Dr. Peter Seufer-Wasserthal, Chief Business Officer
“A multipart AI-based, drug discovery platform, optimising compounds for specificity, efficacy, physiochemical properties, and broad patentability (i.e., novelty), in a parallel process is faster, cheaper and better than serial-step traditional drug discovery,” believe the founders of Origenis, a small molecule drug discovery firm established as a management buyout from a biotechnology company, Morphochem, in 2005. Before the inception of Origenis, Morphochem carried out drug discovery—based on a novel, smart, technology-based approach—as well as drug development processes. With drug candidates in hand, it was decided to fund the development programs and suspend its discovery initiatives. Consequently, the builders of the technology group—Michael Almstetter, Andreas Treml, and Michael Thormann—saw their novel drug discovery approach as an opportunity and co-founded Origenis.
Germany-based Origenis’ mission from day one was to improve and accelerate drug discovery processes to create lead molecules in areas of interest using a unique combination of proprietary in-silico tools for designing compounds (MolMind®
) coupled with immediate synthesis and testing of the designed compounds (MOREsystem®
). A major advantage of these combined technologies is that they are not limited to a number—even a very large number—of compounds that are contained in in-silico libraries, but use a combination of starting materials and chemical reactions. Use of genetic algorithms, machine learning, and other AI tools allow searching through this chemical space to identify compounds that have activity against a target and can be readily synthesised linking in-silico tools to the lab in a unique way. MolMind®
support the design, synthesis and screening of compounds in the Origenis labs with interactive AI feedback loops that can alter and enhance the next cycles of design.
"MOREsystem® is a patented, unique, and highly-integrated technology platform that is supported by a proprietary patent search and analysis tool, Cippix®, to evaluate chemical and biological information about drugs"
Initially, Origenis collaborated with pharma and biotech companies to design and optimise drug compounds with high activity and specificity for different tissues and therapeutic areas. During these collaborations, it became clear that the traditional step-by-step discovery process applied for small molecules leads to high attrition as optimising for any new property is a trade-off of the existing ones. Compounds optimised for activity and specificity often miss the physicochemical property window needed for the targeted therapeutic use. The Origenis team thus studied the properties that a molecule needs to possess in order to impact a specific body part and developed analytical tools allowing the assessment of such properties that only need small quantities of the compound—EYEdeal™ and BRAINstorm™— to be done very early in the discovery process. This allowed a parallel optimisation toward a target property profile, which is confirmed at the start of a program and not only when active compounds are available. This strategy has the properties needed for the planned therapeutic use built into the molecule, resulting in shorter development time to better compounds. The Origenis ophthalmology and CNS portfolios of development candidates reflect the success of approach and platform. Another important element in every discovery project is novelty, which is many a times looked at too late in the process.
Origenis will continue employing interactive AI feedback loops to ensure that the compounds have suitable chemical properties and are improved for biological activity, toxicity, synthesisability, and stability to make them valuable drug candidates
Origenis ensures the originality of the compounds by guiding the process through Cippix®
, a proprietary software that can read and “understand” chemical and biological information from patents in real time. The company compares designed andsynthesised compounds with all patents available, ensuring “freedom-to-operate” for the Origenis programs at every stage of the discovery process. In addition, patent information is used with Cippix®
to gain knowledge about chemistry with abstracting “reaction rules” that define how given combinations or starting materials react under different reaction conditions. Throughout its drug discovery process, Origenis will continue employing interactive AI feedback loops to ensure that the novel compounds have unique chemical properties and are tested for biological activity, toxicity, and stability for real-world application.
Over the past three years, the company has developed a portfolio of compounds inhibiting different kinases that play roles in neurodegenerative diseases such as Parkinson’s and Alzheimer’s. These programs were recently partnered for moving the candidates towards clinical development. Based on this success, Origenis will continue to create portfolios of new chemical entities (NCEs) for clinical development, looking at efficacy, specificity, target tissues, and broad patentability.
