Can drug design replace drug discovery? Current JLABS resident Ordaos thinks so and explains how

Tell us what challenge Ordaōs is trying to solve and how?

Ultimately, Ordaos is trying to find new ways to enhance the speed and efficiency of bringing safe and effective treatments to patients in need. And how? While monoclonal antibodies and the larger proteins have had success and have been approved for treatment across a wide variety of diseases, they do have their limitations. The emergence of mini-proteins is addressing these shortcomings and offers some advantages over larger proteins. For instance, mini-proteins can deliver the power and performance of antibodies, while being more stable, configurable, and easier to manufacture. Their size is actually their greatest strength; typically twenty times smaller than antibodies. They can thus go where other antibodies cannot. These smaller packages also have a tightly packed internal core that improves thermostability and solubility, while minimizing immunogenicity. Given this, mini-proteins provide an effective alternative to the biologic standard of care. They have the potential to make what was once undruggable, druggable. They are also faster to make, easier to produce, and if done right, a lot less costly. And because we design rather than discover them, they can meet the specific target product profiles as specified by the drug hunter. These advantages help Ordaos to de-risk the drug development process, with the aim of getting treatments to patients that are truly effective. In short, Ordaos is in many ways a generative mini-protein asset foundry that is not replacing monoclonal antibodies but rather offering an effective alternative and complement.

Where did the name ‘Ordaōs’ come from?

Our novel technology, the Ordaōs Design Engine, leverages multitask meta-learning and continuous learning loops to translate human-targeted product criteria into machine-designed mini-proteins. It is thus befitting that the Ordaōs name evolved out of the concept that we bring Order to the Chaos of drug discovery and development. This name reflects who we are and what our mission is.

We see several dimensions for Artificial Intelligence (AI) to create value in drug discovery, including greater productivity (faster speed and/or lower cost), broader molecular diversity and improved chances of clinical success. What is Ordaōs’ approach to the drug discovery process? How is it unique compared to other AI/Machine Learning companies?

Ordaōs uses the Ordaōs Design Engine to deliver true protein property design. Starting with amino acids, the Design Engine generates, appraises, and ranks billions of protein sequences and hundreds of thousands of protein structures and properties to create customized miniPRO™ proteins. These proteins are then rapidly evaluated in silico to provide intelligent feedback on multiple design objectives including protein structure, binding specificity and affinity, solubility, stability, immunogenicity, and developability, followed by in vitro validations. This iterative process delivers optimized mini-proteins to meet the partner’s specific molecular target product profile (mTPP). Using this approach, the Ordaōs Design Engine creates more ideal, unseen protein leads than others and can accelerate drug candidate development, increasing the probability of more therapeutically effective candidates.
More simply put, other companies start with words where Ordaōs starts with letters. This allows us to find more protein candidates and ensures that mini-PRO ™ proteins has the ideal physiochemical properties and are better customized.

When you say that future therapeutic advances may come from design rather than discovery, how will the design be applied? Are there therapeutic areas where this approach is more (or less) applicable?

Seeking the desirable requires more than a search through the known. Protein drug hunters today search through vast databases of existing molecules to find a starting point that is “good enough.” The result is molecules riddled with side effects and limited impact. With discovery we are limited by what we can see, what already exists. With design, we are limited only by our imagination. When it comes to the applicability: we started with aggressive cancers that have high unmet needs, including triple-negative breast cancer, esophageal cancer, and AML, and currently have five programs in total in oncology. We aim to widely serve overlooked patients with limited treatment options, and we seek to achieve that goal through partnerships that are currently rapidly developing.

Since you can generate lead assets very quickly and amass a large portfolio of protein binders in many therapeutic areas, how do you prioritize which projects to advance?

Because the Design Engine allows for extremely effective, highly repeatable and rapid results, we have the capacity to run projects concurrently; this means we can help partners address many complex projects at once. For that reason, we haven’t needed to prioritize one project over another and we can truly expand the reach of this amazing science. Any prioritization is based on the needs of our partners and we work with them to deliver on their requirements using our technology to generate, appraise, and rank hundreds of thousands of proteins on their structures and drug-like properties for novel miniPRO™ candidates. Ultimately, Ordaos is most interested in working on challenging targets with equally interesting scientists!

What has been an intractable drug target thus far that you would like Ordaos to successfully develop a therapeutic against?

We recently partnered with Yatiri Bio to create new therapeutics for two novel targets in acute myeloid leukemia. Yatiri Bio has a base of global proteomics from AML patient samples and matched model systems to identify two cell surface therapeutic targets for high-risk AML. Ordaōs is using its proprietary Ordaōs Design Engine to create a panel of in silico de novo protein candidates and will be responsible for in vitro validation of those leads. While this will not be easy, recent advances in machine learning, mass spectrometry, bioinformatics, and ex-vivo model systems have the potential to drive a new era of precision oncology for AML patients and the development of therapeutics specifically for relevant patient subsets. We are really excited for the promise of our engagement with Yatiri for that reason.

How do you see machine learning evolving and advancing drug discovery?

AI and machine-driven systems provide the biopharma industry with the promise to increase their speed and efficacy, while mitigating risks as far down the pipeline as patient side effects. In the area Ordaos works in, our machine-driven technology allows us to leverage multitask meta-learning and continuous learning loops to translate human-targeted product criteria into machine-designed mini-proteins at an accelerated pace. Our human-enabled machine learning (ML) system can generate, appraise, and rank large amounts of protein sequences, structures and properties to create and deliver optimized mini-proteins easier, faster, and more customized than if we used human talent alone. The introduction of AI/ML has truly transformed the way we conduct drug discovery and development and we expect that it will continue to evolve, improve and expand its applications in the biopharma industry and beyond. We believe that this evolution will ultimately lead to nothing less than a true paradigm shift in how early design, no longer discovery, is done. We see the industry moving to a world of design, to a world of customized creation. Our mission at Ordaōs is to define the design standard.

You announced a couple collaborations last year (congrats!). What type of partnerships are you looking for going forward?

Ordaōs is focusing on difficult-to-drug targets in our internal R&D that serve the overlooked patient population. To date, besides developing in-house its proprietary AI/ML-driven Design Engine, the company has benchmarked using the HER2 protein target in a partnership with Vyant Bio, Inc. and conducted a COVID-19 program with Twist Bioscience Corporation to discover and optimize nanobodies. In addition, in September of last year, Ordaōs announced a third partnership to design therapeutics against three never-before-seen targets with specific product profiles with NonExomics, a Massachusetts biotech focused on determining whether novel proteins from the ‘dark’ genome have therapeutic or diagnostic utility.

What are the current challenges you face as a startup company?

We are growing fast and furious and we hope we can recruit the type of talents excited by this relatively new and emerging area of drug development – the creation of mini-proteins. What keeps me up at night is the dichotomy of the speed and virality of progress witnessed this year in machine-driven tech-bio versus the severe contraction in biotech funding and dealmaking. It is clear that with the right support and the right timing, the technology is at the inflection point between pipe dream and true, value-added reality. We hope that investors will see the positive outlook and want to invest in this intelligent reality.

How are you hoping JLABS can help you accelerate?

We are thrilled to be working with Johnson & Johnson Innovation as a JLABS resident. JLABS offers us the support we need to grow and optimize our research and development. They also provide us with their industry connections, access to their entrepreneurial programs and the resources and lab space to support our upcoming in-house wet lab. With their years of experience, they are able to provide us with wisdom and know-how that can help us catapult over what could be major challenges. We feel privileged to be a resident and thankful that we have this opportunity to scale our science to meet this moment in our company’s evolution.