Canadian informatics firm Cyclica is making good on a promise it made last year to make a software-as-a-service version of Ligand Express, its proprietary software for analyzing and evaluating drug-protein interactions, available to the pharmaceutical market.
The company plans to begin beta testing the software product in the fourth quarter of this year with existing clients with a full launch scheduled for the first half of 2017, Naheed Kurji, Cyclica's President and CEO, told GenomeWeb this week. These clients will have a chance to test the solution at a reduced price and to provide feedback that will be used to improve it ahead of the launch. The company has also added new features to its software including a tool for determining the effects of drug compounds on protein activity.
“The real value out of all of this is [customers] can keep the data in house, they don't have to disclose anything to us,” Steven Molinski, an application scientist at Cyclica, told GenomeWeb. Equally importantly “is the visualization and the ability to really tailor the analysis to their needs,” he said.
Historically, Cyclica has provided services based on Ligand Express to customers seeking fresh candidates for their drug development pipelines. The software uses biological, chemical, and protein structure data to identify small molecule candidates for possible development and testing including opportunities for drug repurposing. Given a list of compounds, it looks at known interactions between the test compounds and proteins as well as interactions between similar compounds and proteins. It selects drug-protein interactions that are more likely than others and uses them as a template to explore much larger sets of proteins for possible interactions. As it finds new interactions, the software selects those that will likely be good candidates for research and development. Cyclica recently published a validation study that showcases the efficacy of the software's proteome-docking capabilities using experimental data from kinase inhibitors.
The company targets small- to mid-cap pharmaceutical companies as well as academic institutions. These clients submit lists of molecules to Cyclica which performs the computational analysis internally and reports the results back to the client. But now it wants to make those internal pipelines available to customers so that they can search for those compounds on their own. Kurji told GenomeWeb last year that the company would offer the software under an enterprise licensing model and that it would also provide some additional speciality services to customers.
The company planned to launch the software as early as Q4 this year but decided to hold off on doing so in order to lay some additional groundwork and make some improvements, Kurji told GenomeWeb. For one thing, it wanted to make sure that the platform was secure and that there would be no data leaks once it was accessed. To that end, it tapped a Canadian data security company to assess Cyclica's security mechanisms, bridge gaps, and put all the necessary components in place to ensure that hosted customer data would be safe. “When we agreed to take those necessary steps, we realized that's going to be a six-to-eight month process,” he said.
The company also wanted to ensure that its intellectual property was protected prior to the launch, he said. At the end of last year, it filed an international patent for the core Ligand Express technology that covers the algorithms and scripts that underlie the software. They also wanted to include newer applications that they were developing in the final product. If they had launched the software earlier, they would not have been able to include those features in the initial release, he said.
In the last year, the company has added two new capabilities to its platform, which it says were developed in response to customers' requests for more information to be included in their results. Historically, “what we've been doing is proteome docking,” Kurji explained. “We can dock…a drug or ligand to every protein in the human body [as well as] bacterial and viral proteins. It's a lock-and-key approach where the drug is the key and we are trying to find the best lock.” The output is a list of candidate proteins for a given drug and associated docking scores. Those scores are useful but customers wanted more information, specifically, about the consequences of the drug-protein interactions. “This is really important because certain diseases require inhibition of proteins that are overexpressed whereas [in] other diseases [we need] to turn on a protein and make it work better than it did before,” Kurji noted.
To that end, Cyclica developed a tool called Switchx, which uses machine learning techniques to predict whether a protein will be turned on or off when it interacts with a given drug compound. Specifically, Switchx predicts the most likely category for a compound — inhibitor or activator — by looking at how all drugs have interacted with the protein, Molinski explained to GenomeWeb. It makes an initial prediction based on this data and then refines that prediction based on information about the binding properties of the interaction and the consequences of those interactions, he said. According to Cyclica, it can take four to five months for scientists to run standard assays to determine the effects of drugs on proteins. With Switchx clients get results in minutes.
A second application that Cyclica has added to Ligand Express lets users identify alternative compounds that behave in similar ways to the drug molecules that they are interested in. The new tool that Cyclica has developed lets users generate proteome-binding profiles for compounds and then compare them to one another to identify those that are similar, Molinski said. It offers an alternative to traditional pharmacophore-based searching or mapping which identifiers biosimilars by looking at chemical structures, he said. “Instead of looking at just the structure of the drugs, we can see how each drug interacts with every protein in the human body and find similar drugs which act in that way.”
Other applications available in Cyclica's software include a tool called Divex which can be used to predict combinations of drugs that can be used to treat complex diseases such as AIDS or cancer.
With these tools “we can really look at the whole picture [of] how a drug is interacting with every protein and make some conclusions on how this can streamline or facilitate future drug discovery or drug repurposing and development efforts,” Molinski said. “We can prioritize [compounds] based on mechanism of therapeutic action … we can provide you with a list of every protein that your new drug interacts with [and] we can tell you how it will interact with these proteins.” Furthermore, “we can further enrich the data by using systems biology to map these interactions on different disease pathways and processes,” he said.
