Monday, Oct. 12, 2009 | 2 a.m.
- Budget woes raise issues of cost, value of research (4-5-2009)
- Research park running through cash, still empty (1-25-2009)
- UNLV and research: A chase for the Holy Grail? (4-9-2006)
Beyond the Sun
Martin Schiller, UNLV’s newest trophy hire, is primed to run research projects that will identify the targets for the next generation of HIV/AIDS drugs.
This is not the most important project Schiller is working on.
The most important project is using a pair of search engines that he devised to find and compare prized fragments of proteins.
The search engines, almost as easy to use as Google, can trawl the thousand-odd databases and terabytes of data that contain the recipes of known proteins, the builders and shapers of life on a cellular level. The search engines can then highlight portions of their recipes believed to have a known function and compare proteins across species. These search engines are powerful tools that Schiller calls a “hypothesis generator” because they suggest new possibilities to anyone figuring out the machinery of life.
Schiller’s search engines will help make sense of a field that combines the hardest parts of biology with the hardest parts of chemistry. Biochemistry has a reputation as a difficult science because it is one. Compared with biochemistry, little projects like building a robot, putting it on a rocket, sending it to Mars and driving the robot around remotely are pretty easy and intuitive. Biochemistry can make a nuclear chemist — a scientist who regularly contemplates how to light the fires of the sun — despair and ask to talk about something easier, like particle physics. Life — and biochemistry is life — is subtle, surprising and hard to keep up with.
Schiller’s search engines make it easier to keep up with biochemistry because one of their most basic functions is to serve as a kind of dictionary for motifs — short portions of a protein’s amino acid chain that have been identified as having particular functions.
Schiller is an intense and animated explainer of his work, lured from the University of Connecticut to run his own lab and team of graduate and undergraduate scientists at UNLV.
He would have preferred that someone else write a motif dictionary.
Schiller is a laboratory biochemist, trained to run experiments and analyze the results. A program like Schiller’s is something you would expect a computational biochemist to come up with, except that, 10 years ago, none of them had. Which is why Schiller found himself complaining to his father, Stanley, at one of their regular get-togethers to play chess and drink port.
Schiller’s father was losing at chess and pouring the port more quickly than he might have if he had been winning. Pretty soon, neither father nor son could play chess. Schiller complained to his father that every time he wanted to know what a motif did, he had to go to conferences or pore through journals, hoping someone had found his motif in his protein and figured out what it did. When is someone going build a search engine and database to find and catalog these things? he said.
“I can build you an Excel macro in a weekend,” Schiller’s father said.
“Go ahead,” Schiller said.
“I’ll do it,” his father said.
“And I was like, ‘Ah, you’re drunk,’ ” Schiller recalled.
About a month later, the macro — a miniature program that runs through Excel computer spreadsheets — arrived in an e-mail. It worked. Pretty soon, Schiller had a new project: to create a search engine to hunt through the world of proteins and flag their similarities.
Schiller’s main search engine is called the Minimotif Miner. Proteins are built with 20 amino acids, and are often represented as long chains of letters. An average protein is about 300 letters long. Schiller’s search engine looks for short chains of acids — motifs — that are known to perform specific roles in other proteins, such as modifying or bonding to molecules. If you find a motif in one protein, chances are it performs the same or a similar role in another protein. If you find a motif repeated consistently in versions of the same protein, from the yeast version to the human version, chances are it’s a very important motif.
In this way, the Minimotif Miner is almost like an incomplete but rapidly expanding dictionary. The drawback to it is that it represents these motifs in the standard two-dimensional strings of letters. Proteins, however, are three-dimensional tangles, like a ball of string you might find in your pocket. How a motif functions may depend on its interactions with other motifs, which depends on their relative locations in the tangle.
This is why Schiller’s second search-and-compare engine was created to model proteins in three dimensions. When viewed in this second program, repeated motifs that were scattered randomly in the string are often clustered together in the three-dimensional tangle. Grouped together, they may have a common function, such as hooking onto a DNA molecule.
Schiller suspects that this is like discovering syntax in language, where you can see how words relate to each other and change each other’s meanings.
But life isn’t merely a language, it’s also a construction job. Everyone says DNA is like a blueprint. Fine. We don’t live in blueprints. If you want a house to live in, you’re going to need tools. Proteins do just about everything that needs doing — sticking cells together, dividing cells, forming the superstructure of cells and signaling between cells. Proteins are life’s tools. And life has very different ideas than your average general contractor.
In house-building, more advanced tools replace their crude cousins. Flaked obsidian gave way to metal hand saws, which were replaced by circular saws and so on. In biology, pretty much every tool is still in use.
As Schiller points out, nature is amazingly modular. We share genes and proteins with earthworms and fruit flies. The genes and proteins are not exactly the same, but they’re more similar than they are different. Humans are more complex than the average nematode, so we’re built with a few new and improved tools, but working right alongside them are the same stuff that builds a worm.
If life were to build your house, it would bring along all of its tools and blueprints. Your house would be a hodgepodge of tepee walls, tumbled stone, two-by-four framing, with maybe a steel-beamed skyscraper rising out of the middle. This would not bother life, because life would just go on working on your house and start expanding across the neighborhood in a major act of urban clearance. Life really believes in job security.
And so do viruses and HIV in particular. When HIV infects a cell, it turns it into a factory for new viruses. The replication process is sloppy, creating many different versions of HIV. A person might be infected with 10,000 versions of HIV, a variety that frustrates and overwhelms the human immune systems and our drugs. Say you find a way to kill 90 percent of the viruses — the other 10 percent thrive, repopulate the host body and mutate some more.
This is where Schiller’s search engines become more than dictionaries. They become a way to design weapons.
Schiller says the extraordinary mutability of HIV makes it an ideal target for his team’s search engines, that it has “huge advantages, not for people, but for us.” If you use his engines to compare the proteins in 10,000 versions of HIV, you will find a handful of motifs that exist in every one. Drugs that attack these points will attack all HIV viruses. If the drugs hit key points, such as where the virus attaches itself to a cell, you could cripple all known varieties of HIV.
Schiller won’t be designing those drugs. He’ll be refining the tools to find where they’re needed and identifying targets.
He works about 70 hours a week.