System and method for network association inference, validation and pruning based on integrated constraints from diverse data
A network inference and validation engine is presented which combines data of different types into a network associations' inference and performs validation of existing networks based on constraints from several data sets or previously known linkages. The engine would assist scientists to integrate information from various sources into a network of association, validate previously known associations against the supplied constraint data sets and prune or modify existing connections based on subsequent experiments.
1. A computer-implemented method to combine diverse experimental data sets and to infer a network out of such data, the method comprising the steps of:
a. designating one of the data sets as primary and the rest as secondary,
b. analyzing the secondary data sets to obtain a connection matrix out of the secondary data sets by using statistical methods comprising R-square measures, clustering, correlation studies and mutual information measures,
c. evolving a population of models and choosing a model from the population by proposing a population of plural nodes and connections from representation and representing the population by strings of characters and associated trees, with the string representation including characters representing data entities as well as a choice of mathematical operators,
d. evaluating the values of the associated trees by integrating or iterating differential or difference equations along branches of the trees for a candidate in the population,
e. assigning a fitness measure to each candidate based on i) presence of known motifs in the network, ii) stability of the network as evaluated by a linear stability analysis, iii) statistical measure of correlations in the data and iv) consistency with the prior known connections in the network, and
f. incorporating penalties obtained from connection matrices in step b) into the fitness measure to arrive at an initial guess population consistent with the connections from the experimental data sets.
2. The computer-implemented method of claim 1 where the data sets include gene expression data, protein interaction data and gene knockout experiment data.
3. The computer-implemented method of claim 1 , wherein the data sets include data representative of experimental data, knowledge from the literature, patient data, clinical trial data, compliance data; chemical data, medical data, or hypothesized data.
4. The computer-implemented method of claim 1 , wherein the data sets include multivariate, parameterized data including, but not restricted to time series data, financial data, email or other social network data, simulated data from a known network structure.
5. The computer-implemented method of claim 1 , wherein the primary datasets are gene expression data, gene expression profiles with varying environmental conditions including time, protein interaction data and gene knockout experiment data.
6. The computer-implemented method of claim 1 , wherein the secondary data sets are location analysis, protein-protein interaction, two hybrid data, and pathway information or transcription factor relations obtained from sequence or previous knowledge represented in form of a connection matrix.
7. The computer-implemented method of claim 1 , further comprising the step of interchanging the configuration of primary and secondary data sets and obtaining a consensus network.
8. The computer-implemented method of claim 1 , wherein the method further comprises the step of computer-implemented tuning of various inference or evolution parameters.
9. The computer-implemented method of claim 1 , wherein the tuning of various inference or evolution parameters is computer-implemented.
10. The computer-implemented method of claim 1 , wherein the steps are performed iteratively for a specified number of iterations or until a specified accuracy threshold is reached.
11. The computer-implemented method of claim 7 , wherein the method further comprises the step of computer-implemented tuning of various inference or evolution parameters.
12. The computer-implemented method of claim 8 , wherein the tuning of various inference or evolution parameters is computer-implemented.