve
ly,
a
ls
o
us
ing
the
R
e
L
U
a
c
ti
va
ti
on
f
unc
ti
on
.
And
the
output
lay
e
r
ha
s
a
s
ingl
e
unit
with
a
s
igm
oid
a
c
ti
va
ti
on
f
unc
ti
on,
a
ll
o
wing
f
o
r
bi
na
r
y
c
las
s
if
ica
ti
on.
Dur
ing
tr
a
ini
ng
,
we
uti
li
z
e
d
the
Ada
m
opti
mi
z
e
r
with
a
lea
r
ning
r
a
te
o
f
0.
001
a
nd
e
mpl
oye
d
the
binar
y
c
r
os
s
-
e
ntr
opy
a
s
a
los
s
f
unc
ti
on.
T
he
model
wa
s
t
r
a
ined
f
o
r
100
e
poc
hs
with
a
ba
t
c
h
s
ize
of
100
s
a
mpl
e
s
.
F
ur
ther
mo
r
e
,
the
t
r
a
ini
ng
da
ta
wa
s
s
pli
t
int
o
t
r
a
ini
ng
a
nd
va
li
da
ti
on
s
ubs
e
ts
,
with
a
5
%
va
li
da
ti
on
s
pli
t
.
W
e
a
ls
o
tr
a
ined
two
tr
e
e
-
ba
s
e
d
c
las
s
if
ier
s
,
including
XG
B
a
nd
R
F
models
.
T
he
s
e
models
,
c
ons
tr
uc
ted
us
ing
the
s
klea
r
n
im
p
leme
ntation,
a
r
e
ba
s
e
d
on
e
ns
e
mbl
e
lea
r
ning
tec
hniques
th
a
t
c
ombi
ne
s
mul
ti
ple
de
c
is
ion
tr
e
e
s
to
make
p
r
e
dictions
.
T
o
c
o
ns
tr
uc
t
the
models
,
we
us
e
d
va
r
ious
pa
r
a
mete
r
s
to
opti
mi
z
e
their
pe
r
f
o
r
manc
e
.
W
e
us
e
d
a
max
de
pth
of
2
leve
ls
,
f
o
r
e
a
c
h
c
ons
tr
uc
ted
indi
v
idual
de
c
is
ion
tr
e
e
.
Additi
ona
ll
y,
we
uti
li
z
e
d
a
lea
r
ning
r
a
te
of
0
.
01
,
to
c
ontr
ol
the
s
tep
s
ize
a
t
e
a
c
h
boos
ti
ng
a
nd
ba
gging
it
e
r
a
ti
on.
T
he
number
of
e
s
ti
mator
s
wa
s
s
e
t
to
1
,
0
00,
indi
c
a
ti
ng
the
number
of
de
c
is
ion
tr
e
e
s
to
be
c
r
e
a
ted
in
the
e
ns
e
mbl
e
.
T
his
va
lue
wa
s
c
hos
e
n
f
r
om
a
li
s
t
t
ha
t
include
d
50,
200,
400,
600
,
800
,
a
nd
1
,
000
e
s
ti
mator
s
.
T
he
r
e
c
e
iver
ope
r
a
ti
ng
c
ha
r
a
c
ter
is
ti
c
(
R
OC
)
c
u
r
ve
s
we
r
e
s
ubs
e
que
ntl
y
dr
a
wn
f
or
e
a
c
h
model
tr
a
ined
with
a
dis
ti
nc
t
number
of
e
s
ti
mator
s
to
a
s
s
e
s
s
their
im
pa
c
t
on
the
model’
s
pe
r
f
o
r
manc
e
.
Nota
bl
y,
a
s
the
number
of
e
s
ti
mator
s
incr
e
a
s
e
d
f
r
om
50
to
1
,
000
,
we
obs
e
r
ve
d
a
pr
ogr
e
s
s
ive
im
pr
ove
ment
of
the
a
r
e
a
unde
r
the
c
ur
ve
(
AU
C
)
.
T
h
e
a
c
hi
e
v
e
d
AU
C
v
a
l
ue
s
s
to
o
d
a
t
0
.
6
21
f
o
r
XG
B
a
nd
0
.
59
6
f
or
R
F
a
s
s
h
ow
n
in
F
i
gu
r
e
s
2
a
nd
3
r
e
s
pe
c
ti
ve
ly
.
W
e
a
ls
o
a
d
jus
ted
the
c
l
a
s
s
we
ig
hts
t
o
a
d
d
r
e
s
s
c
l
a
s
s
i
mba
la
nc
e
a
nd
e
nha
nc
e
the
m
o
de
l’
s
pe
r
f
o
r
m
a
n
c
e
,
pa
r
ti
c
u
la
r
ly
o
n
th
e
mi
no
r
it
y
c
las
s
b
y
a
s
s
ig
ni
ng
a
h
ig
he
r
we
ig
ht
to
s
a
mp
les
f
r
o
m
t
he
m
i
no
r
it
y
c
las
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14
,
No.
4
,
Augus
t
2025
:
281
5
-
2825
2820
F
igur
e
2.
R
OC
c
ur
ve
s
f
or
XG
B
with
va
r
ied
numbe
r
s
of
e
s
ti
mator
s
F
igur
e
3.
R
OC
c
ur
ve
s
f
or
R
F
with
va
r
ied
numbe
r
s
of
e
s
ti
mator
s
F
ur
ther
mor
e
,
we
buil
t
a
n
S
VM
model
us
ing
dif
f
e
r
e
nt
ke
r
ne
l
f
unc
ti
ons
.
T
h
is
include
s
li
ne
a
r
,
r
a
dial
ba
s
is
f
unc
ti
on
(
R
B
F
)
,
po
lynom
ial,
a
nd
s
igm
oid
f
u
nc
ti
ons
s
hown
in
F
igur
e
4.
I
t
a
ppe
a
r
s
that
f
or
li
ne
a
r
,
R
B
F
,
a
nd
pol
ynomi
a
l
ke
r
ne
ls
,
the
tr
a
ini
ng
s
c
or
e
is
r
e
latively
high
whe
n
us
ing
f
e
w
s
a
mpl
e
s
f
or
t
r
a
in
ing
a
nd
de
c
r
e
a
s
e
s
whe
n
incr
e
a
s
ing
the
nu
mber
o
f
s
a
mpl
e
s
.
