Private and public sector labs have recently released new biosimulation software tools that promise to bring a new level of user-friendliness to complex in silico modeling and simulation experiments.
Last week, scientific software firm the MathWorks launched SimBiology, a new product built upon its Matlab numerical computing environment that allows modelers and biologists to model, simulate, analyze, and share biochemical pathways.
The release followed the mid-June launch of COPASI (Complex Pathway Simulator), a software package from the Virginia Bioinformatics Institute and EML Research that also enables users to model, simulate, and analyze biological pathways and networks.
Both packages aim to bring biosimulation — a methodology that largely remains a specialty of expert modelers — to the biologist's desktop.
SimBiology: Pathway Graphics with Matlab Under the Hood
"Modelers and scientists don't communicate well," said Kristin Amuzzini, technical marketing manager at The MathWorks. In addition, she noted, while many drug-discovery groups have adopted software that allows them to analyze pathway diagrams, these tools do not enable simulation. Furthermore, the semantics and symbols across different pathway-analysis packages can vary greatly, since there is no established standard for representing biological entities in computational models.
|
"In order to make a program powerful, you normally have an interface that is not so easy to use, and we've been really trying to stay on the edge there, of making it easy enough for the average biologist to use, but then also to have the most advanced and powerful numerical algorithms." |
The goal in building SimBiology, she said, was to create a single environment that would enable "both graphical and programmatic pathway analysis." The package includes graphical elements that biologists are familiar with and can use to build biological pathways. Underlying those friendly graphics, however, is the Matlab programming language for those more comfortable with ordinary differential equations and a command-line interface.
Users can import SBML files into SimBiology or build their own models through a choice of two interfaces: a tabular "behind the scenes" interface that allows users to enter molecular species, reactions, parameters, and other variables into a table; or a block diagram editor that offers a graphical depiction of biological pathways. Individual users can customize the program with their own graphical element settings so that kinases always appear as red squares, for example, regardless of the notation used by whoever built the model.
SimBiology runs simulations with both stochastic and deterministic solvers, and also enables sensitivity analysis to predict the effect of specific parameters on the system and parameter estimation to help fit the model to experimental data. These features are expected to aid hypothesis generation, Amuzzini said. Currently, she noted, "many people don't go as far as they could with a model. They go to the wet bench instead."
Birgit Schoeberl, director of network biology at Merrimack Pharmaceuticals, said that Merrimack was already using Matlab for biosimulation, so the firm opted to give SimBiology a try. She said that the new program offers a number of benefits over Matlab, including a better user interface for graphically representing pathways and the ability to share models.
Likewise, Anna Georgieva, senior scientist at Novartis, said that the company's modeling and simulation group of around 50 people has added SimBiology to a "huge" list of tools that already included Matlab, the MathWorks' Simulink software, and a number of other statistical and mathematical modeling packages.
"Generally speaking, there is nothing that this tool can do that the basic Matlab package cannot do," Georgieva said. "However, as applied to pathway models that are part of a systems biology model, it adds flexibility, readability, user friendliness, and it's easier to pass the results of such a tool to a biologist."
Georgieva noted that the MathWorks "started later than some of the other tool developers" in providing a biosimulation software package, but said that the company has "put their minds to it and they're good in hearing user requirements, so it's great to see this tool evolving."
Schoeberl said that Merrimack considered a few other commercial and academic biosimulation alternatives, but that none of those options offered the flexibility of SimBiology. With other biosimulation packages, "you can only do what the package offers you to do," she said, "whereas with Simbiology or Matlab you can use features that have already been implemented by the MathWorks or you can write your own code and implement whatever you like."
Modeling is an "important" part of Merrimack's discovery process, Schoeberl said. "We use these models to design our antibody therapeutics, to identify which would be the best targets [by] doing sensitivity analysis on pathways, and to do experimental design," she said.
But Schoeberl said that it's still too early to say how well suited the software is for end-user biologists. "We're trying to get there right now," she said.
SimBiology is available for Windows, Linux, and Mac OS. List price is $3,000 for a single-user license, and academic discounts are available.
COPASI: Bleeding-Edge Simulation in a Friendly Package
On the public sector side, the first official release of COPASI is the culmination of six years of co-development by Pedro Mendes at VBI and Ursula Dummer at EML Research.
COPASI is the successor to Mendes' Gepasi simulation program, but includes new capabilities such as stochastic simulation and support for Windows, Linux, Mac OS X, and Solaris.
Like SimBiology, COPASI supports SBML and enables biologists to construct their own biochemical models.
"The main interest is really to target the wider biological community," Mendes told BioInform. "Having said that, we're trying to add a number of quite advanced packages. So there is a little bit of a challenge there, because in order to make a program powerful you normally have an interface that is not so easy to use, and we've been really trying to stay on the edge there of making it easy enough for the average biologist to use, but then also to have the most advanced and powerful numerical algorithms."
The key, he said, is the interface. In order to be of use to biologists, "it can't talk about variables and equations and reactions," he said, noting that "a good interface will do all the necessary work in the background."
Mendes said that COPASI uses the Gillespie algorithm for stochastic simulation, which is particularly useful for gene-expression data "when genes have very low expression and it's very dominated by noise." Mendes said that he is currently adding several tools for nonlinear dynamics, bifurcation analysis, and other capabilities that are not available in other programs.
While COPASI competes with SimBiology in a certain sense, Mendes noted that "competition is healthy in the field" and that there are some important differences that set the two packages apart. "There are some advantages of being an academic group," he said. "One of them is we are always on the lookout for algorithms and in fact doing research on these things, so we will probably be bringing some of these new algorithms into the package quicker than they would."
On the other hand, he noted, "they probably in some ways offer a bit more polished package."
COPASI is free and open source for non-commercial use (http://www.copasi.org/tiki-index.php), with commercial licenses handled on a case-by-case basis.