Doing Science With Virtual Biologists 29
An anonymous reader sends word of new research into automating computational experiments. A team of scientists developed a piece of software, dubbed Eureqa, to help solve complex, computationally-intense biological problems. A new paper in the journal Physical Biology details their success (abstract).
"The researchers chose this specific system, called glycolytic oscillations, to perform a virtual test of the software because it is one of the most extensively studied biological control systems. Jenkins and Vallabhajosyula used one of the process' detailed mathematical models to generate a data set corresponding to the measurements a scientist would make under various conditions. To increase the realism of the test, the researchers salted the data with a 10 percent random error. When they fed the data into Eureqa, it derived a series of equations that were nearly identical to the known equations. 'What’s really amazing is that it produced these equations a priori,' said Vallabhajosyula. 'The only thing the software knew in advance was addition, subtraction, multiplication and division.'"
Pretty impressive (Score:1)
In reality, I'm sure that there is only a very small subset of problems that this system will work for, but, there is no reason that we shouldn't put it to work on those posthaste. It will be interesting to see what it can do with an unsolved problem.
Re:Pretty impressive (Score:5, Informative)
It's neat stuff, but I'm skeptical that it will replace human biologists any time soon. As is often the case, the pop-sci writeup is a lot more dramatic than the article itself. Reading the latter, I'd say that what they've done is a clever bit of data mining combined with mathematical modeling -- they use an evolutionary algorithm to find the best set of differential equations, out of an enormous number of possible models, which describe the behavior of the data.
This is easier, and probably produces better results, than the traditional method of coming up with sets of diff. eqs. to describe the behavior of complex systems, but it's not a replacement for human judgement in coming up with the model space in the first place. (I'll also note that they performed almost the entire "experiment" on simulated data, which is always a valuable first step in the development of any modeling method, but it's not enough to show that the method "works" -- real data is always messier than the best simulations, and biological data is particularly so.) That being said, it's a very nice technique, and I'll be interested to see if the same approach can be applied to building the kinds of statistical models I work with, Bayesian networks and such.
Re: (Score:3)
The only paragraph I found at all useful:
The rest is all mystery meat about the actual method.
This is certainly the way of the future. IBM has pumping that notion right now, as well. I'm not incli
Re: (Score:3)
FWIW, mathematical modeling isn't science. It's quite important, but it isn't science until it's tested against a physical environment to determine whether the predictions are correct.
This is a bit nit-picky, but too many people seem to be forgetting it, and the distinction is extremely important.
Re: (Score:2)
Check into how this software works. It chooses a sparse set of data points, creates its "model", and then brings in more points to test against. I've used it (though not super seriously) and heard a talk by one of its creators. It's based upon a heuristic of finding the _most_surprising_, _worst_ matches to its guesses and then refining the model. In the sense that it is explicitly used to predict how well it fits to further actual, experimentally-obtained data points, your criterion of it being "tested aga
Re: (Score:2)
Yes, that's a mathematical model of science. But it doesn't become science until it makes a prediction that is then tested against the physical world. And it's the complete process that's science, not any one piece of it.
Thus, String Theory is mathematics that is endeavoring to become science. Some parts of String Theory have become science. Failed science, as the predictions were not validated, but still science. Other parts of it can't yet be tested.
Science requires BOTH the prediction of a previousl
Re: (Score:2)
PARTS of computer science are science. Parts are math. Where I went to school the science parts were emphasized in the EE department, and the math parts in the math part. I suspect that by now "Computer Science" has become it's own department, but the name is not the thing.
Geez, this stuff has been around .... (Score:2)
And of course the AI program was recalcitrant about revealing how it does it's thing, haven't all of us from time to time exhibited the exact same behavior?
Now, if only they could couple this with Google's API, and SkyGrabber, I'd bet we'd end up with some really fantas
I ran the events of the last 2 weeks through it... (Score:2)
This has been a very confusing few weeks, so allow me to summarize.
The Attorney General for Chiquita and ExxonMobil, Eric Holder, in a public announcement stated that he'd found the Smoking S**t, the connection between the Underwear Bomber's Calvin Klein's and the Iranian Quds Force.
Meanwhile, Swami Rami of India's Ta Ta Agency said that all religion is B.S. and demanded that President Obama ship 500,000 more jobs to India.
President Obama res
Re: (Score:3)
A priori or ab initio? (Score:3)
'Virtual Biologists' huh (Score:1)
So if I use Matlab to do statistical analysis (econometrics) on my econ dataset, then my computer is a 'virtual economist', huh.
And for the less informed among us... (Score:2)
When they fed the data into Eureqa, it derived a series of equations that were nearly identical to the known equations.
How useful is "nearly identical"?
Re: (Score:2)
Re: (Score:2)
It might be just some constants being a little off.
Random error (Score:2)
The problem with research is that there is no such thing a pure random error. Time and time again, we develop methods that work awesome with random error. In reality, the error in the data is not purely random, but a combination of systematic and random error (that is not Gaussian), so it takes a trained eye to work through this. I would much preferred that they used real data rather than only fabricated data, then you can say that you have something.
okay.. (Score:1)
If I told you I can fit equations with data you wouldn't think it was magical .. (after all thats exactly what any series expansions are guaranteed to do) why do biological modelers its somehow more magical to do when we are in ODE land?
-avi
Doing Intelligent Design With Virtual Creationists (Score:1)
Now that would be impressive!