A computing consultant with a part-time affiliation at the agricultural University of New Hampshire, Will Gilbert hardly seems a likely candidate for posterboy of Apple’s latest marketing campaign aimed at the bioinformatics community (in fact, you may have seen some of the ads in this magazine).
Gilbert wasn’t always the Mac devotee he is now (from the sounds of it, he wouldn’t trade in his G4 for water on a desert island). “I was one of those homebrew guys, back in the ’70s, building my own computers,” he recalls. He started as a biochemist and joined the MIT faculty in 1979, where he got into protein crystallography and learned computer engineering along the way. By the late ’80s, he had become both director of computing at the Whitehead Institute and the Mac aficionado he is today, introducing the Macintosh platform at Whitehead. (To this day, Gilbert has no connection with Apple.)
Gilbert finally left MIT for the life of a consultant, giving up the two-hour commutes he had endured for his Cambridge job. Ever since, he’s been working with startup biotech companies, and also spends about 15 percent of his time doing work for UNH, where he mentors computer science students.
It was his work with Blast that caught Apple’s attention. A student at the UNH genome center looking to study the difference between chimp and human was convinced that finding low-density SNP regions conserved in the human genome “would suggest what genes make us human,” Gilbert says. But the process wasn’t as easy as it looked on paper — comparing low-density regions from the Santa Cruz assembly to genes from the NCBI assembly didn’t work; the coordinates “weren’t even close” between the two. As a solution, “we needed to run Blast with 200KB segments against the human genome,” Gilbert says. But that proved unworkable, too: Gilbert stopped the first of 192 runs when it was still going after 16 hours.
“I remembered reading something about A/G Blast,” Gilbert says. He increased word size by a factor of 10 and did some QC testing to make sure the method wouldn’t sacrifice sensitivity. The first trial came up with results in 30 seconds. To move the experiment from his own Mac to one down the hall, Gilbert loaded the genome onto his iPod and ported it to the other machine. With a basic Perl script, the 192 runs were completed in some three hours, indicating to Gilbert that “if you take two genomes that are fairly similar, you can compare extremely large pieces of DNA very quickly.”
The current ad campaign has Gilbert amused. Through all the Mac years — including the “dark times in the late ’90s” — “I’ve stuck with Apple,” Gilbert says. “So Apple’s sticking with me.”
— Meredith Salisbury