I
n
c
ontr
a
s
t,
the
c
r
os
s
-
va
li
da
ti
on
s
c
or
e
s
tar
ts
a
t
a
moder
a
te
leve
l
a
nd
s
hows
a
s
li
ght
incr
e
a
s
e
whe
n
a
dding
s
a
mpl
e
s
.
W
he
r
e
a
s
the
plot
f
o
r
S
igm
oid
ke
r
ne
l,
the
tr
a
ini
ng
s
c
or
e
r
e
mains
low
r
e
ga
r
dles
s
of
the
s
ize
of
the
t
r
a
ini
ng
s
e
t.
On
the
other
ha
nd,
the
c
r
os
s
-
va
li
da
ti
on
s
c
or
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
N
on
-
s
mall
c
e
ll
lung
c
anc
e
r
ac
ti
v
e
c
ompounds
dis
c
ov
e
r
y
holdi
ng
on
pr
otein
e
x
pr
e
s
s
ion
…
(
Ham
z
a
Hanafi)
2821
de
c
r
e
a
s
e
s
with
the
s
ize
of
the
tr
a
ini
ng
da
tas
e
t.
I
nd
e
e
d,
it
de
c
r
e
a
s
e
s
to
a
point
whe
r
e
it
r
e
a
c
he
s
a
plate
a
u.
T
he
polynom
ial
a
nd
R
B
F
ke
r
ne
ls
e
na
bles
us
to
c
las
s
if
y
the
da
ta
wi
th
c
ompl
e
x
r
e
lations
hips
.
T
he
mo
de
l
wa
s
tr
a
ined
to
f
ind
the
be
s
t
bounda
r
y
that
s
e
pa
r
a
tes
th
e
a
c
ti
ve
a
nd
inac
ti
ve
mol
e
c
ules
.
W
e
s
e
t
the
r
e
gul
a
r
iza
ti
on
pa
r
a
mete
r
to
10
to
s
tr
ike
a
good
ba
lanc
e
be
twe
e
n
t
r
a
ini
ng
a
c
c
ur
a
c
y
a
nd
c
las
s
if
ica
ti
on
p
r
e
c
is
ion,
a
nd
we
us
e
d
the
‘
s
c
a
le’
opti
on
f
o
r
ga
mm
a
to
e
ns
ur
e
a
s
moot
h
d
e
c
is
ion
bounda
r
y.
T
he
s
e
c
hoice
s
a
ll
ow
our
S
VM
model
to
pe
r
f
or
m
e
f
f
e
c
ti
ve
ly,
ove
r
c
omi
ng
the
c
ha
ll
e
nge
of
c
las
s
im
ba
lanc
e
inhe
r
e
nt
in
the
da
ta.
L
a
s
tl
y,
we
uti
li
z
e
d
the
s
klea
r
n
li
br
a
r
y
to
im
pleme
nt
a
NB
c
las
s
if
ier
,
whic
h
is
a
pr
oba
bil
is
ti
c
c
las
s
if
ier
that
a
s
s
umes
f
e
a
tur
e
indepe
nde
nc
e
,
making
it
e
f
f
icie
nt
a
nd
s
uit
a
ble
f
or
lar
ge
da
tas
e
ts
.
T
he
de
f
a
ult
im
pleme
ntation
in
s
klea
r
n
e
mpl
oy
s
the
Ga
us
s
ian
NB
a
lgor
it
hm,
a
s
s
umi
ng
a
Ga
us
s
ian
dis
tr
ibut
ion
f
or
the
f
e
a
tur
e
s
.
F
igur
e
4.
L
e
a
r
ning
c
ur
ve
s
f
o
r
S
VM
with
dif
f
e
r
e
nt
ke
r
ne
ls
(
R
B
F
,
poly
,
li
ne
a
r
)
4.
RE
S
UL
T
S
AN
D
DI
S
CU
S
S
I
ON
T
his
s
e
c
ti
on
pr
e
s
e
nts
the
ke
y
f
indi
ngs
f
r
om
ou
r
s
tudy,
whic
h
f
oc
us
e
s
on
c
ompar
ing
dif
f
e
r
e
nt
a
lgor
it
hms
to
f
ind
the
be
s
t
model
c
a
pa
ble
of
pr
e
dicting
the
a
c
ti
vit
y
of
c
ompounds
tar
ge
ti
ng
a
s
pe
c
if
ic
pr
otein.
Although
numer
ous
models
e
xis
t
in
the
li
ter
a
tur
e
,
ther
e
is
a
notable
ga
p
in
identif
ying
the
opti
mal
one
.
Our
s
tudy
a
ddr
e
s
s
e
s
thi
s
ga
p
by
e
va
luating
va
r
ious
models
us
ing
pe
r
f
or
manc
e
met
r
ics
.
T
a
b
l
e
2
p
r
o
v
ides
a
c
om
p
r
e
he
ns
i
ve
s
u
m
ma
r
y
o
f
t
he
o
ve
r
a
l
l
a
c
c
u
r
a
c
y
,
p
r
e
c
is
io
n
,
r
e
c
a
l
l
,
a
nd
F
1
s
c
o
r
e
.
I
n
c
o
n
tr
a
s
t
,
F
ig
ur
e
5
d
is
pla
ys
a
h
e
a
tm
a
p
il
lu
s
t
r
a
ti
ng
t
he
a
c
c
u
r
a
c
y
,
p
r
e
c
is
io
n
,
a
n
d
r
e
c
a
l
l
a
c
hi
e
ve
d
b
y
e
a
c
h
mo
de
l
a
c
r
os
s
d
i
f
f
e
r
e
nt
c
la
s
s
e
s
.
T
a
ble
2.
Ove
r
a
ll
pe
r
f
or
manc
e
met
r
ics
of
t
r
a
ined
M
L
m
ode
ls
M
ode
l
A
c
c
ur
a
c
y
P
r
e
c
is
io
n
R
e
c
a
ll
F1
s
c
or
e
M
L
P
0.865
0.878
0.845
0.861
X
G
B
0.604
0.601
0.605
0.603
RF
0.564
0.551
0.658
0.600
S
V
M
0.593
0.595
0.598
0.597
NB
0.558
0.546
0. 649
0.593
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14
,
No.
4
,
Augus
t
2025
:
281
5
-
2825
2822
F
igur
e
5.
He
a
tm
a
p
s
howc
a
s
ing
the
a
c
c
ur
a
c
y,
pr
e
c
i
s
ion,
a
nd
r
e
c
a
ll
s
c
or
e
s
a
tt
a
ined
by
e
a
c
h
model
T
he
M
L
P
model
a
c
hieve
d
a
n
a
c
c
ur
a
c
y
of
0.
