PathwayLab offers multi-platform approach to drawing, publishing and research
This month’s commentary will be replaced by a brief software review that is actually a thinly disguised effort to induce life scientists to attempt more mathematics. In future columns, I will be performing a similar stealth act by introducing some really nice freeware programs and usage of the R-language.
A company called InNetics introduced an in silico pathway analysis tool in 2004, yet another entry into the burgeoning fields of bioinformatics and systems biology. While
click to enlarge
Figure 1: A biochemical pathway
we have been introduced to a plethora of bioinformatics software, these usually concentrate on the areas of genomics and biological pathways. The tool described in this month’s column, however, concentrates on biochemical pathways, and these types of studies actually predate the bioinformatics “revolution.” While marketed to the pharmaceutical industry as help in target selection and validation plus reduced time-to-decision, PathwayLab might be found useful by students and biochemists in those areas of research involving modeling, simulations and kinetics. The software comes in three flavors: drawing, publishing and research. The first is a basic drawing tool specialized to chemical reactions, the second includes information management features, and the last has more computational features.
Basically, this software is a biochemical pathways drawing tool layered on Microsoft Visio. As such, it will be readily and easily used by those familiar with Visio, with a short learning curve for the rest. Once the basics of finding the parts, drawing, coloring and connecting are mastered (Figure 1), and a biochemical pathway chart produced, a few simple simulations can be performed and graphics produced (Figure 2).
To really utilize the information gained, however, it is necessary to export it to platforms such as Mathematica or MATLAB in order to do to detailed mathematical analysis. As a Mathematica connect was provided, several of the pre-modeled routines were done on that platform as well as a few further analyses to test the capabilities (Figure 3).
The basic program will perform simple simulations at the push-of-a-button and will allow for user control of such inputs as reaction times, perturbations, tolerances and initial
parameter settings. With these simulations, the analyst can perform transient and metabolic control analysis, visualizing the consumption of reactants, formation of products and overall reaction kinetics. If a more detailed analysis is needed, the model can be exported to Mathematica where the following are just a sampling of the analyses that can be done: time course simulation, approximate and exact steady state computations, linearization, non-linear stability analysis, stoichiometric matrix computations, sweeping, model reduction and symbolic transformations.
A nice feature of the Mathematica language is the ability to quickly convert standard mathematical notation into something readily appreciated by the chemist, i.e., the standard form of reaction rate equations.
While the use of sophisticated mathematical programs adds much to the power of any analysis, there are many features of this multi-platform approach that may be of concern to the prospective client. The first is the purely technical issue of having to use three disparate software programs from three different companies. In case of difficulties, the novice might imagine that there would be quite a bit of finger pointing amongst developers as to whose software was at fault. Although I had no problems in having the three programs talk to each other, there is always a bit of trepidation.
The second problem is the very practical issue of software/technique mastery. It is perfectly reasonable to expect the biochemist to be at ease with Visio and drawing
click to enlarge
Figure 3: Mathematical analysis in Mathematicae
pathways, but how reasonable is it to expect a mastery of techniques involved in the very powerful analyses that Mathematica and MATLAB can perform? There are a variety of pre-coded functions available, but we are facing the hurdles of price-to-acquire, mastery of the programming techniques and, finally, mastery of the mathematics necessary to set up the problem and understand the results. Perhaps in today’s teamwork-driven atmosphere, it is expected that the duties can be apportioned out to various experts on the team.
In summary, this seems like a nice idea whose execution still needs a bit of thought. For further information, visit the InNetics Web site at www.innetics.com. The information presented there is rather limited, but contact information is given for detailed questions.
John Wass is a statistician with GPRD Pharmacogenetics, Abbott Laboratories. He may be reached at editor@ScientificComputing.com