Figure 1.

BioQuali reasoning: logical steps leading to predictions. The tool encodes knowledge and observations into a system of qualitative equations which is solved with an efficient algorithm [25]. The algorithm generates the complete set of solutions of these equations and identifies invariants of the set of solutions, that is, node values that are constant throughout the entire set of solutions. These invariants are called predictions. They can also result from the logical deduction process detailed above. The tool computes predictions without detailing the reasoning steps, which would be impossible for large scale systems. Step 1 (backward deduction). The node mRNA AKT1 is observed as up-regulated. It is regulated by IGF1 only. Therefore IGF1 should be up-regulated to explain the observation. Step 2 (backward deduction). IGF1 increase cannot be derived from an increase of its transcriptional activity since its mRNA is down-regulated. The only possible explanation is a decrease of its inhibitor IGFBP3. Step 3 (backward deduction). Consequently, the only regulator of IGFBP3 should be up-regulated. Step 4 (forward deduction). All incoming regulations on PIK3C tend to increase it. This should result in an increase of the node activity. Final step. No additional input or former deduction propagates. Deductions are all in agreement with each other. They are denoted as predictions and the process ends. Alternatively, if a decrease of PIK3C is observed or deduced, the full process would fail, all deductions would be discarded and an inconsistency diagnosis would be generated.

Baumuratova et al. BMC Systems Biology 2010 4:146   doi:10.1186/1752-0509-4-146
Download authors' original image