## Virtual System analysis and resolution

Systems biology networks have so many restrictions (known data) that their treatment requires advanced mathematical methods, including artificial intelligence and supercomputing. Even so, the mathematical models of biological systems always have more variables than restrictions, so their calculation needs exceed anyone's capacity. For this reason, SIMScells creates many mathematical solutions, whose combination explains what is known about the real biological system.

This approach, besides being affordable in terms of computing power, is also consistent with the behaviour of real populations. Clinical practitioners have long known that patients vary widely in their responses to drug therapy, both in therapeutic effects and adverse reactions. Genomic studies have recently uncovered some of the underlying molecular bases that explain these differences, leading to the creation of *pharmacogenomics*. Also, biomedical research has revealed that multifactorial diseases are caused by several pathophysiological mechanisms with variable contribution to the pathology.

Therefore, instead of using an algorithm to represent the average of the population, SIMScells recreates this variability in the form of multiple mathematical algorithms that account for all the possible physiological responses in the population, creating a virtual population composed of thousands of mostly similar but slightly different virtual individuals, each of which may be seen as the model corresponding to a putative person (Figure 1). For constructing its results, SIMScells only takes into account those solutions that show a better accuracy (i.e. comply with the information stored in the Truth Table). By using the restrictions and the analytic procedure, the virtual population shows on average the same behaviour than the real population, so the average of all the individual conclusions is considered as the final solution. For instance, measuring a response (e.g. headache relief) to a drug (e.g. aspirin), most of the real individuals' headaches will disappear, which is also the case for the virtual population of algorithms.

Thus, for a given biological process, SIMScells calculates the most probable behaviour of the entire population, but also other molecular mechanisms that, though less probable, are expected to be found in certain subpopulations. This flexibility holds a huge potential in the study of molecular mechanisms driving therapeutic effect of drugs or pathological progression of diseases.