865,
indi
c
a
ti
ng
that
it
c
or
r
e
c
tl
y
pr
e
dicte
d
the
a
c
ti
vit
y
o
f
c
ompounds
f
or
inhi
bit
ing
NSC
L
C
in
a
lar
ge
major
it
y
of
c
a
s
e
s
.
I
t
ha
d
a
pr
e
c
is
ion
of
0.
878
,
mea
ning
t
ha
t
whe
n
it
pr
e
dicte
d
a
c
ompound
a
s
e
f
f
e
c
ti
ve
a
ga
ins
t
NSC
L
C
it
de
mons
tr
a
ted
the
model
’
s
a
bil
it
y
to
e
xc
lude
ir
r
e
leva
nt
c
a
s
e
s
.
T
he
F
1
s
c
or
e
of
0.
861
,
whic
h
is
the
ha
r
mon
ic
mea
n
of
pr
e
c
is
ion
a
nd
r
e
c
a
ll
,
indi
c
a
tes
the
good
ove
r
a
ll
ba
lanc
e
d
pe
r
f
or
manc
e
of
the
model
.
T
he
s
e
r
e
s
ult
s
s
ugge
s
t
that
the
M
L
P
model
pe
r
f
o
r
med
the
be
s
t
a
mong
the
tes
ted
models
in
pr
e
dicting
the
a
c
ti
vit
y
o
f
c
ompou
nds
f
or
inhi
b
it
ing
NSC
L
C
.
T
he
XG
B
model
a
c
hieve
d
a
n
a
c
c
ur
a
c
y
of
0.
604
,
i
ndica
ti
ng
moder
a
te
pe
r
f
or
manc
e
in
p
r
e
dicting
the
a
c
ti
vit
y
of
c
ompounds
f
or
inhi
bit
ing
NSC
L
C
.
I
t
h
a
d
a
p
r
e
c
is
ion
of
0.
601
,
r
e
c
a
ll
o
f
0.
605
a
nd
a
n
F
1
s
c
or
e
of
0.
603,
s
ugge
s
ti
ng
a
r
e
latively
ba
lanc
e
d
pe
r
f
or
m
a
nc
e
.
W
hil
e
the
a
c
c
ur
a
c
y
of
the
XG
B
model
i
s
lowe
r
c
ompar
e
d
to
the
M
L
P
model,
it
s
ti
ll
pr
ovides
a
r
e
a
s
ona
ble
leve
l
of
pr
e
dictive
a
bil
it
y
in
identi
f
ying
potential
c
ompo
unds
f
or
NSC
L
C
inhi
bit
ion
.
T
he
R
F
model
a
c
hieve
d
a
n
a
c
c
ur
a
c
y
of
0.
564
,
whi
c
h
is
lowe
r
than
the
M
L
P
a
nd
XG
B
models
.
I
t
ha
d
a
pr
e
c
is
ion
o
f
0.
551
,
indi
c
a
ti
ng
a
higher
r
a
te
of
f
a
ls
e
pos
it
ives
c
ompar
e
d
to
the
othe
r
models
.
How
e
ve
r
,
the
r
e
c
a
ll
va
lue
of
0.
658
s
ug
ge
s
ts
that
the
R
F
model
s
uc
c
e
s
s
f
ull
y
identif
ied
a
higher
pr
opor
ti
on
o
f
tr
ue
pos
it
ives
(
c
ompounds
with
NSC
L
C
inhi
bit
or
y
a
c
ti
vit
y)
c
o
mpar
e
d
to
other
models
.
T
he
F
1
s
c
or
e
of
0
.
600
r
e
f
lec
ts
a
moder
a
tely
ba
lanc
e
d
pe
r
f
or
manc
e
be
twe
e
n
p
r
e
c
is
ion
a
nd
r
e
c
a
ll
.
Ove
r
a
l
l,
while
the
R
F
model
s
hows
potential
in
c
a
ptur
ing
t
r
ue
pos
it
ive
c
a
s
e
s
,
it
s
uf
f
e
r
s
f
r
om
a
h
igher
r
a
te
o
f
f
a
ls
e
pos
it
ives
,
im
pa
c
ti
ng
it
s
ove
r
a
ll
a
c
c
ur
a
c
y
in
pr
e
dicting
c
ompounds
f
or
NSC
L
C
inhi
bit
ion.
T
he
S
VM
a
nd
NB
models
e
xhibi
t
we
a
ke
r
pe
r
f
o
r
manc
e
s
c
ompar
e
d
to
mor
e
a
dva
nc
e
d
models
,
s
uc
h
a
s
M
L
P
.
T
he
S
VM
model
a
c
hieve
d
a
n
a
c
c
ur
a
c
y
o
f
0
.
593
wi
th
a
ba
lanc
e
d
pr
e
c
is
ion
o
f
0.
595
,
r
e
c
a
ll
of
0
.
598,
a
nd
a
n
F
1
s
c
or
e
of
0.
597
.
W
hil
e
the
S
VM
model
de
mons
tr
a
tes
potential
in
c
a
ptur
ing
c
ompounds
with
a
nd
without
NSC
L
C
inhi
bit
or
y
a
c
ti
vit
y,
it
s
a
c
c
ur
a
c
y
f
a
ll
s
be
low
that
of
the
M
L
P
model,
s
ugge
s
ti
ng
li
mi
t
a
ti
ons
in
a
c
c
ur
a
tely
pr
e
dicting
c
ompound
a
c
ti
vit
y.
S
im
il
a
r
l
y,
t
he
NB
model
de
mons
tr
a
tes
a
ba
lanc
e
be
twe
e
n
pr
e
c
is
ion
of
0.
546
a
nd
r
e
c
a
ll
o
f
0
.
649,
r
e
s
ult
ing
in
a
mode
r
a
tely
ba
l
a
nc
e
d
F
1
s
c
or
e
of
0.
593
.
T
he
NB
mode
l,
while
pr
ovidi
ng
s
ome
pr
e
dictive
a
bil
it
y
,
lags
be
hind
oth
e
r
models
in
ter
ms
of
a
c
c
ur
a
c
y
a
nd
p
r
e
c
is
ion.
