Gene Network Analysis

A forward modeling is a form of knowledge-driven model construction while reverse-modeling tries to use the behavior of the system itself to directly infer the interactions of the natural system. 

The characteristic path length L is the measure of the typical separation between any two nodes in the network. The value of the characteristic average shortest path length L is equal to the number of edges in the shortest path coimected by any two nodes, averaged over all pairs of nodes. Here, L determines the effective size of a network, the most typical separation of one pair of nodes therein. 

---

Gene Expression Profiling and Gene Network Analysis... 

Generally, two or more objective functions are defined for gene expression profiling and gene network analysis. Usually, these objectives are conflicting in nature. Use of traditional single objective optimization techniques to solve these multi-objective optimization problems suffer from many drawbacks. Single objective problems either use penalty function approach or use some of the objectives as constraints. Both of these approaches have user-defined biases. Thus, multi-objective optimization techniques are definitely needed to model and solve these and similar other problems. 

All major network sciences seem to concur on the view that real networks are not random but are based on robust and strong organizational principles. As explained above, major metabohc network analysis techniques could be strongly influenced by the network reductionist approach, where the behavior of the network could be potentially predicted by its elementary constituents and their interactions alone. Therefore, the hierarchical and scale-free property of gene networks can effectively complement techniques like metabolic control analysis and become a vital tool in future gene network analysis (Almaas and Barabasi, 2006). 

Konig R, Eils R (2004) Gene expression analysis on biochemical networks using the Potts spin model. Bioinformatics 20 1500-1505... 

Baginsky S, Hennig L, Zimmermann P, Gruissem W (2010) Gene expression analysis, proteomics, and network discovery. Plant Physiol 152(2/402 10... 

Li Z, Li P, Krishnan A, Liu JD (2011) Large-scale dynamic gene regulatory network inference combining differential equation models with local dynamic Bayesian network analysis. Bioinformatics 27 2686-2691. doi 10.1093/ bioinformatics/btr454... 

Traditionally, to identify the function of a protein the first step was to obtain a mutant and study its properties. This same procedure can also be applied to whole-cell models. A detailed understanding of the metabolic network is possible by understanding the behavior of the system without a key component. The FBA has been used to predict the whole-cell metabolic flux distribution for a number of mutants under different conditions. The results of the computational gene deletion analysis indicate that the model accurately predicts the qualitative growth behavior in over 85% of the cases. Genome enabled studies, such as the analysis of large sets of mutants in parallel and whole-cell transcript profiling, can be further aided by the interpretation of the data from a metabolic model. ... 

Even though Eq. (5) is more realistic than Eq. (3) as a model for biological systems, this equation still is highly oversimplified. Yet this equation has remarkable mathematical properties that facilitate theoretical analysis. Moreover, there is an expectation, demonstrated in some simple examples like those discussed above, that the qualitative dynamics in the model system wiU be preserved in more realistic versions, for example when the discontinuous step functions are replaced by continuous sigmoidal functions [33, 34, 37]. As mentioned above, synthetic gene networks have been created that show some of the simple types of dynamical behavior found in our class of networks—in particular, bistability (two fixed points) [26] and oscillation in an inhibitory loop [27]. 

---