An interview with Alexander Stepanov on fIVE Calculator, a web application designed to automatically calculate kinetic parameters from in vitro experiment

Alexander Stepanov, QSP modeler at InSysBio and fIVE DB team member, has told about the development of fIVE Calculator, its benefits and capabilities.

fIVE DB (from In Vitro Experiments DataBase) is an InSysBio project for in vitro experiments information collection and processing. fIVE DB is designed for Quantitative Systems Pharmacology (QSP) modelers and scientists who deal with in vitro/ex vivo data analysis to simplify their hard work and avoid potential errors. 

fIVE Calculator is a web application designed to automatically calculate kinetic parameters from in vitro experiments for modeling of biosystems.

First of all, what is the core idea of fIVE Database? How is it useful? Who will most benefit from it?

Alexander Stepanov: In general, the core idea of fIVE Database is to simplify modelers’ need to search for the data and to provide assessment of cell processes rate in vitro. fIVE Database is mainly designed for QSP (Quantitative Systems Pharmacology) modelers but at the same time it can be helpful for experimental biologists, for instance, to optimize their experimental protocols as the DB contains information on a wide range of experiments, the averaged assessments are provided that can be useful for experimental biologists to plan the experiments. Still the main target audience is modelers. In view of In vivo data lack the necessity to resort to in vitro data quite often occurs in order to evaluate some parameters and to use them in QSP model development as starting values for fitting or just to fix parameters and not to fit them further.

How do you choose the articles to review for data extraction? What is the guarantee of data quality and are there the criteria for selection of articles to include as the data source?

Alexander Stepanov: It is important to mention that at the heart of the base is the framework in frames of which we used the receptors theory, derived formulas that allow to calculate cell processes rate parameters from in vitro experiments data such as cells migration, proliferation, death, differentiation. It takes much time to search for the necessary data during model development. Often it is impossible to use in vitro data directly. Or for instance, a modeler can face controversial data. From this point of view, the DB value is that it accumulates in vitro data and extracts parameters from many articles on a wide range of experimental data. The published data can include mistakes or some inaccuracies, the experiments are conducted under different protocols and different experimental conditions so that the obtained values are different. But when we collect a large amount of such data, we see the spread and can average out the data and parameters and expect that the average estimate is obtained which is quite precise and describes the cell process. Thus, the power is to accumulate a big volume of even controversial data which allow us to find the averaged estimation for the parameter and use it in model development. It gives us confidence that despite data quality within a particular article can be poor or even contain mistakes we won’t be quite wrong if we take an averaged estimation based on many articles. So, our position is to trust the authors.

How did the idea of fIVE Calculator development come about?

Alexander Stepanov: Initially, we faced the question if we could use in vitro data. So, when we are developing in vivo model in frames of which some cell processes are being described we need to have and assess parameters. Often, we can’t extract these parameters from in vivo data because they are integrated and include many other effects and processes. When we describe cell processes, we need to get parameters, for instance, constant of migration, proliferation or cytokine production rate. Thus, the question if we can evaluate these parameters from in vitro data occurred. In InSysBio the method to perform it was developed. The receptors theory is in the basis of the framework which is described in our Blog. This instrument allows to calculate such parameters values as K_base, K_max, IC₅₀, EC₅₀, I_max, E_max via in vitro data. The DB contains calculated by the formulas parameters values. It is important that the calculation is performed without fitting. Thus, only digitized experimental data are introduced into the DB and the parameters are calculated automatically (from them).

The idea to develop fIVE Calculator come about during my first year of work in fIVE DB team. Initially, the Calculator’s aim was to demonstrate the capabilities of the DB so that users could try to calculate on their own, namely, to input data into the Calculator and to get the constants. Later the functions were expanded. And now the main aim is to provide an opportunity for users to calculate parameters values themselves. So, if there are no necessary data in the DB a user can obtain the constants by the input of their own data or unpublished data.

What are the benefits of fIVE Calculator? Who is it useful for?

Alexander Stepanov: First of all, fIVE Calculator is useful for modelers for parameters estimation from in vitro data. If there are no necessary data in the DB users can find any appropriate data and calculate the parameters from them. A two-point parameters calculation proceeds in the DB now while the Calculator allows to input the whole digitized experimental curve and identify the parameter based on it. Thus, fIVE Database identifies the first point as the starting point of an experiment (time zero) and it can’t be changed. While in fIVE Calculator the first point is identified by the applied algorithm and it not necessarily coincides with the start of an experiment. It only applies to time series data. As for the DB, data collection, digitization and input in the DB are performed by the annotators and controlled by the reviewers. We have formulated the rules for points selection which are implemented into the DB and on which parameters calculation is based. I tried to perform automatic selection of such points from the whole experimental curve in the Calculator. Thus, a parameter is calculated by two points but it is possible to input the whole curve as the Calculator provides flexibility in selection of these two points. Due to algorithms flexibility, it is possible to select points for parameter calculation. However, there is still the capability for a two-point input to definitely calculate the parameter.

fIVE Calculator content: what fields does it contain? Where does it take the data from?