T
his
a
na
lys
is
highl
ight
s
the
c
ha
ll
e
nge
s
f
a
c
e
d
by
both
S
VM
a
nd
NB
models
in
e
f
f
e
c
ti
ve
ly
c
a
ptur
ing
the
c
ompl
e
xit
ies
o
f
the
da
ta
f
or
a
c
c
ur
a
te
p
r
e
dictions
in
NSC
L
C
inhi
bit
ion
.
Among
thes
e
models
,
the
M
L
P
model
pe
r
f
or
me
d
the
be
s
t
with
the
highes
t
a
c
c
ur
a
c
y
of
0
.
865,
indi
c
a
ti
ng
it
s
a
bil
it
y
to
make
a
c
c
ur
a
te
pr
e
dictions
.
I
t
a
ls
o
a
c
hieve
d
the
highes
t
p
r
e
c
is
ion
a
nd
F
1
s
c
or
e
va
lues
.
On
the
other
ha
nd,
the
NB
model
pe
r
f
o
r
med
the
l
e
a
s
t
with
a
n
a
c
c
ur
a
c
y
o
f
0
.
558,
indi
c
a
ti
ng
a
lowe
r
leve
l
o
f
pr
e
diction
a
c
c
ur
a
c
y
c
ompar
e
d
to
the
o
ther
mo
de
ls
.
I
t
ha
d
the
lowe
s
t
pr
e
c
is
ion
a
nd
F
1
s
c
or
e
va
lues
,
s
ugge
s
ti
ng
it
s
li
mi
tations
in
a
c
c
ur
a
tely
p
r
e
dicting
t
he
a
c
ti
vit
y
f
o
r
NSC
L
C
.
Us
ing
the
M
L
P
model,
w
e
r
a
nke
d
top
-
10
highl
y
a
c
ti
ve
mol
e
c
ules
in
NSC
L
C
.
T
a
ble
3
s
hows
a
li
s
t
of
thes
e
dr
ugs
.
T
he
r
a
nking
method
is
ba
s
e
d
on
the
pr
oba
bil
it
ies
r
e
tur
ne
d
by
the
M
L
P
M
od
e
l,
whe
r
e
thes
e
pr
oba
bil
it
ies
r
e
pr
e
s
e
nt
t
he
pe
r
c
e
ntage
of
be
longi
ng
to
the
pos
it
ive
c
las
s
,
de
mons
tr
a
ti
ng
the
l
ikelihood
of
a
c
ompound’
s
a
c
ti
vit
y
a
ga
ins
t
NSC
L
C
.
T
he
li
s
t
of
dr
ugs
pr
e
dicte
d
by
our
M
L
P
mode
l
to
tar
ge
t
NSC
L
C
a
li
gns
we
ll
with
the
dr
ugs
mentioned
in
the
medic
a
l
li
ter
a
tu
r
e
.
S
e
ve
r
a
l
of
t
h
e
dr
ugs
in
the
top
-
10
li
s
t,
s
uc
h
a
s
Os
im
e
r
ti
nib,
B
r
igatini
b,
Ale
c
ti
nib,
E
r
lot
ini
b
,
C
e
r
it
ini
b
,
A
f
a
ti
nib,
T
r
a
s
tuzuma
b,
Ada
gr
a
s
ib
a
nd
Ge
f
it
ini
b
that
a
r
e
r
e
c
ogn
ize
d
a
s
im
por
tant
ther
a
pe
uti
c
a
ge
nts
f
or
NSC
L
C
[
33]
.
T
he
li
ter
a
tur
e
highl
ight
s
the
e
f
f
e
c
ti
ve
ne
s
s
of
thes
e
dr
ugs
in
va
r
ious
s
e
tt
ings
,
including
a
dva
nc
e
d
NSC
L
C
wi
th
s
pe
c
if
ic
ge
ne
ti
c
mut
a
ti
ons
(
s
uc
h
a
s
E
GFR
m
utations
,
AL
K
-
po
s
it
ive
or
R
OS
-
1
-
pos
it
ive
NSC
L
C
)
a
nd
meta
s
tatic
NSC
L
C
.
T
he
li
ter
a
tu
r
e
a
ls
o
p
r
ovides
s
up
por
ti
ng
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
N
on
-
s
mall
c
e
ll
lung
c
anc
e
r
ac
ti
v
e
c
ompounds
dis
c
ov
e
r
y
holdi
ng
on
pr
otein
e
x
pr
e
s
s
ion
…
(
Ham
z
a
Hanafi)
2823
e
videnc
e
f
or
the
e
f
f
ica
c
y
of
thes
e
dr
ugs
,
with
inf
or
mation
on
ove
r
a
ll
s
ur
vival
,
a
nd
im
p
r
ove
d
s
ur
vival
ti
me
[
33]
.
M
o
r
e
o
ve
r
,
the
f
a
c
t
th
a
t
s
om
e
o
f
t
he
dr
ug
s
i
n
t
he
t
op
-
10
li
s
t
a
r
e
a
p
pr
o
ve
d
s
uc
h
a
s
Os
im
e
r
t
in
i
b
[
34
]
,
B
r
iga
t
in
ib
[
35
]
,
Al
e
c
ti
ni
b
[
3
6
]
,
E
r
l
ot
in
i
b
[
3
7]
,
C
e
r
i
t
in
i
b
[
3
8]
,
A
f
a
ti
n
ib
[
39
]
,
a
nd
Ge
f
it
in
i
b
[
4
0
]
,
u
nde
r
s
c
or
e
s
th
e
i
r
e
s
ta
bl
is
he
d
e
f
f
i
c
a
c
y
in
N
S
C
L
C
t
r
e
a
t
men
t
.
W
h
i
le
o
t
he
r
d
r
u
gs
s
u
c
h
a
s
S
ot
o
r
a
s
ib
[
41
]
,
T
r
a
s
t
uz
um
a
b
[
4
2
]
,
a
nd
A
da
g
r
a
s
ib
[
4
3
]
a
r
e
c
u
r
r
e
n
t
ly
un
de
r
g
oi
ng
c
l
i
nic
a
l
t
r
i
a
l
s
f
u
r
t
he
r
c
on
f
i
r
ms
th
e
i
r
r
e
le
va
nc
e
in
NS
C
L
C
t
r
e
a
t
me
nt
.
T
a
ble
3.
