Dec. 2006 / Jan. 2007 | As coworkers at Silicon Genetics, Saeid Akhtari and Ilya Kupershmidt had a vague idea of producing new software that would allow non-experts to ask sophisticated questions while removing complexity. When Agilent acquired the company in 2004, the pair decided to put their plan into action.
“We sat down [at the] bench at Stanford all day and figured out our approach,” says Kuperschmidt. “Companies build very sophisticated products and expect people to learn how to use them. We want to make software easy for anyone to use.”
To do so, they co-founded NextBio, which this fall officially introduced its platform after a year of beta testing by a handful of select organizations. “Many companies come out and say they intend to do something great,” says Akhtari, but NextBio took a different approach by getting customers under its belt before releasing its product. “We decided first to establish enough users. We’re confident that [the platform] is working and the systems are doing what we hoped, so we could get out and tell the world.”
This new approach to bioinformatics is in essence to provide high-throughput information to researchers without them having to learn anything. Akhtari recalls: “We went to lots of people, outlining the problems. We knew about their pain... they’ve invested a lot in producing [‘omics] information, but not many were tapping into that knowledge base.”
One of NextBio’s primary goals is to enable biologists to query complex information without difficulty. “Other companies build complex systems that can do very sophisticated analysis and allow users to run many different algorithms. Our approach was completely different — we feel that complexity has to be hidden from the user,” says Kupershmidt.
A second distinction, he says, focuses on the information source. “We are not focused on storing or analyzing raw data. There are many companies that do that well. Instead, we focus on study results — information generated by statisticians and computational biologists analyzing high-throughput datasets from diverse platforms,” says Kupershmidt. “This enables us to bypass the initial complexity of dealing with unprocessed data.”
The NextBio platform enables biologists to examine a gene, protein, compound or disease of interest in a global, biological, or clinical context. “The current informatics space doesn’t allow biologists to look at information in a systematic way,” says Kuperschmidt. Take a gene, for example: “What are the key tissues where this gene is significantly expressed? What are the known public and internal compounds that potentially target that gene or influence its expression? What disease states are associated with this gene’s activity? Even these fairly simple queries represent a complex task for today’s biologists.”
Kuperschmidt explains: “If you have a compound, you have to design an entire project to understand what are the potential targets, tissues, etc. to form an in silico exploration of that compound. It’s an information challenge.” He points to the Broad Institute’s Connectivity Map as a good example (see “Compute for the Cure,” Bio-IT World, Nov. 2006).
“It’s a hypothesis generating or hypothesis confirming tool. NextBio provides powerful meta-analysis and query capabilities on top of a knowledgebase that integrates and correlates information from internal projects and public information,” says Kuperschmidt. “From the scientists’ standpoint, you can query that entire space of internal and public information within seconds to better understand the science of a drug target, disease gene, or a clinical phenotype.”
In 2005, NextBio invited some industry and academic groups to serve as beta testers. Feedback came from three constituencies. “The researchers themselves — both scientists and clinicians — form the largest and most important group of users,” says Akhtari. “We understand the types of questions and queries they are interested in. We also work with IT guys and bioinformaticians who have different issues and pain.”
Early users included research and clinical scientists at Scripps Florida, Stanford, Princeton, Yale, the Institute for Systems Biology, and Genentech. “Most customers have a pretty good base of raw data, LIMS and good tools, but when it comes to getting results there’s not a repository to capture that information. One of the attractions for Scripps was to capture that data,” says Akhtari.
Following the Scripps
Nick Tsinoremas is senior director of informatics at Scripps Florida. Aside from his own interests in high-throughput and high-content screening data, Tsinoremas’ chief interest is understanding microarray data to build new signaling pathways. Tsinoremas knew the NextBio founders from his earlier stints at DoubleTwist and Rosetta Inpharmactics.
“It’s a very interesting concept,” says Tsinoremas. “Until now, there’s been no system to store processed data — results essentially — and format it in an organized way. All of the [existing] tools deal with individual experiments, so if you want to [compare] experiments, it’s very difficult.”
By contrast, Tsinoremas says the NextBio platform deals with metadata. He offers an example: “If you do an experiment, you have to decide what is signal and noise. Then, you want to compare these data with results from one month ago. Before you couldn’t, you had to reprocess the data. Now you can do that.”
Another benefit is that NextBio allows users to compare their own data with data in the public domain. “That’s a huge plus — previously it was a custom job that required specialists to download new data, which would live in Excel spreadsheets, or Spotfire. Now, it enables our scientists to ask questions without having to write code or normalize data.”
Tsinoremas says he wants to roll out the software to a larger audience with other applications above and beyond gene expression, such as chemical biology, HTS, and HCS. He has joined the NextBio scientific advisory board to help in that capacity.
Other users include stem cell researchers at the Burnham Institute, where NextBio serves as the knowledge hub. Uploading internal results links the data with other large-scale datasets on stem cells, tissue profiling, and so on. And a biopharma client is studying the activity of various compounds in animal models. The NextBio system was used to upload legacy data, and compare internal data with projects outside compound profiling.
Last October, NextBio announced the general availability of its platform. “We have a very healthy pipeline,” says Akhtari. “We’re being careful that every customer is a reputable account. We don’t want to stretch too thin.” To that end, Akhtari is looking for 6-12 month pilot projects, with groups of 50-100 users. Akhtari says he is not aware of anything that is a direct competitor — his only concern is internal development at major pharmas.
NextBio offers ASP models or can install the software internally. The platform is delivered with pre-loaded and processed public data.
Email Kevin Davies.
Subscribe to Bio-IT World magazine.