Receptor.AI “democratizes” automated AI solutions for drug discovery
The Receptor.AI platform automates the major stages of AI-based drug discovery workflow
Artificial Intelligence (AI) in drug discovery is now on a steep rise. A growing number of companies compete to develop new drugs faster, cheaper, and with a much higher success rate by using AI tools at all crucial stages of the drug discovery pipeline.
Most of the players in this quickly expanding market are oriented towards big pharma, which is routinely investing billions into drug development. Such a partnership is tempting not only for startup companies but also for established leaders in the field of AI-based drug development because it provides stable multi-year contracts backed up by the financial resources and infrastructure of the pharmaceutical giants.
As a result, end-to-end AI-based drug discovery services are tailored for large corporate customers. For example, confidentiality and security issues push big pharma to carry out the on-premise installation of the software with long-term on-site maintenance contracts. Somewhat paranoid fear of data leaks in big pharma substantially limits the potential of cloud scaling and requires the usage of complex access rights management and advanced encryption schemes. All this makes such solutions much more expensive and much less universal than their straightforward SaaS equivalents. The excessive price and complexity drive them out of reach for medium biotech and academic science.
Although orientation on big pharma is very convenient and profitable for AI-based drug discovery companies, it largely overlooks the innovative potential of small and medium pharmaceutical companies. In 2020, 133 new drugs were approved by EU and US authorities. Still, only 37% of them were developed by pharmaceutical giants, while 63% originate from small and medium enterprises, often in collaboration with academic research institutions. The usage of AI drug discovery technologies in this sector is still insufficient, and there is a strong demand for solutions which are friendly to small companies and academic scientists.
One of the new players in the field, intending to change this situation and make AI-based drug discovery services equally available for big pharma, medium enterprise, and academic researchers, is Receptor.AI.
The company develops a scalable and modular drug discovery platform based on the synergy of AI, computational chemistry and molecular simulations. Such a platform will satisfy the needs of different categories of potential customers, including those which are currently overlooked by competitors.
The company’s CEO, Alan Nafiiev, explains:
The end-to-end AI-based drug discovery pipeline could be very efficient but also very expensive. Big pharma is concerned about the security of its intellectual property, preferring on-premise solutions, which are especially costly, poorly scalable and less flexible in comparison to SaaS platforms.
In contrast, mid-sized biotech companies prefer to combine cloud solutions with their own resources, employing specific AI-based services at certain stages of their pipeline.
We believe that AI-based drug discovery solutions are essential for the industry and society and have to be available for all categories of potential customers.
Receptor.AI is a British company founded in early 2021 with an R&D department based in Kyiv, Ukraine. Despite the ongoing devastating Russian invasion of Ukraine, the R&D team managed to relocate partially to Europe and continues to operate. The team currently employs 20 experts in machine learning, data science, bioinformatics, medicinal chemistry, biophysics and molecular modelling.
The team’s exceptional professional level and motivation allowed us not only to survive the onset of the catastrophic war in our country but also to increase the pace of development. We launched the first public version of our AI platform for drug design in July 2022 and plan to add more and more features to it in the coming months. — Alan Nafiev says.
Currently, the public version of the platform integrates the ultra-fast high-throughput virtual screening based on AI-assisted assessment of drug-target interactions and prediction of the ADME-Tox parameters of hit candidates using an extensive set of AI models. The company claims that these solutions are on par or even superior to market leaders’ products as far as synthetic benchmarks are concerned. The virtual screening of the large multi-billion chemical databases against one of more than 10K target proteins takes only an hour with the Receptor.AI SaaS platform and costs as little as $10K- $20K.
Due to orientation on several market segments at once, such a solution should be highly modular and customizable to adapt to the needs of a particular customer. An additional requirement for such flexible systems is extensive cross-validation by several independent AI models, which significantly improve the reliability and robustness of obtained results.
Thorough validation of this complex technological stack on real-world drug discovery problems is vital. That is why the company is currently engaged in more than 10 scientific collaborations with research institutions worldwide and several commercial projects with CROs and medium-sized biotech companies.
Such collaboration has already yielded promising results. We have obtained hit compounds in fully automated mode for several challenging proteins and successfully modeled several novel targets. Our average hit rate in these projects is over 10% from the first iteration, which is really encouraging. This gives us confidence that our approach works and is ready to be marketed,
— notes Receptor.AI CSO, Dr. Semen Yesylevskyy.
Currently, each drug discovery project still remains unique and requires a lot of human resources and customization of the tools, despite the power of AI under the hood. The drug discovery workflow itself is far from being automated. Big market players often hide this complexity under their expensive turnkey solutions and opaque long-term service plans. However, diversification of the customer base beyond big pharma requires a much higher level of automation, resulting in a cheaper and less laborious workflow.
The Receptor.AI platform automates the major stages of AI-based workflow, namely data preparation, model training and pipeline construction for drug discovery. There are fully automated modules for “extract-load-transform-analyze” data lifecycle using both structured and unstructured sources, as well as for training, tuning and deploying AI models using the latest deep learning architectures, data analytics and visualization. The model training is focused on the real-world endpoints avoiding the proxy metrics, which commonly lead to the sub-optimal predictive power of AI models.
On top of this, a visual pipeline constructor allows for creating fully customizable AI pipelines for drug discovery. Following the needs of pharma companies, this system can be integrated on-premise to secure not only the data, but also the AI pipelines.
Sergii Starosyla, the CTO of Receptor.AI, mentions 40+ AI methodologies which are currently used, including known and trusted solutions, as well as proprietary developments:
The ultimate goal of Receptor.AI is to develop a fully automated drug discovery platform being driven by an advanced AI and combining the computational discovery of new drug candidates with their experimental validation. The Receptor.AI envisions radical minimization of human intervention at all stages of the drug discovery process in the near future.
The emergence of companies which propose AI-based solutions for small and medium pharmaceutical companies and research collaborations is important for the global development of the healthcare industry. There is no doubt that AI and machine learning will eventually transform healthcare on different levels, including drug discovery. Still, the pace of these changes depends on the availability of these technologies. Small and medium enterprises are currently the most productive and innovative players in the drug discovery field. It is important to give them access to the whole spectrum of modern AI technologies to make the world healthier and happier.