T
op
-
10
r
a
nke
d
d
r
ugs
in
lung
c
a
nc
e
r
R
a
nk
D
r
ug n
a
me
1
O
s
im
e
r
ti
ni
b
2
B
r
ig
a
ti
ni
b
3
A
le
c
ti
ni
b
4
E
r
lo
ti
ni
b
5
C
e
r
it
in
ib
6
A
f
a
ti
ni
b
7
S
ot
or
a
s
ib
8
T
r
a
s
tu
z
uma
b
9
A
da
gr
a
s
ib
10
G
e
f
it
in
ib
5.
CONC
L
USI
ON
Our
s
tudy
s
howe
d
the
potential
of
int
e
gr
a
ti
ng
p
r
otein
e
xpr
e
s
s
ion
a
na
lys
is
a
nd
M
L
tec
hniques
f
or
a
c
ti
ve
c
ompounds
dis
c
ove
r
y
in
lung
c
a
nc
e
r
tr
e
a
tm
e
nt.
B
y
leve
r
a
ging
ge
ne
e
xpr
e
s
s
ion
da
ta
a
nd
tar
ge
ted
pr
otein
a
na
lys
is
,
we
s
uc
c
e
s
s
f
ull
y
identif
ied
bioac
t
ive
c
ompounds
that
s
pe
c
if
ica
ll
y
tar
ge
t
pr
oteins
a
s
s
oc
iate
d
with
NSC
L
C
.
T
hr
ough
us
ing
va
r
ious
M
L
models
,
including
M
L
P
,
XG
B
,
R
F
,
S
VM
,
a
nd
NB
,
we
c
ompar
e
d
their
pe
r
f
or
manc
e
s
in
pr
e
dicting
the
a
c
ti
vit
y
of
c
o
mpounds
.
Among
thes
e
models
,
the
M
L
P
model
e
xhibi
ted
the
highes
t
F
1
s
c
or
e
,
a
c
hieving
a
n
im
p
r
e
s
s
ive
va
lue
of
0.
861
,
de
noti
ng
it
s
a
bil
it
y
to
a
c
c
ur
a
tely
p
r
e
di
c
t
a
c
ti
ve
c
ompounds
f
or
NSC
L
C
tr
e
a
tm
e
nt.
F
ur
ther
mor
e
,
o
ur
s
tudy
pr
ovides
a
li
s
t
of
10
d
r
ugs
pr
e
dicte
d
a
s
a
c
ti
ve
in
NSC
L
C
,
a
ll
of
whic
h
a
r
e
s
uppor
ted
by
r
e
leva
nt
s
c
ientif
ic
e
videnc
e
.
T
he
s
e
f
indi
ngs
c
ontr
ibut
e
to
the
dr
ug
dis
c
ove
r
y
pipeline
f
or
lung
c
a
nc
e
r
,
o
f
f
e
r
ing
va
luable
ins
ight
s
int
o
the
de
ve
lopm
e
nt
of
tar
ge
ted
ther
a
pies
.
Ac
c
or
dingl
y,
the
int
e
gr
a
t
ion
of
c
omput
a
t
ional
methods
with
bioi
nf
or
matic
tool
s
pr
ovides
a
powe
r
f
ul
a
ppr
oa
c
h
to
a
c
c
e
ler
a
te
the
identif
ica
ti
on
a
nd
e
va
lu
a
ti
on
of
nove
l
c
ompounds
,
ult
im
a
tely
a
dva
nc
ing
pr
e
c
is
ion
medic
ine
in
the
tr
e
a
tm
e
nt
of
lung
c
a
nc
e
r
.
F
utu
r
e
r
e
s
e
a
r
c
h
will
f
oc
us
on
va
li
da
ti
ng
the
p
r
e
dicte
d
c
omp
ounds
in
pr
e
c
li
nica
l
a
nd
c
li
nica
l
s
tudi
e
s
to
f
ur
ther
c
onf
ir
m
t
he
ir
e
f
f
ica
c
y.
AC
KNOWL
E
DGE
M
E
NT
S
T
he
a
utho
r
s
e
xpr
e
s
s
their
g
r
a
ti
tude
to
the
r
e
vie
we
r
s
f
or
thei
r
c
ons
tr
uc
ti
ve
f
e
e
dba
c
k
dur
ing
the
de
ve
lopm
e
nt
of
thi
s
wor
k
.
T
his
r
e
s
e
a
r
c
h
wa
s
c
onduc
ted
indepe
nde
ntl
y,
without
e
xter
na
l
f
unding
or
f
inanc
ial
s
uppor
t.
F
UN
DI
NG
I
NF
ORM
AT
I
ON
Author
s
s
tate
no
f
unding
invol
ve
d.
AU
T
HO
R
CONT
RI
B
U
T
I
ONS
S
T
AT
E
M
E
N
T
T
his
jour
na
l
us
e
s
the
C
ontr
ibut
o
r
R
oles
T
a
xo
nomy
(
C
R
e
diT
)
to
r
e
c
ognize
indi
vidual
a
uthor
c
ontr
ibut
ions
,
r
e
duc
e
a
utho
r
s
hip
dis
putes
,
a
nd
f
a
c
il
it
a
te
c
oll
a
bor
a
ti
on.
Nam
e
of
Au
t
h
or
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Ha
mza
Ha
na
f
i
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
M
’
ha
med
Aït
Kbir
✓
✓
✓
✓
✓
✓
✓
✓
B
a
dr
Dine
R
os
s
i
Ha
s
s
a
ni
✓
✓
✓
✓
✓
✓
✓
C
:
C
onc
e
pt
ua
li
z
a
ti
on
M
:
M
e
th
odol
ogy
So
:
So
f
twa
r
e
Va
:
Va
li
da
ti
on
Fo
:
Fo
r
ma
l
a
na
ly
s
is
I
:
I
nve
s
ti
ga
ti
on
R
:
R
e
s
our
c
e
s
D
:
D
a
ta
C
ur
a
ti
on
O
:
W
r
it
in
g
-
O
r
ig
in
a
l
D
r
a
f
t
E
:
W
r
it
in
g
-
R
e
vi
e
w
&
E
di
ti
ng
Vi
:
Vi
s
ua
li
z
a
ti
on
Su
:
Su
pe
r
vi
s
io
n
P
:
P
r
oj
e
c
t
a
dmi
ni
s
tr
a
ti
on
Fu
:
Fu
ndi
ng a
c
qui
s
it
io
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14
,
No.
4
,
Augus
t
2025
:
281
5
-
2825
2824
CONF
L
I
CT
OF
I
NT
E
RE
S
T
S
T
AT
E
M
E
N
T
Author
s
s
tate
no
c
onf
li
c
t
of
int
e
r
e
s
t
.
