PALO ALTO, Calif.--Roche, the multinational pharmaceutical and diagnostics company headquartered in Basel, Switzerland, this month hired Michael Liebman to help it meet an objective set by the company’s head of pharmaceutical research, Jonathan Knowles: To knit together its global computational biology divisions.
Liebman vacated a position as director of pharmaco-genomics and biomedical informatics at Wyeth-Ayerst Research to start his new job November 1. He will serve as both director of computational biology for Roche Bioscience here and as global head of computational genetics at Roche Genetics.
In his global role, Liebman will contribute to the endeavor, initiated a year ago when Roche Genetics was formed, to link the company’s sites and technologies--bioinformatics and clinical data on one level, and information from diagnostics and pharmaceutical therapeutics on another. The job entails overseeing computational genetics activities spanning Roche’s major sites, including in Penzburg, Germany; Basel, Switzerland; Welwyn Garden City, UK; Kamakura, Japan; Nutley, NJ; and at his home office.
With major businesses in therapeutics and diagnostics, Roche is in a unique position to integrate, Liebman said. "The goal of all of this is to look at how science and technology is going to change the practice of medicine, and then the interaction with therapeutics and diagnostics going forward. This includes pharmacogenomics."
Liebman spoke last week with BioInform about his new mandate and about Roche’s bioinformatics infrastructure.
BioInform: Which genomic databases are you using?
Liebman: We have commercial relationships with Affymetrix, CuraGen, DeCode Genetics, Incyte Pharmaceuticals, and the SNP Consortium. With DeCode we’re looking at population genetics and the integration of health records and genealogy in a fairly closed society. With Affymetrix, we’re not only looking at using arrays, but also at how to enhance analysis methods to get the best data possible out of the experiments. We’ve developed databases related to our Affymetrix work. We’re also using public domain databases. We’re working with a number of external academic organizations to put together clinical and tissue data as appropriate.
The real key is not just accessing databases, but to potentially develop the data model that will allow databases to work together efficiently to address questions and issues. That’s really what we’re focusing on, more than just what databases do we access.
Something that comes under my domain is a problem endemic to the pharmaceutical industry: How do we make all of the information work together across the organization to make the pipeline more efficient at the beginning and all along the drug development process?
Bioinform: What is an example of that?
Liebman: A pet area would be understanding how toxicology can be extended from being a filter in the drug development pipeline to feeding back information to the discovery process of successes and failures. This could enhance selection of drugs and targets to make the process more efficient. The end result of all that is to produce better drugs. We’re trying to understand the entire disease process because what you want to do in the future is improve your diagnostics to stratify and stage the disease, indicate the genetic variance of the patient, and then have the appropriate therapeutic occur earlier in the critical processes that need to be modulated. That’s really the goal of what pharmacogenomics is potentially directed towards.
Bioinform: Are you using gene chips from Affymetrix, Incyte, Gene Logic, and CuraGen?
Liebman: I can’t guarantee that at any given time we’re using all of them. I can probably guarantee that we have evaluated each of them for different projects along the way and they continue to be evaluated.
Bioinform: How large is Roche’s computational genetics group?
Liebman: It’s an integration and interfacing of groups, so we’re in the process of hiring. It’s something that was created recently so we’re just staffing up.
Bioinform: How is your staff organized?
Liebman: Initially, we’re staffing as a centralized effort, but potentially we would look at either coordinating or having staff in a more distributed fashion. The first thing we have to do is put together a strategy. With that in place, we can start to figure out the best implementation plan, which is to have the core people centralized and move from there to figure out how best to interact with the different therapeutic areas at different sites. Along with these therapeutic sites, we also have sites in Indianapolis, which is the diagnostics division, as well as Roche Molecular Systems in Alameda, Calif. We’re doing this coordination with diagnostics, so those sites are an important part of it. The other sites I mentioned are primarily therapeutics or pharmaceuticals.
Bioinform: Do you oversee software development?
Liebman: We do software development and research. We’re looking at how the research development area is coordinated with production implementation and trying to work closely with the information technology organization to establish that relationship.
Bioinform: Which third-party software do you use?
Liebman: We use so many different packages across the organization that I’d hate to give you a list that seemed to imply we weren’t using something somewhere. We’re constantly updating that because we’re always looking at new packages. What we’re trying to do is solve a problem and figure out the best way to solve it, as opposed to having a commitment to certain single packages. One direction we hope to go in is towards incorporating the best tools from each of the packages behind a more common interface. That allows the scientists to get their jobs done through a common entryway while having access to a wide range of tools--the best tools available to solve the problem, not just a package that’s commercially prepared.