Creation of a Virtual System

A virtual biological system is a mathematical simulation that mimics a real biological system, containing all its elements and, more importantly, behaving like it.

Virtual biological systems are created in three steps:

  • [1] Constructing a protein–protein interaction map
  • [2] Loading the network with biological information
  • [3] Creating a dynamic model

1. CONSTRUCTING A PROTEIN–PROTEIN INTERACTION MAP

The first step consists in building the "skeleton" of the model as a network whose nodes are functional genes or proteins of the real system, linked according to their relationships (activation, downregulation…). This generates a static picture of the real system (Figure 1).

For the construction of this map, SIMScells uses public and private external databases (KEGG, BIND, BioGRID, IntAct, MatrixDB, MINT, REACTOME, MIPS…) and proprietary link information, extracted from scientific literature, manually curated by our expert team.

The number of nodes (proteins/genes) contained in the human map with which SIMScell works is 12,086 and the number of links used for its construction is 246,335 (with an average of 18.8 links per node).

step1

Figure 1. Construction of a protein–protein interaction map (virtual system skeleton)

2. LOADING THE NETWORK WITH BIOLOGICAL INFORMATION

After the construction of the map, SIMScells loads to this network all the information compiled in the Anaxomics Health database (AX-Health DB), which contains all the updated biological and biomedical knowledge associated with the network's elements (proteins/genes), including drugs that target them, if they are involved in pathologies or adverse events, results from clinical trials, information of microarrays, their use as biomarkers, metabolic information... (Figure 2).

Figure 2. Load of the network with biological information

3. CREATING A DYNAMIC MODEL

Besides containing identical elements, a real and a virtual model should also behave similarly. Since biological systems are dynamic, we transform their static virtual counterparts into dynamic models through systems biology methods, by training them to respond to specific stimuli in the same way that the real system does.

This requires knowing enough stimulus–response exhibited by the real biological system (whether it is a cell type, a tissue, a complete organism…), which are manually extracted by SIMScells' staff from the scientific literature. These relationships are then included in the Truth Table, which we use alongside manually curated information derived from gene expression profiling to restrict the virtual model's behaviour (Figure 3).

SIMScells works with a general Truth Table that compiles all the available information about the human organism (SIMScells virtual system: Global Human Cell), and constructs the Truth Tables for each cell type by adding the specific information about the behaviour of each biological system.

Figure 3. Creation of a dynamical model (dynamic virtual system)

At the end, we have a virtual biological system able to faithfully simulate the behaviour of its real counterpart, what allows effective, mechanistically-driven analyses of the researcher's data by means of systems biology.

NOTE: Some genes or proteins are poorly studied and there is no information about them in scientific databases. In terms of systems biology, this means that nodes corresponding to these genes or proteins are not connected in the map and thus cannot be considered in the analysis. Consequently, if such proteins or genes are introduced in the system, SIMScell analyses cannot take them into account.