DA
T
A
AV
AI
L
A
B
I
L
I
T
Y
T
he
da
ta
that
s
uppor
t
the
f
indi
ngs
of
thi
s
s
tudy
we
r
e
obtaine
d
f
r
om
the
C
hE
M
B
L
da
taba
s
e
,
whic
h
is
publi
c
ly
a
va
il
a
ble
a
t
htt
ps
:/
/www
.
e
bi
.
a
c
.
uk/che
mbl
.
RE
F
E
RE
NC
E
S
[
1]
T
.
C
he
ng,
Q
.
L
i,
Z
.
Z
hou,
Y
.
W
a
ng,
a
nd
S
.
H
.
B
r
ya
nt
,
“
S
t
r
uc
tu
r
e
-
ba
s
e
d
vi
r
tu
a
l
s
c
r
e
e
ni
ng
f
or
dr
ug
di
s
c
ove
r
y:
a
pr
obl
e
m
-
c
e
n
tr
ic
r
e
vi
e
w
,”
A
A
P
S J
our
nal
, vol
. 14, no. 1, pp. 133
–
141, 2012, doi:
10.1208/s
12248
-
012
-
9322
-
0.
[
2]
M
.
M
.
R
a
hma
n
e
t
al
.
,
“
E
me
r
gi
ng
pr
omi
s
e
of
c
omput
a
ti
ona
l
t
e
c
hni
que
s
in
a
nt
i
-
c
a
nc
e
r
r
e
s
e
a
r
c
h:
a
t
a
G
la
nc
e
,”
B
io
e
ngi
ne
e
r
i
ng
,
vol
. 9, no. 8, 2022, doi:
10.3390/bi
oe
ngi
ne
e
r
in
g9080335.
[
3]
G
.
H
us
s
a
in
a
nd
Y
.
S
hi
r
e
n,
“
I
de
nt
if
yi
ng
A
lz
he
im
e
r
di
s
e
a
s
e
de
me
nt
ia
le
ve
ls
us
in
g
ma
c
hi
n
e
le
a
r
ni
ng
me
th
ods
,”
M
e
di
c
al
R
e
s
e
ar
c
h
A
r
c
hi
v
e
s
, vol
. 11, no. 7.1, 2023, doi
:
10.18103/m
r
a
.v11i7.1.4039.
[
4]
M
.
S
ugi
mot
o
a
nd
T
.
S
ue
yos
hi
,
“
D
e
ve
lo
pme
nt
of
hol
oe
ye
s
hol
ogr
a
phi
c
im
a
ge
-
gui
de
d
s
ur
ge
r
y
a
nd
te
le
me
di
c
in
e
s
ys
te
m:
c
li
ni
c
a
l
be
ne
f
it
s
of
e
xt
e
nde
d
r
e
a
li
ty
(
vi
r
tu
a
l
r
e
a
li
ty
,
a
ugme
nt
e
d
r
e
a
li
ty
,
mi
xe
d
r
e
a
li
ty
)
,
th
e
me
ta
ve
r
s
e
,
a
nd
a
r
ti
f
ic
ia
l
in
te
ll
ig
e
nc
e
in
s
ur
g
e
r
y
w
it
h a
s
ys
te
ma
ti
c
r
e
vi
e
w
,
”
M
e
di
c
al
R
e
s
e
a
r
c
h A
r
c
hi
v
e
s
, vol
. 11,
no. 7.1, 2023, doi:
10.18103/m
r
a
.v11i7.1.4045.
[
5]
A
.
H
.
U
r
ba
ns
ki
,
J
.
D
.
A
r
a
uj
o,
R
.
C
r
e
ig
h
to
n,
a
nd
H
.
I
.
N
a
ka
ya
,
“
I
nt
e
gr
a
ti
ve
bi
ol
ogy
a
ppr
oa
c
he
s
a
ppl
ie
d
to
huma
n
di
s
e
a
s
e
s
,
”
C
om
put
at
io
nal
B
io
lo
gy
, pp. 19
–
36, 2019, doi:
10.15586/compu
ta
ti
ona
lb
io
lo
gy.2019.c
h2.
[
6]
A
.
K
.
A
A
lAbduls
a
la
m,
J
.
H
.
G
a
r
vi
n,
A
.
R
e
dd,
M
.
E
.
C
a
r
t
e
r
,
C
.
S
w
e
e
ny,
a
nd
S
.
M
.
M
e
ys
t
r
e
,
“
A
ut
oma
te
d
e
xt
r
a
c
ti
on
a
nd
c
la
s
s
if
ic
a
ti
on
of
c
a
nc
e
r
s
ta
ge
me
nt
io
ns
f
r
omuns
tr
uc
tu
r
e
d
te
xt
f
ie
ld
s
in
a
c
e
nt
r
a
l
c
a
nc
e
r
r
e
gi
s
tr
y.,”
A
M
I
A
J
oi
nt
Sum
m
it
s
on
T
r
ans
la
ti
onal
Sc
ie
nc
e
pr
o
c
e
e
di
ng
s
. A
M
I
A
J
oi
nt
Sum
m
it
s
on T
r
ans
la
ti
onal
Sc
ie
nc
e
, vol
. 2017, pp. 16
–
25, 2
018.
[
7]
H
.
H
a
na
f
i,
B
.
D
.
R
.
H
a
s
s
a
ni
,
a
nd
M
.
A
.
K
bi
r
,
“
P
r
e
di
c
ti
ng
a
c
ti
ve
c
ompounds
f
or
lu
ng
c
a
nc
e
r
ba
s
e
d
on
qua
nt
it
a
ti
ve
s
tr
uc
tu
r
e
-
a
c
ti
vi
ty
r
e
la
ti
ons
hi
ps
,”
I
nt
e
r
nat
io
nal
J
our
nal
of
E
le
c
tr
ic
al
and
C
om
put
e
r
E
ngi
ne
e
r
in
g
,
vol
.
13,
no.
5,
pp.
5755
–
5763,
2023,
d
oi
:
10.11591/i
je
c
e
.v13i5.pp5755
-
5763.
[
8]
H
.
H
a
na
f
i,
B
.
D
.
R
.
H
a
s
s
a
ni
,
a
nd
M
.
A
.
