Oscillator



Permalink

Truth and Beauty


Paul Krugman’s recent NYTimes Magazine article, “How Did Economists Get It So Wrong?” starts with criticizing economists for confusing beauty for truth in the simplified mathematical models of market economies. I’ve been thinking a lot about “beauty” and “elegance” in science since overhearing a discussion of synthetic biology where someone said that it would be possible for scientists to come up with a more “elegant” solution to biological information processing than already exists in evolved living systems.

In science (particularly physics), elegance is defined as mathematical simplicity. Here’s a fun TED talk by Murray Gell-Mann all about it beauty and truth in physics:

I think that systems biology is driven by a similar underlying principle; the idea that there is an underlying beauty to living systems that can be explained by simple, beautiful equations. Synthetic biology has been described as the “electrical engineering” to systems biology’s “physics” and I think it’s mainly for this reason. Synthetic biology aims to take the “fundamentals” from systems biology and turn them into an engineering discipline, where the simple equations that govern living systems can be used to design new ones.

Unfortunately, I think the notion that there is underlying simplicity in biological systems is flawed. This is demonstrated best in the work of that most famous physicist turned biologist, Francis Crick. Crick discovered the “central dogma of molecular biology,” a beautifully simple description of how heredity works, where DNA can replicate itself, is transcribed into RNA, which can then be translated into proteins that make up the functions of living cells.

His discovery was tremendous, but in the decades since Crick coined the phrase, the uni-directional, dogmatic simplicity has turned into a chaotic jumble. RNA can make DNA, RNA can act like some proteins, proteins can alter how genes are expressed and how RNAs are processed and more, and there are likely many more complicating factors that have yet to be discovered. In a way, it’s still beautiful, but it’s definitely not simple, and all of these complexities definitely make synthetic biology very challenging, but also fascinating in terms of what we can learn. By trying to understand why even the most successful synthetic biological systems fail to behave exactly as predicted, we may be able to more deeply understand the complexity inherent in biological systems and realize that stripping it away is counterproductive. We’re lucky that so far the failure of oversimplified biological models, unlike the consequences of sticking to oversimplified economic models, only really hurts the graduate students working on the project.


Tags: social studiesaestheticsartmodels
Comments (View)



Permalink
Creating novel protein folds (or even being able to predict how a natural one will fold) is a huge problem for basic science and for biological engineering. With the game “fold it”, not only can anyone with a computer participate in the search for better protein structure algorithms, but soon may be able to use something like this to design totally new proteins, at least at small scales. This stuff is great, but I think that the main problem with the current approaches to protein structure determination and prediction is that it is totally static. It’s still very hard to incorporate the flexibility of proteins into models of what proteins look like, and protein structures are currently determined when the proteins have been artificially crystallized into a non-biological lattice. What are we missing in our models of how proteins work? Are we losing something important to protein function by focusing computational models on predicting crystallized structures? Will we be able to actually create new protein folds from scratch using these kinds of methods? Can everything in biology even be explained by the equations of thermodynamics? One big thing that I wonder about is how a synthetic biology approach to protein structure will change the way we think about proteins. Importantly, will the focus on engineering and computational aspects of biological systems maintain the static view of how proteins look and function, or will the ability to test models in a biological context promote a more fluid, flexible understanding of how proteins work? Solve Puzzles for Science | Foldit

Creating novel protein folds (or even being able to predict how a natural one will fold) is a huge problem for basic science and for biological engineering. With the game “fold it”, not only can anyone with a computer participate in the search for better protein structure algorithms, but soon may be able to use something like this to design totally new proteins, at least at small scales.

This stuff is great, but I think that the main problem with the current approaches to protein structure determination and prediction is that it is totally static. It’s still very hard to incorporate the flexibility of proteins into models of what proteins look like, and protein structures are currently determined when the proteins have been artificially crystallized into a non-biological lattice. What are we missing in our models of how proteins work? Are we losing something important to protein function by focusing computational models on predicting crystallized structures? Will we be able to actually create new protein folds from scratch using these kinds of methods? Can everything in biology even be explained by the equations of thermodynamics?

One big thing that I wonder about is how a synthetic biology approach to protein structure will change the way we think about proteins. Importantly, will the focus on engineering and computational aspects of biological systems maintain the static view of how proteins look and function, or will the ability to test models in a biological context promote a more fluid, flexible understanding of how proteins work?

Solve Puzzles for Science | Foldit


Tags: proteinsmodels
Comments (View)



Permalink
National Science Foundation awards $1.4 million for GenoCAD development

The GenoCAD point-and-click user interface guides the user through the process of designing new sequences. By successively clicking on icons representing structural features or functional blocks, complex DNA sequences composed of dozens of functional blocks can be designed in a matter of minutes. Upon completion of the design process, the sequence can be downloaded for synthesis or further analysis.

The idea of computer aided design in synthetic biology is fascinating, intellectually interesting and important, and infuriating at the same time. As a biologist who works full time trying to build synthetic gene networks, the idea of being able to computationally design a system of genes in minutes that actually works is a beautiful dream. The problem is that right now it’s not coming up with ideas or designs that is the hard part, it’s actually getting the designed systems to work the way we think that they will in the cellular context. I wish this money was going into trying to understand some of the underlying biology—What constitutes a “part”? How do genes from different organisms work in your “chassis” strain? How can you regulate gene dynamics and cellular noise in a way that allows genetic networks to function predictably? I do hope that the community building aspect of the project does take off so that people can more easily share this kind of information with each other, because only through an enormous and concerted effort to collect this kind of data will we be able to understand genetic networks and be able to build new ones. At that point, software will only be icing on the cake.


Tags: computersCADmodelspolitics
Comments (View)