Feb 15, 2006
Keith: I think that standards are rarely successfully proposed, but are usually adopted when someone successfully implements something of use. TCP/IP is a great example. Nobody initially came out and said everybody who is doing networking must use TCP/IP.I think what the industry will do is standardize upon successful approaches that are shown to work. I think SBML is great in academic circles, but we don’t use it. We haven’t found a whole lot of use for it internally.
Alex: We support it within our software platform, but for the kinds of models it was designed to be used for. If you’re trying to develop a biochemical reaction model, you can use SBML. If you’re developing a high-level model that’s designed for [something] different, that isn’t just biochemical reactions, SBML doesn’t fit.
Herbert: I agree. It was designed over five years ago for specific reasons. But there was demand, and that’s why probably it’s been relatively successful. At least 100 [software] packages now support it. Nature endorses it, and a number of other journals now are endorsing it. If you want to submit a model, you have to submit SBML in with it. But it doesn’t try to do everything.
Keith: If we focus on developing technology and trying to provide technology by developing standards, we will have no industry development. I think the bioinformatics industry is a good example of that, where people tried to focus on technology. I think what we’ll see in this space is if there is real commercial value, this space will explode and go very rapidly. If there isn’t, it will be a slow burn in academia until it gets to that flash point.
Return to main article.