AI is Reshaping Drug Discovery in the Pandemic Era: Accelerating the Development of Oral Antivirals for COVID-19

Paxlovid, the first Pfizer oral medication for COVID-19 to be approved for use in the United States, has been hailed by the WHO as "the best option so far for treating high-risk patients" and has subsequently been approved for marketing in numerous countries and regions worldwide. On June 7, Pfizer announced an investment of $120 million to significantly boost Paxlovid production in the U.S., with projected sales for 2022 expected to reach $22 billion. The success of Paxlovid is inseparable from its remarkable therapeutic effects and rapid development process. Behind its success lies the acceleration and support provided by XtalPi's AI algorithms, completing in just six weeks what traditionally would take several months.

Under the pandemic, the competition in drug innovation has become increasingly fierce. Has AI subtly created a new normal for drug research and development? Amidst the continuously surging wave of development in the AI pharmaceuticals sector, how is AI actually being implemented in assisting drug R&D? XtalPi, as a representative company in the AI pharmaceuticals field, has frequently announced news of pharmaceutical company collaborations and incubations. VCBeat spoke with two core team members of XtalPi to explore the development trends in the AI pharmaceuticals track.

Paxlovid is a protease inhibitor with the main component being the compound PF-07321332, which exhibits antiviral activity against all known coronaviruses capable of infecting humans. According to previously published results from Phase II/III clinical trials, Paxlovid reduced hospitalizations and deaths by 89% in patients with mild to moderate symptoms who were at high risk of severe illness. In February this year, Paxlovid received emergency review and approval from the National Medical Products Administration and was included in the national medical insurance system in March, playing a significant role on the front lines of the fight against the pandemic.

According to XtalPi, the development of this drug was accelerated by its AI pharmaceutical algorithms. As the world's first solid oral drug to receive FDA approval for marketing, during the research and development process of Paxlovid, the Pfizer team chose to collaborate deeply with XtalPi. By leveraging XtalPi's AI prediction algorithms combined with experimental validation, the traditionally months-long R&D cycle was shortened to six weeks, assisting Pfizer in quickly identifying the optimal crystal form suitable for subsequent drug development and production.

Dr. Zhang Peiyu, Chief Scientific Officer of XtalPi, introduced that XtalPi and Pfizer have maintained a long-term strategic partnership since 2017, which includes a customized small molecule drug simulation algorithm platform for Pfizer. "The Pfizer team has an open attitude towards exploring intelligent R&D methods and industry-leading successful experience. This has enabled our AI prediction methods to be quickly implemented and widely adopted." According to Zhang Peiyu, XtalPi has now established a comprehensive "trinity" drug R&D system that closely integrates intelligent algorithms, automated experiments, and expert experience. In the R&D collaboration on Paxlovid, it was the mutual verification between XtalPi's intelligent algorithms and the wet lab experiments completed by Pfizer scientists, combined with professional interpretation and rapid decision-making from the R&D expert teams in both China and the U.S., that significantly shortened the R&D cycle. This allowed the COVID-19 drug to complete extensive research exceeding industry standards in a shorter time and move into subsequent development phases earlier.

As one of the most favored unicorn companies in the AI pharmaceuticals field by the capital market, XtalPi has raised hundreds of millions of dollars in total financing, leading globally and paving the way for numerous AI pharmaceutical enterprises. However, beneath the company's brilliant "fund-raising" exterior, what capabilities and characteristics have led many pharmaceutical companies both in and outside China to initiate drug discovery collaborations with XtalPi?

Algorithms, computing power, and data are the three core elements of artificial intelligence development. XtalPi, as one of the earliest innovative companies to enter the field of AI drug discovery, proposed the concept of a "digital twin" R&D system in 2018 — enabling the algorithmic world to gradually approach a perfect mapping of the real world. XtalPi announced that it will independently acquire the ability to obtain abundant real-world data for drug research and development through self-built laboratories. Combined with quantum mechanical calculations, this will achieve efficient interactive iteration between experimental exploration and drug simulation algorithms, supporting the continuous innovation and upgrading of its algorithms.

Recently, at the ITBT (Digital Drug R&D) sub-forum of the VB100 Conference hosted by VCBeat and others, Zhang Peiyu stated that XtalPi has established an intelligent, automated drug R&D system integrating smart computing, automated experiments, and expert experience over the past few years. This includes an automated laboratory spanning thousands of square meters. Through a closed-loop of dry and wet experiments, it can significantly reduce experimental costs and time, obtain high-precision experimental data promptly, and provide feedback to the algorithm platform.

