perform.hmmsearch: perform.hmmsearch - Do system state optimisation using a...

In ddepn: Dynamic Deterministic Effects Propagation Networks: Infer signalling networks for timecourse RPPA data.

Description

Use an HMM to perform search for optimal system state sequence. getE and replicatecolumns are two helper functions for computing the emission probabilities and doing some internal matrix operations. perform.hmmsearch is usually called internally.

Usage

perform.hmmsearch(phi.n, bestmodel)

perform.hmmsearch_R(phi.n, bestmodel)

perform.hmmsearch_C(phi.n, bestmodel)

perform.hmmsearch_globalest(phi.n, bestmodel)

getE(x, datx, thetaprime)

replicatecolumns(mat, replicates=4)

is_consistent(phi, gammax)

checkC(phi, gammax, ch, pa)

updateA(maxima.ind, pseudocount, M, A)

viterbi(gammaposs, datx, thetaprime, T, R, M, A)

Arguments

phi.n	Adjacency matrix describing the candidate network.
datx	The data matrix.
bestmodel	Model object containing the current model.
x	State vector.
thetaprime	Model parameters for gaussians.
mat	Matrix object.
replicates	How often should each column be replicated.
phi	An adjacency matrix.
gammax	A state transition matrix.
ch	Integer denoting a child node.
pa	Integer denoting a parent node.
maxima.ind	Vector of integer. Series of state indices describing the optimal path through the system states.
pseudocount	Integer. The pseudocount value used in the viterbi update of the state transition matrix.
M	Integer. Number of states in A. Used for normalisation of the transition probability update.
A	The transition probability matrix.
gammaposs	The system state matrix containing M system states.
T	Integer. The number of time points.
R	Integer. The number of replicates.

Details

perform.hmmsearch is called whenever an HMM is to be run. Depending on the global variable IMPLEMENTATION, it decides upon the implementation of the HMM to use. IMPLEMENTATION can be set by passing command line argument implementation in the function call to ddepn. See the documentation there for a description.

is_consistent checks whether a state transition matrix is consistent with the structure. This means, for instance, that a child node ch can never be activated whenever a parent pa is active and connected via an inhibition edge. The function returns the number of found inconsitencies and some extra information. It is called in implementation R_globalest to improve the assignment of active and passive states.

Value

TODO

Note

TODO

Author(s)

Christian Bender

References

TODO

See Also

TODO

Examples

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

ddepn documentation built on May 2, 2019, 4:42 p.m.