The TIMES formalism is a shell program in which a large variety of individual simulators for different species, tissues or conditions could be nested. Each of the simulators might be based on subset of the library of biotransformations and be recalibrated to metabolism data specific to the species, tissue or conditions of interest. SimBuilder is a generalized approach of building simulators for smaller training sets of data under specific conditions.
The two stages of building this generalized metabolism simulator are illustrated in the Figure 1. The first stage consists of collecting and generalizing all simple transformations which could occur in the specific tissue/organ environment whereas the second is assigning probabilities to these transformations to accurately prioritize their competing implementation.
 |
| Figure 1. The two stages of a metabolism simulator building |
The first stage involves the development software which uses a training set of documented maps to deduce larger set of transformations which mostly closely reproduces the input data, makes mechanistic sense, and can be reasonable extrapolated to molecules outside the dataset. This stage is underdevelopment.
The second stage provides an automated optimization of the transformations hierarchy from an initial priority of metabolic transformations suggested by experts. To solve this complex optimization problem, we have uses a genetic algorithm (GA). In this approach, each candidate in the population of tentative solutions is one ordering of transformations (i.e., one simulator called also Parent ordering in Figure 2). Each candidate simulator is applied to all parent chemicals having documented maps in the training set to produce the respective theoretical maps. The generated maps are compared with the observed ones and the average similarity is calculated for all maps in the training set thus assessing the performance of the simulator. The simulators providing best performance are selected as new parents for next genetic population.
 |
| Figure 2. The scheme of application of genetic algorithm (genetic progress) for optimizing hierarchy of metabolic transformation and assigned probabilities. |