"Currently, data has become a key bottleneck in AI-driven drug research and development. We believe that automated experiments will be the solution to this bottleneck, and XtalPi has been making strategic moves and explorations in this field for quite some time," introduced Zhang Peiyu. XtalPi is one of the few AI pharmaceutical companies in the industry that has mastered the independent ability to acquire experimental data and complete experimental validation, providing assurance for the practical implementation and reliable delivery of AI in drug R&D. By combining automated experiments with AI, XtalPi can help scientists overcome certain limitations of traditional R&D, allowing them to more quickly achieve better and more challenging results—those 'high-hanging or hidden fruits.'

"The core of drug development lies in experimental trial and error and exploration. What we are doing is using a large number of high-precision computational predictions to quickly achieve massive trial and error that traditional R&D cannot accomplish, performing virtual screening across a broader chemical space, rapidly focusing on the most promising candidate molecules, reducing experimental needs by up to 90%, thereby improving R&D efficiency. Automated experiments can further reduce human and time costs for experiments and form a closed loop between algorithms and data."

In early April this year, XtalPi announced significant progress in its collaboration with Chia Tai Tianqing on a challenging anti-tumor new drug, achieving key R&D milestones in just 50% of the originally planned time. Underpinned by an intelligent and automated drug discovery platform that integrates expert experience, XtalPi successfully ensured the delivery of R&D outcomes despite the pandemic. Zhang Peiyu predicted that drug R&D will shift from being "labor-intensive" to "computation-intensive" and "automation-intensive," driven by advancements in automation and intelligent tools, offering AI-driven pharmaceutical companies the opportunity to tap into a vast drug R&D market with limited resource investment.

According to statistics from VCBeat, in 2021 alone, XtalPi announced 18 significant collaborations and advancements with 16 pharmaceutical companies, including Jacobio, 3D Medicines, Jingxin Pharmaceutical, Genhouse Biotech, Singleron Biotechnologies, Geode Therapeutics, and PhoreMost. These projects span various therapeutic areas such as oncology, psychiatric disorders, and autoimmune diseases, covering multiple research and development categories ranging from small-molecule innovative drugs to large-molecule monoclonal antibodies, peptides, and ADCs. According to XtalPi, by 2022, the company had served over 140 pharmaceutical clients in China and abroad, with its automated and intelligent platform simultaneously supporting more than 60 new drug discovery projects.

Wen Shuhao, Chairman of XtalPi, introduced that XtalPi's partners can mainly be divided into three categories — international pharmaceutical giants, powerful pharmaceutical companies in China, and innovative biopharmaceutical companies. Foreign large pharmaceutical enterprises have rich R&D experience and pipeline layouts, with clear pain points and direct needs, and pay close attention to the development of innovative companies and cutting-edge "enabling technologies." A search for academic papers on XtalPi reveals collaborations with global pharmaceutical giants such as AbbVie, Johnson & Johnson, and AstraZeneca.

Wen Shuhao introduced that in business cooperation with such enterprises, the process of securing an order from scratch is the most arduous, but also the most valuable. For instance, XtalPi once spent more than half a year passing through the stringent and meticulous data security review of a certain international pharmaceutical company. "Our research and development does not involve patient data related to target discovery, so we only deal with drug-specific R&D data. These data come from public databases, collaborators, and our own experiments and computational accumulations. When we initially built the platform, we gave full consideration to information security, and we also pass authoritative industry audits annually. However, pharmaceutical companies have even stricter requirements than certification bodies." It was these pharmaceutical companies that helped XtalPi refine its algorithms and validate its platform and technology in the early stages. "Once they recognize your technology and team, and trust is established, subsequent progress becomes much smoother," Wen Shuhao explained. As projects are delivered and technologies verified, these pharmaceutical companies become repeat customers of XtalPi, gradually expanding the scale and scope of cooperation.

The second category is China's leading pharmaceutical companies. These enterprises possess strong capabilities and experience in drug development, production, and sales. As national policies encouraging innovative drugs continue to emerge, they face both opportunities and pressures for innovation transformation and acquiring new pipelines. The AI-driven de novo drug discovery capability aligns closely with their strengths in the later stages of R&D, enabling these companies to expand their pipeline sources. XtalPi's "one-stop new drug R&D service" ensures a seamless integration of AI processes with traditional R&D workflows—XtalPi receives targets from pharmaceutical companies and delivers preclinical candidate compounds that have been experimentally validated, allowing the pharmaceutical companies to further complete IND and clinical application processes.

