David Schwartz, a physical chemist at the University of Wisconsin, aspires to be the Rand McNally of genomics.
Following the parade of researchers hoping to cash in on academically developed technologies, Schwartz is counting on a high-resolution mapping technique he developed that pins down even the most diffi-cult repeat-rich DNA sequences to specific sites on a genome.
Wed like to start a company, said Schwartz. Theres a big demand for it, he said, referring to his optical mapping technique and a suite of custom-made software to analyze the resulting data.
Although Schwartz still has to hammer out a business plan and attain financing, he said the company would at first provide optical mapping services and offer map database sub-scriptions to academia and pharmaceutical companies. Diagnostics would ideally follow.
Currently, institutions such as the Institute for Genomic Research send sequence contigs to Schwartzs lab to anchor them to specific points on the genome and to solve one of the biggest problems they face--repeats, repeats, repeats. If a genome contains lots of repeats, the assembler cannot disentangle the information and, as a result, it often makes bogus assemblies.
Its a very nice validation tool for convincing ourselves that the assemblies work well, said Owen White, deputy director of bioinformatics and head of assembly at TIGR.
With as much as 20-30 percent of the human genome made up of repeats, Schwartz said, these regions provide a veritable goldmine for researchers who can make sense of them. Although some people refer to these regions as junk, Schwartz maintains that they contain lots of interesting biology.
Its just a bunch of scientists who are scared of the dark, said Schwartz. What scares us is that we dont have good tools to really go in there and make sense of it.
Thats where maps come in. The higher the resolution of the map, the greater the precision of the sequence location. If the genome were a map of the United States, what you want to find out at the end of the day, is everyones address, said Schwartz.
So whats happened with all this shotgun sequencing is that theyve gotten a lot of addresses, said Schwartz. Well, where do all these addresses belong?
The first step in piecing the puzzle back together is bacterial artificial chromosome ordering. After tracing the sequences back to the clones they originate from, BAC ordering puts the states in the right order, said Schwartz.
Zoom in and BAC fingerprinting narrows the range of the sequence fragment to areas within the clone. Its like a listing of all the counties in the state, said Schwartz. But where these counties are relative to each other is still not clear.
Genetic markers, or landmarks, can also be found, but the distance between the landmarks is unknown.
But with high resolution optical mapping you can give me an address and, very precisely, I can tell you where that address is, said Schwartz. I can tell you this person is on the eighth floor, this person is on the seventh floor and so on.
That combined with sequence gives you everything you need in terms of putting a genome together, he added.
The optical-mapping technique Schwartz developed involves stretching single fluorescence-labeled DNA molecules over a glass surface in a way that they remain biochemically accessible. A series of restriction enzymes cut the DNA in various places.
Because the DNA molecule is like a stretched rubber band on sticky flypaper, if you cut the rubber band with a razor blade, it will relax back a bit at the cut site to produce a visible gap, explained Schwartz.
And then by measuring the fluorescence, Schwartz determines the mass and length of the restriction fragments.
To assemble the fragments and complete the maps, Schwartz has turned to computer-scientist mathematicians Thomas Anantharaman of University of Wisconsin and Bhubaneswar Mishra of New York Universitys Courant Institute of Mathematical Sciences. As collaborators and potential business partners they developed Gentig--a map assembler--and a package of other analysis software.
Gentig is a Bayesian-based program that automatically builds contigs from the optical-mapping data. It repeatedly combines the two islands that produce the greatest increase in probability density and excludes false-positive overlaps.
It corrects for the error rates whenever it can, but it also knows when it failed to correct them, said Mishra. If you give it molecule maps that are not very good, it tells you that, so you dont waste your time.
Other software products include Convex (Contig Visualization and Exploration), a contig viewer that allows you to see and zoom in on the maps and the un-derlying statistics, and Validator, which compares sequence assemblies with the Gentig-generated map and points out discrepancies.
Mishra said they are also about halfway through developing a map-based sequence assembler. Celera uses a map-based sequence assembler. And the University of Southern California, TIGR, and others are developing their own versions too.
Despite the methods value to institutions such as TIGR, some people question the commercial value of Schwartzs optical mapping technology.
The rate limiting step for him is that it is by no means high-throughput, White said.
But Mishra countered that it is only a matter of time and resources until they achieve high-throughput.
The nice thing about optical mapping and all the related technologies is that its what a computer scientist would call mindlessly parallelizable. A lot of the components can proceed simultaneously without interfering with each other. So that makes it very easy for us to im-prove the throughput, he said.
Eventually, Mishra thinks the system will be able to crank the throughput to 1,000 human maps a day. And at these levels studies on chromosomal variations as well as diagnostics would be possible.
By aligning and analyzing maps for two different people, Schwartz said he would be able to see disease-inducing differences that dont show up in SNP scoring, such as gene duplication or translocation.
If he could make it possible to score a lot of chromosomes quickly, then sure, it would be good way to look for abnormalities, said White. That sounds believable.
--Aaron J. Sender