BERKELEY, Calif.--Neomorphic Software's office here has the casual look of graduate student quarters--no surprise considering the company's academic origins and existing ties. Cyrus Harmon, CEO, and Gregg Helt, chief technology officer, got their start on the University of California, Berkeley, Drosophila Genome Project. Martin Reese, chief scientific officer, is an alumnus of the Lawrence Berkeley National Laboratory Human Genome Center.
While in graduate school, the three developed computational tools to speed up different phases of their sequencing projects. "We thought if we could put the three tools together we could build some nice applications," Harmon said.
The trio left graduate school to found Neomorphic in the fall of 1996. Their company now supports a staff of 10, and Harmon said he expects continued growth.
Harmon attributed part of the company's success to its strong biology base. "We are very biologically driven," he said. "We are looking to solve problems in a way that really makes sense to biologists."
All of Neomorphic's products grew out of frustrations Harmon, Helt, and Reese experienced in their work. If they were not finding the tools they needed on the market, other people probably weren't either, the team guessed. And because they were designing tools for themselves, they designed them in a way that was most useful to biologists.
Harmon said that the team needed to leave the academic setting in order to make the most of their preliminary work. "The focus of academics and industry are different," he explained. Grant money being in short supply, academics don't have the time or budget to turn good ideas into products, Helt added. "They can't do all those things that come after the big idea," he said.
"Traditionally, the really great ideas come out of academics," Harmon elaborated. "We want to hang on to that." The three are maintaining their ties to academia to remain tapped into innovation. Neomorphic has formal ties with the Drosophila Genome Project, where the curator uses Helt's tools to display genome data over the web.
Tools more interesting than research
As a graduate student, Helt said, he wanted to view the results of several database searches in the same screen. He told BioInform that he started working on visualization tools on the side to help him view search results. "After a while the tools were more interesting, so I focused on those," Helt explained. He said that because his product came about as a solution to a real problem he faced, it is proving useful to other biologists. His tools became the Neomorphic Genome Software Development Kit, a set of five Java-based visualization tools that developers can use to build their own applications. Helt said Neomorphic's Development Kit is designed to be easily integrated into other applications developers might have.
With one application built from three of the tools, a user can view a stretch of chromosome that shows all identified genes in the region. Below it on the screen are results from several different database searches showing homologues lined up beneath the newly identified genes. By selecting a homologue, users can view the score for that match and see any annotations about the gene in the database.
Helt said he hopes individual users will be able to add their own annotations about what did or did not work with each clone so others in the company or academic group don't make the same mistakes. "The people who are the most qualified to annotate regions are the ones who have been working in those regions," he noted. Neo morphic also markets a system to keep track of the entire sequencing process. The product, Labtrack, was first designed by Peter Brokstein at the Drosophila Genome Project.
David Kulp, Neomorphic's first employee and vice-president of research, turned the original tool into a versatile database that monitors a sequencing project from plates in a freezer to sequences in a computer. The Drosophila Genome Center uses a version of the product that is integrated with the lab's bar-coding machine. When a lab tech makes plates, Labtrack prints a set of bar code labels for the plates and tracks where the plate is placed in the freezer. Technicians record each step by bar code number, so when they have the final sequence they can look back on everything that happened to that clone.
Harmon said Labtrack presents all information in a consistent user interface, so lab technicians or lab managers all see the same set of information.
Harmon said Neomorphic will integrate Labtrack into a customer's existing bar coding system or tailor the product to meet a particular lab's needs by adding steps or changing how it records information.
The company's third product, Neomorphic Genie, identifies genes in unknown DNA sequences and compares those predicted genes with mathematical models of gene families. Reese said that for many years, scientists guessed which stretches of DNA contained genes and where the gene began within that region of DNA.
Genie uses these comparisons to identify a gene's 5-prime end and to locate possible promoter regions. Researchers can still get this information without Genie, "but it is very labor intensive," Reese observed.
Harmon said Genie can also handle inaccurate data. "We put a lot of effort into making sure our technology is pretty robust in the face of messy data," Harmon re mark ed. This is important because sequences in published databases aren't always accurate since, accord ing to Reese, older methods for reading gene sequences were prone to errors.
Reese said he hopes that if people can identify problematic data, they can clean it up, making databases more accurate.
Neomorphic is working toward integrating its three products to provide customers with a complete solution. Exactly what direction the company will take depends on what customers want, Harmon said. Some just want to buy a piece of the solution, while others seek an entire functioning system from one vendor. "We will try to provide what a customer wants," he said.
Over time, Harmon, Helt, Reese, and their team hope to provide a complete solution while still selling individual pieces to customers who prefer them. The trio is also considering forming partnerships with other companies to help provide additional pieces to a complete solution.
As part of keeping its options open, Neomorphic has not tied itself to any single platform. The company's visualization tools are all in Java, which a client can view on a Macintosh or PC. Server and database applications are written in C, to compile on any platform. Harmon said most customers use Unix systems, but Neomorphic will write for NT if a customer prefers it. "We try to be platform agnostic," he said, "but right now we don't see people jumping into NT. We see NT as more of a client platform."