Moving beyond the Traditional Approach
Today, artificial intelligence (AI), machine learning, in-silico, and deep learning are the buzzwords in the pharmaceutical space even though such methods are not yet prevalent in the small molecule drug discovery segment, especially not linked to quick realisation in the lab, hence linking the in-silico to the real world. “Although companies have been using in-silico methods for quite some time, they mainly use them to support single steps in the traditional approach instead of a new process that the technologies allow,” states Dr. Peter Seufer-Wasserthal, chief business officer at Origenis. These approaches lack the capability to translate the designs into synthesisable compounds, not just single compounds, but whole libraries. Often, they rely on experienced medicinal chemists due to lack of faith in software systems. Additionally, physicochemical properties are often analysed late in the optimisation process, and thus can lead to losses in activity, specificity, or other areas that have already been optimised. These issues can be avoided if target properties for new candidates are optimised in parallel, saving significant time and money and considerably improving the probability of success in small molecule drug discovery.
Origenis has developed and uses a new way of identifying novel compounds as development candidates. It adopts a proprietary method to study the chemical space and design the principles of drugs, providing unique solutions to solve pharmacological problems. “It is important to understand that the data such as physicochemical properties of small molecules needed to reach a target tissue, even in big pharma companies, is limited and often lacks negative information, which is as valuable as positive results for the process. Hence, to build in-silico models, organisations rely on foreign data and limit themselves to two things—data and its quality in different model domains. But we did not want to be restricted,” says Seufer-Wasserthal.
Origenis followed an unrestrained approach and attempted to collect data from other companies only to find out that it was still not enough. Being aware of this limitation Origenis decided to build a high-throughput measurement system to generate the data needed. With this massive characterisation of compounds in its lab, Origenis was able to populate datasets with meaningful compounds and their properties, which led to improved prediction tools and thus a better compound library design at Origenis.
At present, the chemical space is growing significantly owing to the availability of numerous compounds. Chemists gain a broader understanding and knowledge of the right compounds and ways of identifying them, but this is still limited. “This is the reason we have developed tools that abstract how appropriate compounds are made from which starting materials using which chemical reactions from our electronic lab notebooks, literature, and patents. For instance, if a chemist at Origenis performs a certain chemical reaction, all the information associated with it goes into our system. We analyse the result of that reaction, enabling the system to learn simultaneously and apply the rules in all future programs,” explains Seufer-Wasserthal.
The system turns disparate information of chemical compounds into structured, computer-usable data and defines how the compounds are synthesised from available starting materials, which reaction classes need to be used, how frequently those reactions are used, which combination of starting materials works under given reaction conditions and which don’t, and much more. This helps predict a vast number of synthesisable compounds from chemical spaces that fit the program worked on.
The approach is not limited to target classes worked on
In its collaboration with Expansion Therapeutics, Inc., Origenis develops highly selective RNA-targeted small molecule compounds, an area completely new for the Origenis platform. The companies will collaboratively identify drug development candidates for Expansion’s portfolio of RNA-targeted small molecule medicines. This partnership will provide Origenis with an opportunity to extend its competence further into a unique therapeutic space that will differentiate it significantly from other AI-based drug discovery companies.
The Momentous Milestones
The most recent success of Origenis in drug discovery is associated with the pipeline programs of brain-penetrating kinase inhibitors in neurodegenerative diseases. The company has built these programs by selecting certain targets with the use of AI. “Utilising our patent-reading tool, Cippix®, we identified the targets and used BRAINstorm™ to measure and compare the properties of competitor compounds to establish the physicochemical window needed for the application. The use of our chemistry-driven approach has allowed us to design compounds that are active against the targets and have properties that, although outside of established industry rules, make them promising drug candidates in a field of very high interest,” mentions Seufer-Wasserthal.
Rather than only assisting third-party companies with their drug discovery, Origenis plans to focus on its own pipeline programs. Origenis intends to work as an incubator, in collaboration with partners and investors, using MolMind®, MOREsystem®, Cippix®, BRAINstorm™, and other technologies right from design via chemical synthesis to biological screening in order to create candidates that match specific predefined therapeutic area needs and accelerate further development. The company’s objective is to build new portfolios of drug candidates using novel chemical core structures that can be derivatised for use in multiple programs rather than needing to develop novel chemistry for every new target.
“We, at Origenis, will continue to strengthen our discovery platform and make it more effective, right from selecting targets to developing candidates,” concludes Seufer-Wasserthal.