K
bi
r
,
“
P
r
e
di
c
ti
ng
ge
ne
-
dr
ug
-
di
s
e
a
s
e
in
te
r
a
c
ti
ons
by
in
te
gr
a
ti
ng
he
te
r
oge
ne
ous
bi
ol
ogi
c
a
l
da
ta
th
r
ough
a
ne
twor
k
mode
l,
”
I
nt
e
r
nat
io
nal
J
our
nal
of
A
dv
anc
e
s
in
Sof
t
C
om
put
in
g
and
it
s
A
pp
li
c
at
io
ns
,
vol
.
14,
n
o.
1,
pp. 35
–
48, 2022, doi:
10.15849/I
J
A
S
C
A
.220328.03.
[
9]
A
.
S
iv
a
kuma
r
a
nd
R
.
G
una
s
unda
r
i,
“
A
s
ur
ve
y
on
da
ta
p
r
e
pr
oc
e
s
s
in
g
te
c
hni
que
s
f
or
bi
oi
nf
o
r
ma
ti
c
s
a
nd
w
e
b
us
a
ge
mi
ni
n
g,”
I
nt
e
r
nat
io
nal
J
our
nal
of
P
ur
e
and A
ppl
ie
d M
at
he
m
a
ti
c
s
, vol
. 117, no. 20, pp. 785
–
794, 2017.
[
10]
Y
.
Z
ha
ng
a
nd
J
.
H
ong,
“
C
ha
ll
e
nge
s
of
de
e
p
le
a
r
ni
ng
in
c
a
nc
e
r
s
,”
T
e
c
hnol
ogy
in
C
anc
e
r
R
e
s
e
ar
c
h
and
T
r
e
at
m
e
nt
,
vol
.
22,
2023,
doi
:
10.1177/153303382
31173495.
[
11]
A
.
G
a
ul
to
n
e
t
al
.
,
“
T
he
C
hE
M
B
L
da
ta
ba
s
e
in
2017
,”
N
uc
le
ic
A
c
id
s
R
e
s
e
a
r
c
h
,
vol
.
45,
no.
1,
pp.
945
–
954,
2017,
doi
:
10.1093/nar
/g
kw
1074.
[
12]
D
.
K
.
S
im
a
ns
hu,
D
.
V
.
N
is
s
le
y,
a
nd
F
.
M
c
C
or
mi
c
k,
“
R
A
S
pr
ot
e
in
s
a
nd
th
e
ir
r
e
gul
a
to
r
s
in
huma
n
di
s
e
a
s
e
,”
C
e
ll
,
vol
.
170,
no.
1,
pp. 17
–
33, 2017, doi:
10.1016/j
.c
e
ll
.
2017.06.009.
[
13]
L
.
M
a
ns
i,
E
.
V
ie
l,
E
.
C
ur
ti
t,
J
.
M
e
di
oni
,
a
nd
C
.
L
e
T
our
ne
a
u,
“
C
ib
la
ge
de
la
voi
e
de
s
ig
na
li
s
a
ti
on
R
A
S
pour
le
tr
a
it
e
me
nt
de
s
c
a
nc
e
r
s
,”
B
ul
le
ti
n du C
anc
e
r
, vol
. 98, no. 9, pp. 1019
–
1028, 20
11, doi:
10.1684/bdc.2011.1
380.
[
14]
M
.
P
.
D
i
M
a
g
li
a
no
a
nd
C
.
D
.
L
ogs
don,
“
R
ol
e
s
f
or
K
R
A
S
in
p
a
nc
r
e
a
ti
c
tu
mor
de
ve
lo
pme
nt
a
nd
pr
og
r
e
s
s
io
n,”
G
as
tr
oe
nt
e
r
ol
ogy
,
vol
. 144, no. 6, pp. 1220
–
1229, 2013, doi:
10.1053/j
.ga
s
tr
o.2013.01.071.
[
15]
G
.
P
.
W
a
y
e
t
al
.
,
“
M
a
c
hi
ne
le
a
r
ni
ng
de
te
c
ts
pa
n
-
c
a
nc
e
r
r
a
s
pa
th
w
a
y
a
c
ti
va
ti
on
in
th
e
c
a
nc
e
r
ge
nome
a
tl
a
s
,”
C
e
ll
R
e
por
ts
,
vol
.
23,
no. 1, pp. 172
–
180, 2018, doi:
10.1016/j
.c
e
lr
e
p.2018.03.046.
[
16]
T
.
A
.
K
ni
jn
e
nbur
g
e
t
al
.
,
“
G
e
nomi
c
a
nd
mol
e
c
ul
a
r
la
nd
s
c
a
pe
o
f
D
N
A
da
ma
ge
r
e
p
a
ir
de
f
ic
ie
nc
y
a
c
r
os
s
th
e
c
a
nc
e
r
ge
nom
e
a
tl
a
s
,”
C
e
ll
R
e
por
ts
, vol
. 23, no. 1, pp. 239
–
254, 2018, doi:
10.1016/j
.c
e
lr
e
p.2018.03.076.
[
17]
J
.
D
.
A
m
a
r
a
l,
J
.
M
.
X
a
vi
e
r
,
C
.
J
.
S
te
e
r
,
a
nd
C
.
M
.P
.
R
odr
ig
ue
s
,
“
T
a
r
ge
ti
ng
th
e
p53
pa
th
w
a
y
of
a
popt
os
i
s
,”
C
ur
r
e
nt
P
har
m
ac
e
ut
ic
al
D
e
s
ig
n
, vol
. 16, no. 22, pp. 2493
–
2503, 2010,
doi
:
10.2174/138161210
791959818.
[
18]
P
.
M
ont
i
e
t
al
.
,
“
H
e
te
r
oge
ne
it
y
of
T
P
53
mu
ta
ti
ons
a
nd
P
53
pr
ot
e
in
r
e
s
id
ua
l
f
unc
ti
on
in
c
a
nc
e
r
:
doe
s
it
ma
tt
e
r
?
,”
F
r
ont
ie
r
s
in
O
nc
ol
ogy
, vol
. 10, 2020, doi:
10.3389/f
onc
.2020.593383.
[
19]
A
.
K
uş
oğl
u
a
nd
Ç
.
B
.
A
vc
ı,
“
C
a
n
c
e
r
s
te
m
c
e
ll
s
:
a
br
ie
f
r
e
vi
e
w
of
th
e
c
ur
r
e
nt
s
ta
tu
s
,
”
G
e
n
e
,
vol
.
681,
pp.
