Biomedical Engineering Reference
In-Depth Information
lapatinib
LY294002
(7)
Temsirolimus
U0126
PTEN
(19)
GSK3
CCND
1
BCL2L1
(4)
(10)
(11)
(12)
ERBB2/3
PIP3
AKT
(24)
NRG1
PDPK1
(20)
(21)
(22)
BAD
TSC1/2
RHEB
mTOR
(3)
(8)
(9)
BCL2
(23)
IGFRIA/B
IRS1
PIK3CA*
IGF
RP6SKB1
(6)
Ras*
SRF−ELK4
(
5)
(2)
(16)
(17)
(18)
HBEGF
EGFR
Raf
MEK1
ERK1/2
GRB2/
SRF−ELK1
SP1
POS−JUN
SOS
(1)
(13)
(14)
(15)
EGFR ERBB2
EGF
MEKK1
MKK4/7
JNK1
AG1024
AG825
Fig. 5.3
Logic circuit stuck-at fault model for GF signaling pathways
The POs will be defined as a 7-bit binary vector:
Z
=
[
FOS
−
JUN, SP1, SRF
−
ELK1, SRF
−
ELK4, BCL2, BCL2L1, CCND1
]
In all tests, the PIs are fixed to X
=
00001 as this input leads to the non-proliferative
output in the fault-free case.
For this network, six drugs are available, defined as a 6-bit vector. Each bit
corresponds to a drug, such that a value of 1 on the
i
th
bit indicates that drug
i
is
selected, and a value of 0 indicates that drug
i
is not selected. The drug vector is:
D
=
[
lapatinib, AG825, AG1024, U0126, LY249002, Temsirolimus
]
All the methods (Case 1 through 4) were implemented using an open-source weighted
partial Max-SAT solver called Maxsatz [
15
], [
16
]. Our procedure consists of scripts
which take the initial CNF, selects desired fault variables, sets output and drug
weights, and solves the CNF using Maxsatz. The satisfying assignments are then
parsed for the output and drug vectors, and reported in the results. In all examples
listed in this section, the WPMS runtime was significantly less than 1 s per CNF.
5.4.2
Simulation Results
5.4.2.1
Case 1: Single Stuck-at Fault Identification
In the single stuck-at fault model, each net was simulated for s-a-0 and s-a-1 with
no drugs, and results compared with the fault-free circuit. For fault-free circuit with
X
00001, the output vector is
Z
0
0000000. All single nonredundant stuck-at
faults, which have an output different from the fault-free circuit, are recorded and
shown in Table
5.1
. In this table, the first three columns show the affected net, the
stuck-at value, and the faulty output, respectively.
From this table, we observe that nets 13, 14, and 15 are not listed. The presence of
a fault (s-a-0 or s-a-1) on these nets does not generate an incorrect PO, and as such,
these are redundant faults. From a therapy standpoint, the genes corresponding to
these faults can be ignored.
=
=
