Table 1.
Methods | Information richness | Scalability | References |
---|---|---|---|
Correlation/mutual information | Low | High (thousands of genes) | [20,28] |
Partial correlation | Medium | Medium (up to 100 genes using heuristics) | [29,33] |
Differential equations | Medium | Medium | [2,32,34,36] |
Linear regression | Medium | Medium | [38] |
Non-linear regression | High | Low (up to 25 genes) | [38] |
Boolean | High | Low (up to 25 genes) | [11,35] |
It is clear that there is a trade-off between information richness (the number of factors that can be applied to predict gene expression) and the size of the analyzed network. Small networks can be handled by methods that are highly complex and information rich (many linear and non-linear factors can influence a gene within the method). Combining several small network modules holds the potential to analyze a large network [5], although this might not always work.