Do It Yourself

I've always said that you don't have to be a professional working in a sterile lab in order to contribute to the advancement of science and technology, and it looks like others are starting to catch on. The Associated Press ran an article a few days ago about amateurs using relatively inexpensive equipment, working out of their homes or garages, to genetically engineer new life forms.

The best quote in the article is from computer programmer Meredith L. Patterson, who is working to alter the bacteria that create yogurt to glow in the presence of melamine, who said, "People can really work on projects for the good of humanity while learning about something they want to learn about in the process." And that's exactly the point: making a contribution, yes, but also learning something.

It sounds interesting, but the biological sciences are not really an area that I've had any training (other than what I've taught myself and what I learned in biology class in 9th grade). I have been thinking, though, about picking up some equipment for a little physics experimentation out in my garage.

The point is that you can do something. Maybe you can work on genetic engineering, or maybe you can experiment with radioactive decay in your garage. Maybe you can write computer software for scientific simulations, or maybe you could just run the BOINC software (and, by the way, the organization responsible for BOINC is looking for help with programming, translating, testing, and documenting their software, if you have any of those skills). But do something.

Hey Buddy, Wanna Be a Gamer?

"But wait," you say. "Your other posts in the 'Hey Buddy, Wanna Be a...' series have been about contributing to scientific research. And now you want us to just play games?"

Not just play games, I say. But by playing foldit, you can both entertain yourself and contribute to scientific research, specifically in the realm of protein folding.

Those fine folks at David Baker's laboratory at the University of Washington (the same people who brought you the BOINC-based Rosetta@Home project for protein-folding simulation have created foldit to take advantage of the fact that there are some things that humans are just inherently better at (like image analysis and recognition) than computers are.

It turns out that because computers are not very good a visual processing, the Rosetta@Home software sometimes returns incorrect results. But humans—even with no training in biology at all—can do a better job of identifying things visually than modern computers can. In fact, many of the best players have no training in science at all.

The game is free to download and takes about 20 minutes to learn.

I love this kind of creative approach to solving one of the great problems in science today. Proteins are responsible for almost everything that happens in our bodies, but we understand so little about them. And now, thanks to people like me and you—who don't have to know anything about them can help the scientists advance our overall understanding and possibly find new ways to cure diseases.

Rosetta@home Branches Out

The Rosetta@home project (which I blogged about in December) has branched out from its original mission of predicting protein structures. David Baker reports in the Rosetta blog that they are working on a way to convert carbon dioxide into simple sugars using enzymes computationally engineered using Rosetta@home.

David writes:

Graduate student Justin Siegal and postdoc Eric Althoff have come up with a very clever new reaction cycle using new enzymes we would collectively engineer that in total carries out the following reaction:

2C02 + 2e- + H20 -> C2O3H2 + O2

the product is a simple sugar that could be used in a variety of ways, and the removal of C02 from the atmosphere would be great for countering global warming. A nice thing about this compared to current ideas of forming inorganic carbonate compounds is that it requires no other inputs. However, it does require electrons, and hence a source of energy. We are currently assessing the energy requirements of this process and comparing them to those of other proposed carbon sequestration mechanisms.

PlanetQuest Update

I received an email today from Dr. Laurance Doyle, who (among other things) serves as the President of PlanetQuest. PlanetQuest is a nonprofit 501(c)(3) organization whose mission is to inspire global participation in the discovery of planets.

Their goal is to launch a distributed computing project using the BOINC platform that I've blogged about here before. Their software--dubbed Collaboratory--will analyze data from telescopes focused on extremely dense star regions, such as the center of the galaxy in Sagittarius in the hopes of finding planets around other stars.

From their website:

Discovering a new delta Scuti star, for example, will help astronomers better understand the stability of stars; a new Cepheid variable star would help astronomers determine how far away stars are. Most exciting of all, you could discover a new planet—a never-before-seen world beyond our solar system! You will be credited for your discovery, and your find will be entered into the PlanetQuest catalog.

Dr. Doyle's email (which came in response to my donating money toward their work on the software) contained some information on the status of their work on the software. The information was long overdue, as they haven't done a very good job of keeping the public up-to-date on their progress (although they have responded to email requests for information). Dr. Doyle writes:

We have the eclipsing binary system classifier running very well, and are now interfacing the circum-binary planet discriminator with the the binary classifier. We'll soon be going straight onto the BOINC platform with this and at that time can release an alpha version of the Collaboratory. The beta should not be far behind with a ready number of testers interested in helping us, and we are shooting for this summer to release the beta test.

Project Profile: ClimatePrediction.Net

This week's project profile is for ClimatePrediction.net, a massive project to model and forecast weather in the 21st Century.

The software, like the other projects profiled so far, operates on the BOINC platform for distributed computing. The ClimatePrediction.net system uses the unused background cycles of its members' computers to simulate a large number of possible climate scenarios to determine how each individual variable affects the overall climate picture.

These simulations are then studied individually and merged together into one and tweaked as additional data becomes available. The more data runs that are performed, the more accurate the models will become.

Accurate prediction of climate change could be vital on both the short- and long-term scales. For example, more precise climate modeling could have shown that Hurricane Katrina would strike New Orleans, rather than the predicted path that showed it striking Texas. Advanced warning could have led to better evacuation and preparation and given people an expectation of the damage before it happened.

Similarly, in the long-term time scale, climate change predictions can give us better understanding of potential warming effects such as rising ocean levels, increased storm activity, etc.

If you're interested in participating in this type of science project, you can download the software here.