The third category is innovative biotech companies. Wen Shuhao stated that much of the innovation in current drug development comes from innovative biotech firms, including top academic researchers commercializing their scientific findings. These companies need to quickly validate their concepts and obtain compelling milestone data to secure funding and resources to advance their projects towards specialized assets and drug pipeline development. The advantages of AI—low cost, high efficiency, and reliable delivery—align perfectly with their needs. "When intelligent automated R&D tools are placed in the hands of experienced experts, they can deliver even greater value. We place special emphasis on collaborating with truly innovative biotechs to empower innovation at the source of the industry."

In October 2021, Signet Therapeutics, an innovative targeted cancer drug R&D company based on disease models, announced the completion of nearly ten million US dollars in angel+ round financing. In less than a year since its establishment, the company has raised over 100 million yuan in cumulative financing, becoming one of the new batch of fast-growing "small but beautiful" innovative drug companies. It was also announced that XtalPi would be collaborating again, continuing the drug R&D model of "AI drug discovery + disease model platform" to develop an entirely new first-in-class drug pipeline.

Deeply Collaborating with XtalPi, Dr. Haisheng Zhang, Founder and CEO of Signet Therapeutics, told VCBeat that what he values is not only the efficiency improvement in AI-driven drug discovery: It took just over half a year from Signet Therapeutics providing the target to XtalPi delivering the first-in-class candidate compound. According to Dr. Zhang, using traditional drug screening methods, this process would have been delayed by at least one to one and a half years.

The discovery of a new target often gives rise to a batch of innovative drug research and development projects entering the pipeline. To outpace competitors, obtaining molecules with clinical potential earlier is one of the major reasons why an increasing number of pharmaceutical companies are opting for AI-driven drug discovery. However, "speed" is only one aspect; "more importantly, it lies in AI's enhancement of 'quality,'" introduced Zhang Haisheng. Higgs Biotech is researching a novel target for diffuse gastric cancer, whose structure poses significant challenges for traditional drug discovery. Nevertheless, he believes that, "A difficult target isn't necessarily a bad thing. AI is not restricted by existing small molecule libraries, allowing it to create de novo molecular scaffolds. It can also predict and optimize molecular drug-like properties related to clinical performance ahead of time, offering clear advantages in innovation, speed, and molecular performance."

Currently, the first drug developed by Signet Biotech in collaboration with XtalPi is preparing to enter the IND stage. Zhang Haisheng stated that when facing a new target, small Biotech companies often struggle to compete on the same level as well-established Big Pharma. However, AI algorithms have significantly reduced the trial-and-error costs of drug discovery, enabling Biotech firms to achieve milestones quickly despite limited resources. Zhang Haisheng remarked, "The clinical candidate molecules obtained through this powerful collaboration have shown data in trials that are truly encouraging."

"The biopharmaceutical industry has no winter," Wen Shuhao said at the unveiling ceremony of Signet Biotech. "Defeating diseases is an eternal and core pursuit of humanity."

XtalPi is perhaps one of the "most well-funded" companies in the AI pharmaceuticals field. With abundant capital flow and years of accumulated sharp insight into the pharmaceutical industry, this innovative company, highly favored by investors, has also begun "investment banking-style" endeavors. In recent years, it has incubated and invested in a group of domestic and international biotech startups, including Jitai Pharmaceuticals, PhoreMost, Signet Therapeutics, LaMang Bio, and ModaBio. These companies are involved in cutting-edge areas such as organoid target discovery, AI-driven drug delivery and formulation, undruggable targets, and immunometabolism development. Leveraging XtalPi's financing capabilities, they have collectively built a distinctive innovation-driven drug R&D ecosystem.

XtalPi Co-founder and Chairman Shu-hao Wen introduced the internal logic of XtalPi's investment incubation to VCBeat: First, XtalPi’s cooperation with numerous pharmaceutical companies has led to the accumulation of first-hand R&D experience, providing richer data support for project understanding and judgment, allowing them to see "points" that general investment institutions cannot. XtalPi is more likely to help these high-quality projects, which are part of its upstream and downstream industrial chain, obtain the necessary resources at an early stage, enabling rapid growth. Second, XtalPi hopes to empower truly innovative Biotech companies through its self-developed digital drug R&D engine, helping them quickly achieve the transformation of scientific research results. By building a bridge across the "valley of death" in drug R&D, they can focus on rapidly converting scientific achievements into pipeline assets even with limited resources. XtalPi, positioned between target discovery and clinical trials, also brings itself more R&D business opportunities. Third, while leveraging its own AI drug discovery platform to empower projects, XtalPi continuously refines and enhances the capabilities of its AI platform.