Cyclica has completed a number of projects in recent months one of which was submitted by Naturalis, a Canadian nutraceutical company that tapped Cyclica to find natural molecules that are similar to known anesthetics that could be used to treat acute alcoholic hangovers. “The results that Cyclica have provided us have opened completely new opportunities of R&D for my company, and have saved us almost a year of R&D — and with that a lot of money,” Arun Anand, Naturalis co- founder and president, said in a statement. He also told GenomeWeb that he and a colleague have submitted a paper for peer-reviewed publication that describes compounds that Cyclica's software identified as likely candidates. Naturalis is now exploring partnership opportunities with pharmaceutical companies to develop compounds identified by Cyclica's analysis into commercial products.
Another client is a non-profit organization called Cures Within Reach that supports research efforts to repurpose US Food and Drug Administration-approved drugs and devices to provide safe and affordable treatments for common and rare disorders that currently do not have effective treatments. Cyclica worked on a pilot project with Cures Within Reach and Tulane University that aimed to asses five currently approved drugs for possible use in treating idiopathic pulmonary fibrosis (IPF), a chronic and fatal lung disease that affects the elderly. Cyclica used its software to evaluate the interactions between the drug compounds and proteins and to assess the effects of the drugs on proteins. It identified three potential candidates that could be potentially work as treatments for IPF.
Kurji said that the researchers have since validated the compounds in vitro and are preparing to publish a paper describing their results. The partners are now working together to see how Cyclica's software can be used with the Cures Within Reach network to move forward with multiple undisclosed projects, Kurji said. GenomeWeb reached out to one of the Tulane researchers for additional details about the past project but did not receive a response as of press time.
Cyclica is currently in talks with 14 other unnamed companies that are interested in running their compounds through its platform. The company is also working on another agreement with a research group at the University of Toronto that is working on a small molecule therapeutic for X- linked myotubular myopathy, a genetic muscular disease that affects 1 in 50,000 males and is always fatal. It also has projects with researchers at other institutions who are studying rare genetic conditions among other conditions.
“One of our unique value propositions is that we are able to address the rare orphan disease space … [but] we have multiple focus points,” Kurji said. For example, the company is working with at least one pharmaceutical company based in Canada that is developing a new cancer therapy. “Whether its target identification, reprioritization, drug repurposing or adverse effect elucidation, that's our sweet spot,” he said. Competition comes from older firms such as NuMedii and newer players like TwoXar who are also building a business around using computational tools to match compounds to diseases.
But Kurji believes that one of the things that sets Cyclica's solution apart is that it takes advantage of a much wider pool of data than its competitor's offerings. Some of those solutions select compounds by looking only at known and well-supported disease targets. “Once we get outside those [roughly] 1,500 disease targets to the orphan diseases or the rare diseases, machine learning technologies typically have a harder time because the quantum relevance and the quality of the datasets that they are looking at taper off,” he said. Cyclica on the other hand provides a “proteome- wide view of how a drug interacts with all aspects of human biology.”
Beyond the life sciences, Cyclica has accepted projects from the consumer packaged goods market. One client in that space tapped Cyclica to identify a compound in a skin cream product that caused some adverse side effects in a subset of the population. Cyclica was able to identify the offending compound in the product and has signed a longer term agreement with the unnamed company, Kurji said.
Although it plans to offer Ligand Express under a subscription model, Cyclica will continue to take on projects for companies who prefer to access services as needed rather than purchase software. For first-time users, the company offers an introductory price of roughly $25,000 for the first five compounds. After that, it charges between $10,000 and $20,000 per compound, depending on factors such as the size of the compound and the nature of the project, but those numbers are flexible. “We have a framework in place but we are nimble enough that we are able to be amenable to the needs of that customer,” Kurji said.
Following the launch of the software next year, Cyclica expects that initially, about 25 percent of its revenues will come from sales of its SaaS, while 75 percent will come from projects; however, it expects those numbers to flip by 2019 with about 75 percent of revenues coming from software sales and 25 percent coming from projects, Kurji said.
To date, Cyclica has raised about C$3.1 million — about $2.4 million — from investors. The company plans to raise between C$5 million to C$7 million in a new round planned for the fourth quarter of this year, Kurji said.
It has also hired a number of application scientists, including Molinski, whose role, in part, is to test components of the product internally before the company rolls them out and to run analyses for customers who submit projects and transition them to the subscription-based version of software. The company has also hired Andreas Windemuth as senior vice president and chief scientist who will be responsible for current and future development of Ligand Express.
Cyclica plans to open a new office in California in the near future. This is in addition to its headquarters in Toronto and existing offices in Connecticut and Massachusetts.