Abstract
Many technological sectors underwent recent exponential growth because of digital disruption, a phenomenon Peter Diamantis characterized as the ‘Six Ds of Exponentials’: digitization, deception, disruption, demonetization, dematerialization, and democratization. In contrast, drug discovery has been marked by rising costs and modest growth, if any, of annual drug approvals. We argue that the exponential growth of drug discovery can be also achieved through digital disruption brought by data expansion, mature artificial intelligence (AI), automation of experiments, public–private partnerships, and open science. We detected the emergence of all ‘Six Ds of Exponentials’ within modern drug discovery and discuss how each of the ‘Six Ds’ can further empower the field and forcefully address the societal demand for novel, potent, affordable, and accessible medicines.
Keywords: drug discovery, digital disruption, AI-driven methods, open science, Six Ds of Exponentials
Graphical Abstract
Rapid acceleration of the current drug discovery pipeline requires digital disruption catalyzed by the ‘Six Ds of Exponentials.’

Introduction
It is generally recognized that, despite many innovative technological advances and a growing demand for new therapeutics, drug development continues to be slow and expensive.1 Conversely, it is also widely accepted that in the last few decades many technological sectors have experienced explosive growth, with new useful products produced in higher quantities and more affordably.2 This phenomenon, commonly regarded as an example of Moore’s law, is typically illustrated by the evolution of the computer industry, where processors experienced explosive growth in the number of transistors present in an integrated circuit over time while the manufacturing costs exponentially decreased.3,4 Another example of Moore’s law is provided by the evolution of graphics processing unit (GPU) performance per dollar over time (Figure 1), which had a transformative effect on drug discovery through GPU computing and deep learning (DL) algorithms, as we discussed recently.5
FIGURE 1.

Evolution of GPU performance per dollar began to follow Moore’s Law in 2021. GPU performance measured as floating-point operations/second (FLOP/s). Source of data: https://ourworldindata.org/grapher/gpu-price-performance.
In contrast to the computer industry, drug development has become increasingly more expensive over the years, with the cost doubling every 9 years and currently standing at a staggering ~2.6 billion dollars per drug.6 Yet, the number of approvals for new therapeutics has grown very modestly in the last 50 years.7 Together, these observations imply a longstanding decreasing trend in the number of drugs approved per billion USD spent on research and development (Figure 2). This disappointing trend was coined ‘Eroom’s law’ (reversed from ‘Moore’) by Scannell et al.8,9 One grim implication of ‘Eroom’s’ law is that with approximately 10 000 known diseases and only ~2500 approved drugs in circulation, as many as 7500 human conditions remain without treatment.10,11 Arguably, the continuing digitization in subsequent years might have been responsible for gaining the momentum toward breaking Eroom’s law, as was noted in a recent study12 (Figure 2). However, considering that it takes 10–15 years to develop a new medication,13 with the current rate of drug discovery being ~50 new drugs per year at best,14 human civilization would need hundreds of years to develop every essential drug to treat all known diseases. Notably, this estimate does not even consider the emergence of novel health threats like severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). This observation alone suggests that there is an urgent need for technological breakthroughs that can dramatically accelerate the discovery of novel medications.
FIGURE 2.

