The merger of Structural Bioinformatics and GeneFormatics to create Cengent, finalized last week, creates a new structure-based drug discovery company with around 70 employees, at least two years’ worth of cash, an experimental platform combining both NMR and X-ray crystallography, and an expanded portfolio of computational proteomics tools that will accelerate the company’s discovery activities, according to VP and CSO Kal Ramnarayan.
Although bioinformatics comprised the largest area of redundancy between the two firms — more than half of the 40-50 employees laid off in the wake of the merger were computational scientists and engineers — Ramnarayan said that the overlap was actually minimal from a technology standpoint. A number of proprietary computational tools from GeneFormatics will be integrated with the Augmented Homology Modeling platform Ramnarayan oversaw in his previous position as CSO of Structural Bioinformatics.
Cengent claims that Augmented Homology Modeling creates models that are under 2 angstroms root-mean-square deviation from a corresponding X-ray structure — compared to about 5 angstroms for conventional homology modeling methods. However, Ramnarayan noted, the approach starts with a template protein that must have “at least 20-25 percent sequence homology with a known 3D structure of a protein” — a requirement that has limited the number of proteins SBI could model. This drawback will be addressed by an algorithm from GeneFormatics called Prospector that uses threading and fold-recognition methodologies to provide a putative structure assignment that is not very accurate, “but it’s a good starting point for the Augmented Homology Modeling method,” Ramnarayan said. Cengent plans to use Prospector for low-sequence-homology proteins, and then improve the quality using SBI’s method. “That way we are able to broaden the scope of protein models that we can cover,” Ramnarayan said.
Also new to the SBI stable of tools is the Fuzzy Functional Form methodology. The approach is essentially a dictionary of 3D functional motifs for enzymes that have been solved by X-ray crystallography. Once a protein structure is predicted computationally, the Fuzzy Functional Form library helps researchers assign function based on its structural elements. GeneFormatics is additionally contributing a similarity-based searching technology for small molecules that Ramnarayan said will be useful for finding molecules that have similar chemical characteristics — and potentially better biological activity — compared to a lead molecule. Ramnarayan estimated that the new technologies would expand the modeling capacity of SBI’s technology alone by 20-25 percent.
Making it Work Together
Cengent has already begun integrating its computational platforms, and has initiated a proof-of-concept project using a novel phosphatase that it modeled using the combined approach. The company plans to generate several lead molecules using small-molecule virtual screening within two to three weeks, then go into biological testing, and have the whole project completed within the next 60 days, Ramnarayan said.
Cengent plans to integrate the databases of the two firms that gave it life, but the details of that process “will depend on future business,” Ramnarayan said. Currently, SBI offers a relational database called StructureBank that it provides in three forms: a standalone “empty” repository; with a curated version of the Protein Data Bank; or with any number of 6,000 novel 3D protein models that it formerly marketed as a separate database called ProMax. GeneFormatics brings to the table a database of protein sequences for nearly 100,000 human sequences and sequences from nine other genomes. This information could be combined with StructureBank to create species-specific versions, Ramnarayan said, or the company may use GeneFormatics’ Fuzzy Functional Form library to annotate the SBI models.
Right now, Ramnarayan said, Cengent’s executives are evaluating “what new products and services we can offer to increase our chances of making more revenues.” While the details of the company’s future offerings may not be entirely clear, one thing is certain: Cengent will continue to move in the direction of pre-clinical drug discovery. While the ratio of computational work to experimental work at the newly merged firm is around 40/60 now, Ramnarayan said the computational side of the equation will likely drop as the company “transitions more to a traditional drug discovery company.” The computational technology will remain a key part of the company’s service offering in the future, but “we’ll have to enhance the number of chemists and biologists that we have on staff,” he said.
— BT and JK