"The companies invested in and incubated by XtalPi are basically closely integrated with and highly complementary to XtalPi's business. For example, Signet Therapeutics. Through our initial collaboration, we discovered that its unique cancer disease model can obtain data that is closer to the real drug response within the human body. This allows XtalPi’s AI platform to make more accurate predictions of clinical drug efficacy without involving any real patients. We later decided to invest in Signet Therapeutics and helped it complete a new round of financing." Wen Shuhao stated that the advantages of this "1+1>2" model are very clear, which can help XtalPi打通在药物研发上更多的环节.

In recent years, the development of the AI pharmaceuticals field has been extremely hot, and the "presence" of "AI pharmaceuticals" seems indispensable in various forums related to the development of the pharmaceuticals industry. In the capital market, a weathervane for industrial development, the "heat" in the AI pharmaceuticals field is even more evident. According to incomplete statistics from VCBeat, in the first half of 2021, there were 33 transactions in the global AI + pharmaceuticals track, with a cumulative financing balance of $2.57 billion, already close to the total financing amount in 2020 ($2.78 billion). Based on this trend, it is highly likely that the financing amount for the AI + pharmaceuticals track will reach a new high in 2021.

In this regard, Wen Shuhao stated that behind the heated development of the track lie the underlying demands of industrial development and the breakthroughs in AI technology.

In recent years, large pharmaceutical giants have been facing the same issue—the overall return on investment in drug development is steadily approaching 1%. On the other hand, the biopharmaceutical field has seen vigorous growth, especially in China. With the rapid development of China’s economy, continuous changes in living environments, shifts in people’s health concepts, and the acceleration of aging populations, among other factors, the domestic biopharmaceutical industry is growing rapidly. Whether it’s the expedited review and approval process for innovative drugs, the implementation of the Marketing Authorization Holder system, the new policies under Hong Kong Stock Exchange Rule 18A, or the launch of the STAR Market, both national and local governments are actively encouraging and supporting the development of the biopharmaceutical industry. Although China's biopharmaceutical industry needs further upgrading, the risks associated with developing innovative drugs remain high, and pharmaceutical companies urgently need new solutions that lower trial-and-error costs.

"The pharmaceutical industry urgently needs new technologies to break the current deadlock in drug development. AI is expected to solve this problem and is already in the process of solving it," said Shuhaohao Wen.

Accompanying the underlying demands of industrial development is the breakthrough of AI technology in recent years. "The AlphaFold program launched by Google uses AI technology for protein structure prediction, and many of the new ideas and methods will bring inspiration to the upstream and downstream of the industry. Ten years ago, most structural biologists would have thought it impossible to predict protein structures using artificial intelligence, but AI has not only achieved this but also done it exceptionally well," said Shuhaohao Wen.

According to Wen Shuhao, XtalPi's one-stop AI drug discovery business has surpassed the crystal form research and development of its flagship product, becoming the growth driver of XtalPi.

In response to some skepticism about the AI pharmaceuticals industry, Wen Shuhao stated, "Any new development goes through a process from emergence to acceptance. Much like the early internet, which saw many bubbles, but after several rounds of consolidation, it eventually gave rise to trillion-dollar giant companies in areas like search, social media, and e-commerce once the right application scenarios were identified. The same is true for AI in pharmaceuticals. Its growth won’t happen overnight; there will certainly be a developmental process."

Wen Shuhao believes that AI will play an important, even transformative role in the pharmaceutical industry, moving from local optimization in drug development to global optimization. However, this requires patience from investors and the industry, as well as a more open and inclusive attitude from the industry.

"XtalPi is positioned as an AI industrial platform service company rather than a pharmaceutical enterprise. We hope to make the best use of our physical underlying capabilities, leverage AI, and utilize automation technology to continuously enhance the efficiency of new drug development, thereby fostering the creation of more new drugs that will ultimately benefit patients. AI-driven drug discovery is a difficult but correct path. The biopharmaceutical industry itself is a long-cycle, high-investment sector. As milestones are achieved one by one, an increasing number of pharmaceutical companies are recognizing the value of AI. Drugs like Paxlovid, which were accelerated by AI in their development, have already entered the market. In the coming years, a batch of new drug molecules genuinely designed from scratch by AI will gradually enter the ultimate test of clinical trials," said Shuhaotem Wen.