PharmaDesign, a Tokyo-based drug-discovery firm with a focus on bioinformatics, plans to jump into the commercial software market this fall with the launch of POODLE (prediction of order and disorder by machine learning), a software package for predicting unfolded regions in proteins.
A free version of the software, developed in collaboration with the Computational Biology Research Center at Japan’s National Institute of Advanced Industrial Science and Technology, was launched this week at the CBRC website (http://mbs.cbrc.jp/poodle).
The commercial version, which is expected to be of interest to pharmaceutical firms that do not want to submit their protein sequences to a public server, will be available in the fall, the company said. Pricing has yet to be determined.
PharmaDesign has a three-pronged business model built upon a suite of in-house bioinformatics software. The firm uses these tools to develop libraries of predicted peptides and small molecules, which it markets under the brands PharmaGPep and PharmaGChem, respectively. In addition, the company offers contract research services based on its bioinformatics expertise, and also maintains an in-house drug-discovery effort.
The company serves as the Japanese distributor for Inpharmatica’s bioinformatics software products, but historically has not had much of an interest in selling its own software. At one time, PharmaDesign marketed a SNP-annotation package called PharmaSNP, but now uses it solely for in-house projects, Satoru Kanai, director, chief scientific officer, and head of R&D, told BioInform via e-mail this week.
Nevertheless, the company thinks that there is strong demand among pharmaceutical firms for POODLE. Many proteins that contain unfolded regions are involved in gene transcription and DNA binding, and are therefore of great functional interest. However, it is nearly impossible to crystallize proteins with these regions, so it has been very difficult to study them.
According to PharmaDesign, accurately predicting these regions will enable scientists to design and synthesize proteins without disordered regions, making them easier to crystallize.
The firm is currently collaborating with an experimental lab to validate this approach, Kanai said. The company has validated POODLE with data from the sixth Critical Assessment of Techniques for Protein Structure Prediction experiment, he said, noting that “in many cases, the results were better than any other programs.”
PharmaDesign is also participating in CASP 7, to be held Nov. 26-30, and Kanai said that the firm’s self-evaluation results so far have been “satisfactory.”
Disorderly Conduct
POODLE’s emphasis on disordered regions distinguishes it from many other protein structure prediction packages, Kanai said, though he acknowledged that there are other programs, such as DISOPRED2 from University College London and PONDR (Predictors of Natural Disordered Regions) from Molecular Kinetics, that offer similar capabilities.
|
According to PharmaDesign, the accurate prediction of these regions will enable scientists to design and synthesize proteins without disordered regions in order to make them easier to crystallize. |
PharmaDesign and CBRC began developing the software last April, Tamotsu Noguchi, leader of CBRC’s protein function team, told BioInform via e-mail. He said that he had previously been a senior researcher at PharmaDesign, and the collaboration grew out of that relationship.
POODLE actually contains three separate programs: POODLE-L, which predicts long disordered regions; POODLE-S, for short disordered regions; and POODLE-W, which predicts whether an entire protein is wholly disordered. PharmaDesign primarily contributed to POODLE-L, while CBRC concentrated on POODLE-S and POODLE-W, Kanai said.
PharmaDesign will initially distribute POODLE-L, Kanai said, and reserve the other two versions for its contract research business in order to provide “more precise results” for customers.
The free version of the software is currently running on Linux, but Kanai said that the commercial version will also run on Windows. He said that the firm is targeting pharmaceutical companies for the software, so the security on the free site is “insufficient” for these customers who would rather install and run the software behind a firewall.