Project Profile: Seti@Home

This week's profile is for the SETI@home project, probably the most famous of all distributed computing software projects and the one that gave birth to the BOINC system.

In case you didn't already know, the SETI@home software is a distributed computing project that combs through massive amounts of data returned by radio telescopes (including the famous telescope at Arecibo) in search of signals that could have an intelligent origin. Due to the enormous amounts of data these telescopes collect, the need for computing power to analyze it is immense, and thus the idea of distributed computing was born.

I don't run SETI@home at the moment, although I did run the original SETI@home application many years ago (now called SETI@home classic). I think SETI's methods are somewhat limited, in that they only scan a very narrow band of data, and only radio waves. Any advanced society attempting to communicate would likely use a different method, so I put my computing resources into other projects. I'm not one to only push the projects that I support, however, so if searching for signals from aliens is your thing, go ahead and head to their website and get the software.

Project Profile: Einstein@Home

This week's profile is for the Einstein@Home project. Like last week's profiled project, Rosetta@home, Einstein@Home uses the BOINC software as a distributed computing platform.

The Einstein@Home project uses the idle resources of a distributed network of computers to search for pulsars, which are spinning neutron stars. The software analyzes data gathered by the LIGO and GEO gravitational wave detectors.

From the Einstein@Home Screensaver description:

The Einstein@Home Screensaver has a number of elements related to current efforts to detect gravitational radiation from periodic sources such as pulsars. The primary element of the screensaver is a rotating celestial sphere showing the known constellations, along with the current zenith positions of three gravity wave detectors. The positions of the detectors relative to the stars changes periodically over a 24 hour period. If you went to one of the detector sites, the stars visible directly overhead at any time are the same ones that appear next to the detector on your screensaver. (This assumes of course that your computer's time and timezone are correctly set!) Also shown are the positions of the known pulsars and supernovae remnants, and a marker indicating the positions being searched as the calculations proceed. When the graphics are shown in a separate window (not as a screensaver) the user can control the display with the mouse and keyboard.

Project Profile: Rosetta@home

This week's profile is for the Rosetta@Home project. Rosetta@home uses the BOINC software platform for distributed computing to determine the 3-dimensional shapes of proteins in research, with the stated goal of finding cures for some major human diseases.

Basically, by running the BOINC software and setting up the Rosetta@home project, your wasted CPU cycles can be put to use helping cure some of the most devastating diseases affecting humans today, such as cancer and Alzheimer's disease.

From the Rosetta@home Science FAQ:

What is Rosetta?

Rosetta is a protein structure prediction and design program.

What is a protein?

A protein is a polymer of amino acids that is encoded by a gene.

What are amino acids?

Amino acids are chemical moieties that form the basic building blocks of proteins. There are 20 different amino acids that are specified by the genetic code. These 20 amino acids fall into different groups based on their chemical properties: acidic or alkaline, hydrophilic (water-loving) or hydrophobic (greasy).

What do proteins do?

Proteins perform many essential functions in the cells of living organisms. They replicate and maintain the genome (DNA), they help cells grow and divide, and stop them from growing too much, they give a cell its identity (eg liver, neuron, pancreatic, etc.), they help cells communicate with each other. Proteins, when mutated or when affected by toxins can also cause disease, such as cancer or alzheimer's. Bacterial and viral proteins can hijack a cell and kill it. In short, proteins do everything.

How do proteins perform all their different functions?

Each protein folds into a unique 3-dimensional shape, or structure. This structure specifies the function of the protein. For example, a protein that breaks down glucose so the cell can use the energy stored in the sugar, will have a shape that recognizes the glucose and binds to it (like a lock and key). It will have chemically reactive amino acids that will react with the glucose and break it down, to release the energy.

Why do proteins fold into unique structures?

It's long been recognized that most for most proteins the native state is at a thermodynamic minimum. In English, that means the unique shape of a protein is the most stable state it can adopt. Picture a ball in a funnel - the ball will always roll down to the bottom of the funnel, because that is the most stable state.

What forces determine the unique native (most stable) structure of a protein?

The sequence of amino acids is sufficient to determine the native state of a protein. By virtue of their different chemical properties, some amino acids are attracted to each other (for example, oppositely charged amino acids) and so will associate; other amino acids will try to avoid water (because they are greasy) and so will drive the protein into a compact shape that excludes water from contacting most of the amino acids that "hide" in the core of this compacted protein.

Why is it so difficult to determine the native structure of a protein?

Even small proteins can consist of 100 amino acids. The number of potential conformations available to even such a (relatively) small protein is astronomical, because there are so many degrees of freedom. To calculate the energy of every possible state (so we can figure out which state is the most stable) is a computationally intractable problem. The problem grows exponentially with the size of a protein. Some human proteins can be huge (1000 amino acids).

So how does Rosetta approach this problem?

The rosetta philosophy is to use both an understanding of the physical chemical properties different types of amino acid interactions, and a knowledge of what local conformations are probable for short stretches of amino acids within a protein to adopt, to limit the search space, and to evaluate the energy of different possible conformations. By sampling enough conformations, Rosetta can find the lowest energy, most stable native structure of a protein.

Why is distributed computing required for structure prediction by Rosetta?

In many cases where the native structure of a protein is already known, we have noticed that Rosetta's energy function can recognize the native state as more stable than any other sampled state. When starting from a random conformation, however, we've observed that the native state is never sampled. By applying more computing power to the problem, we can sample many more conformations, and try different search strategies to see which is the most effective.

How will Rosetta@home benefit medical science?

Please see our Disease Related Research page for information on how Rosetta is being applied to medical problems.