Systems and methods for identifying drug targets using biological networks
View Patent ↗Certain embodiments of the invention may include systems and methods for identifying drug targets using biological networks. According to an example embodiment of the invention, a method is provided for predicting the effects of drug targets on treating a disease. The method can include constructing a structure of a Bayesian network based at least in part on knowledge of drug inhibiting effects on a disease; associating a set of parameters with the constructed Bayesian network; determining values of a joint probability distribution of the Bayesian network via an automatic procedure; deriving a mean Bayesian network with one or more averaged parameters based at least in part on the joint probability values; and calculating a quantitative prediction based at least in part on the mean Bayesian network.
1. A method, comprising executing computer executable instructions by one or more processors for automatically determining a multinomial distribution in a Bayesian network, the method further comprising:
constructing a graphical structure of a Bayesian network using a set of joint probability parameters associated to the constructed Bayesian network structure;
designing an automatic procedure to determine values associated with said joint probability parameters;
simultaneously deriving an equivalent mean Bayesian network or a class of Bayesian networks; and
performing quantitative predictions and reasoning simulations using the equivalent mean Bayesian network or the class of Bayesian networks wherein a matrix is constructed which places a relative order over the child's probability under a combination of activating and inhibiting parent notes, wherein matrix values P i,j are calculated as
P
0
,
0
=
P
(
D
,
A
_
)
P
(
D
,
A
_
)
+
P
(
D
_
,
A
_
)
,
and
P
0
,
1
=
P
(
D
,
A
)
P
(
D
,
A
)
+
P
(
D
_
,
A
)
wherein P i,j , indicates child's probability in the presence of j activating parent nodes and i inhibiting parent nodes; P 0,0 indicates child's probability in the presence of 0 activating parent nodes and 0 inhibiting parent nodes, and P 0,1 indicates child's probability in the presence of 1 activating parent nodes and 0 inhibiting parent nodes; and wherein value P i,j is represented by the joint probability.
2. The method of claim 1 , wherein the graphical structure of the Bayesian network is constructed at least in part by adding directed edges between variables according to a domain expert or knowledge resource.
3. The method of claim 1 , wherein values of the joint probability are determined at least in part based on a local structure in the Bayesian network, wherein the local structure comprises one or more activating parents, which when present, increase the probability associated with corresponding child nodes.
4. The method of claim 1 , wherein values of the joint probability are determined at least in part based on a local structure in the Bayesian network, wherein the local structure comprises one or more inhibiting parents, which when present, decrease the probability associated with corresponding child nodes.
5. The method of claim 1 , wherein the equivalent mean Bayesian network comprises a Bayesian network with averaged parameters from the Bayesian network class.
6. The method of claim 1 , wherein the constructed Bayesian networks are defined by the structures and parameters: B 1 =(G i , Θ 1 ); B 2 =(G 2 , Θ 2 ); B 3 =(G 3 , Θ 3 ); B 4 =(G 4 , Θ 4 ); B 5 =(G 5 , Θ 5 ) where joint probabilities comprise: Θ 1 =P(A,B,C,D); Θ 2 =P(A,B,D); Θ 3 =P(A,B,D); Θ 4 =P(A,D); Θ 5 =P(A,D), given structures G1-G5.
7. The method of claim 1 , wherein an order between the joint probabilities comprises
P
(
D
,
A
_
)
P
(
D
,
A
_
)
+
P
(
D
_
,
A
_
)
≤
P
(
D
,
A
)
P
(
D
,
A
)
+
P
(
D
_
,
A
)
.
8. The method of claim 1 , wherein the joint probability comprising P(D,Ā)+P(D,A)+P( D ,A)+P( D ,Ā) sums to 1.