80
–
85,
2019,
doi
:
10.1016/j
.ge
ne
.2018.09.052.
[
20]
T
.
M
.
M
a
lt
a
e
t
al
.
,
“
M
a
c
hi
ne
le
a
r
ni
ng
id
e
nt
if
ie
s
s
te
mne
s
s
f
e
a
t
ur
e
s
a
s
s
o
c
ia
te
d
w
it
h
onc
oge
ni
c
de
di
f
f
e
r
e
nt
ia
ti
on,”
C
e
ll
,
vol
.
173,
no. 2, pp. 338
–
354, 2018, doi:
10.1016/j
.c
e
ll
.2018.03
.034.
[
21]
R
.
T
homa
s
a
nd
Z
.
W
e
ih
ua
,
“
R
e
th
in
k
of
E
G
F
R
in
c
a
nc
e
r
w
it
h
it
s
ki
na
s
e
in
de
pe
nde
nt
f
unc
ti
on
on
boa
r
d,”
F
r
ont
ie
r
s
in
O
nc
ol
ogy
,
vol
. 9, no. AU
G
, 2019, doi:
10.3389/f
onc
.2019.00800.
[
22]
R
.
Q
ur
e
s
hi
e
t
al
.
,
“
M
a
c
hi
ne
l
e
a
r
ni
ng
ba
s
e
d
pe
r
s
ona
li
z
e
d
dr
ug
r
e
s
pons
e
pr
e
di
c
ti
on
f
or
lu
ng
c
a
nc
e
r
pa
ti
e
nt
s
,
”
Sc
ie
nt
if
ic
R
e
por
ts
,
vol
. 12, no. 1, 2022, doi
:
10.1038/s
41598
-
022
-
23649
-
0.
[
23]
R
.
Y
a
ng,
X
.
X
io
ng,
H
.
W
a
ng,
a
nd
W
.
L
i,
“
E
xpl
a
in
a
bl
e
ma
c
h
in
e
le
a
r
ni
ng
mode
l
to
pr
e
di
c
ti
on
E
G
F
R
mut
a
ti
on
in
lu
ng
c
a
nc
e
r
,”
F
r
ont
ie
r
s
i
n O
nc
ol
ogy
, vol
. 12, 2022, doi:
10.3389/f
onc
.2022.9
24144.
[
24]
M
.
H
.
W
id
ya
na
nda
e
t
al
.
,
“
Q
ue
r
c
e
ti
n
a
s
a
n
a
nt
ic
a
nc
e
r
c
a
n
di
da
te
f
or
gl
io
bl
a
s
to
ma
mul
ti
f
or
me
by
ta
r
ge
ti
ng
A
K
T
1,
M
M
P
9,
A
B
C
B
1,
a
nd
V
E
G
F
A
:
a
n
in
-
s
il
ic
o
s
tu
dy,”
K
ar
bal
a
I
nt
e
r
nat
io
nal
J
our
nal
of
M
ode
r
n
Sc
ie
n
c
e
,
vol
.
9,
no.
3,
pp.
450
–
459,
2
023,
doi
:
10.33640/2405
-
609X.3312.
[
25]
J
.
E
b
e
r
h
a
r
d
t
,
D
.
S
a
n
t
o
s
-
M
a
r
t
i
n
s
,
A
.
F
.
T
i
l
l
a
c
k
,
a
n
d
S
.
F
o
r
li
,
“
A
ut
o
D
o
c
k
V
i
n
a
1
.
2
.
0
:
n
e
w
d
o
c
k
i
n
g
m
e
t
h
o
d
s
,
e
x
p
a
n
d
e
d
f
o
r
c
e
f
i
e
l
d
,
a
n
d
P
y
t
h
o
n
b
i
n
d
i
n
g
s
,
”
J
o
u
r
n
a
l
o
f
C
h
e
m
i
c
a
l
I
n
f
o
r
m
a
t
i
o
n
a
n
d
M
o
d
e
l
i
n
g
,
v
o
l
.
6
1
,
n
o
.
8
,
p
p
.
3
8
9
1
–
3898, 2021, doi:
10.1021/
a
c
s
.jc
im
.1c
00203
.
[
26]
V
.
V
.
K
le
a
ndr
ova
,
L
.
S
c
ot
ti
,
F
.
J
.
B
.
M
.
J
uni
or
,
E
.
M
ur
a
to
v,
M
.
T
.
S
c
ot
ti
,
a
nd
A
.
S
pe
c
k
-
P
la
nc
he
,
“
Q
S
A
R
mode
li
ng
f
or
m
ul
ti
-
ta
r
ge
t
dr
ug
di
s
c
ove
r
y:
de
s
ig
ni
ng
s
im
ul
ta
ne
ous
in
hi
bi
to
r
s
of
p
r
ot
e
in
s
in
di
ve
r
s
e
pa
th
oge
ni
c
pa
r
a
s
it
e
s
,”
F
r
ont
ie
r
s
in
C
he
m
is
tr
y
,
vol
. 9, 2021, doi:
10.3389/f
c
he
m.2021.634663.
[
27]
H
.
K
a
ne
ko,
“
M
ol
e
c
ul
a
r
de
s
c
r
ip
to
r
s
,
s
tr
uc
tu
r
e
ge
ne
r
a
ti
on,
a
nd
in
ve
r
s
e
Q
S
A
R
/QS
P
R
ba
s
e
d
on
S
E
L
F
I
E
S
,”
A
C
S
O
m
e
ga
,
vol
.
8,
n
o. 24, pp. 21781
–
21786, 2023, doi:
10.1021/ac
s
ome
ga
.3c
0133
2.
[
28]
M
. R
. I
r
e
s
ha
e
t
al
.
, “
M
a
c
hi
ne
l
e
a
r
ni
ng mode
l
a
nd mol
e
c
ul
a
r
do
c
ki
ng f
or
s
c
r
e
e
ni
ng me
di
c
in
a
l
pl
a
nt
s
a
s
H
I
V
-
1
r
e
ve
r
s
e
t
r
a
ns
c
r
ip
ta
s
e
in
hi
bi
to
r
s
,”
K
ar
bal
a I
nt
e
r
nat
io
nal
J
our
nal
of
M
ode
r
n S
c
ie
nc
e
,
vol
. 10, no. 1, pp. 79
–
90, 2024, doi:
10.33640/2405
-
609X.3341
.
Evaluation Warning : The document was created with Spire.PDF for Python.