The evolution of pharmaceutical research and development cost over time. The trend shows a historical decline in the total number of new drugs approved by the FDA per year, per billion USD. Data within the black box indicate the potential emergence of a break in this trend.12
This striking contrast between the continuing growth of technological sectors such as semiconductor or GPU manufacturing (Figure 1) and the lack thereof in the pharmaceutical industry (Figure 2) begs the question of whether the former sectors could offer some insights that could enable the transformation of the latter. The sustainable, exponential growth of many industries relying on the use of computers has been explained in the ground-breaking book Bold by Kotler and Diamantis.15 The authors observed that technologies that experience exponential growth typically go through six phases, defined as the ‘Six Ds of Exponentials’: digitization, deception, disruption, demonetization, dematerialization, and democratization. Given the tremendous impact of the pharmaceutical industry on the wellbeing of mankind, and the expressed need for the transformative acceleration of drug discovery, we find it timely to explore the ‘Six-Ds’ paradigm as it applies to the pharmaceutical industry.
In this contribution, we analyze current or emerging impacts of each individual component of the ‘Six Ds’ paradigm on drug discovery. We observe that although computers have been a prominent and ever-growing component of drug discovery for decades, it might now be time to review and rethink the computerization of the pharmaceutical industry through digital disruption, guided by the principles of the ‘Six Ds of Exponentials.’ We argue that conscientiously bringing ‘Six Ds’ factors into the current drug development workflow will have a catalytic effect on the pharmaceutical industry. We reason that the transformative and actionable realization of these key principles might reverse Eroom’s law, enable exponential growth in the number of approved medications and continuously increase the number of treatable diseases, with a concordant reduction in the cost of therapeutics development.
Relevance of the ‘Six Ds of Exponentials’ paradigm to drug discovery
A technology is defined as exponential when its performance doubles over a specific period of time while its cost is concurrently reduced by half.16 There are many remarkable examples of such technologies, including the aforementioned GPU computing (Figure 1), 3D printing (the cost of which has reduced 400-fold in 7 years), solar panels (cost reduction of 200-fold in 20 years), and drones (cost reduction of 42-fold in 6 years), among many others.17 One familiar life science example is the cost of human genome sequencing, which went from $100 million to a few hundred dollars in just 20 years.17 We have also seen a quick rise in large language models (LLMs). In 2016, Amazon hosted a challenge where developers were offered 1 million dollars for a chat-bot that could hold a conversation for 20 minutes.18 Just less than 6 years later, ChatGPT amassed 1 million users within 5 days of its launch.19 These and many other examples (social media, cryptocurrencies, smartphones, digital cameras, autonomous vehicles, and more20) have been enabled by massive application of digital and digitized technologies that rapidly evolved in a manner that can be described by the ‘Six Ds of Exponentials.’ Below we go step-by-step through how similar technological acceleration, coupled with significant cost reduction, can be achieved in drug discovery by adhering to those principles.
In describing the impact of the ‘Six Ds of Exponentials’ on drug discovery for each of the six components of this paradigm below, we realize that we might not see, at least initially, a dramatic acceleration of the drug approval timeline as this process is highly regulated and controlled by agencies such as the FDA. Therefore most of the discussion and examples illustrating the emergence of the ‘Six Ds of Exponentials’ in drug discovery discussed below relate to early-phase studies in which the target and respective assay have been established. Our key objective here was to introduce this concept, which has been shown to have transformative potential in other technological sectors. We hope that readers involved in all phases of the drug development pipeline will find this paper stimulating and continue to explore it. Thus we expect that eventually, digital transformation will significantly increase the number of drug candidates entering and successfully passing through the entire drug research and development pipeline. Recent analysis suggests that preclinical drug candidates developed with a substantial contribution from artificial intelligence (AI) tools show a success rate of up to 90% in Phase I clinical trials, much higher than historical rates.21 This increase will eventually lead to a substantial increase in the number of approved drugs. Furthermore, we expect that digitizing the testing technologies used in drug development, as well as gradual transformation of regulatory science toward the replacement, reduction, and refinement (the ‘3Rs’) of animal testing using so-called NAMs (new approach methods), will lower the cost and accelerate the process of drug approval.22
Digitization
This first component of the ‘Six Ds’ paradigm implies the transfer of physical information into digital form. When information is digitized, it can be readily stored, processed, distributed, accessed, and modeled, enabling the exponential growth of data-rich technologies.
Digital transformation in health sciences has begun; indeed, it began a long time ago. Notable examples include digital health (digitalized clinical records, insurance records, disease registries, and more23), biological sequence databases (GenBank24, RefSeq25), and protein structure [Protein Data Bank (PDB)26], chemical structure (ZINC,27 Enamine28), and chemical bioactivity databases (PubChem29, ChEMBL30). The growth and utility of chemical bioactivity databases has been enabled by revolutionary algorithms for simple and compact molecular digitization (e.g. SMILES and SMARTS). Furthermore, we underscore the extreme importance of data curation, as data quality remains a major hurdle for building reliable and informative computational models to support drug discovery.
The digitization of chemistry, biology and other life sciences enabled efficient integration and analysis of heterogeneous data. Accessing these digital records enables the application of multiple machine learning (ML) and AI algorithms, recently including LLMs for trend analysis and knowledge mining. Some of these models are already beginning to supersede humans in certain capacities such as image recognition, reading comprehension and language understanding, with reasoning catching up with human experts. In recent years, the challenges of complex data integration have been addressed by the construction of biomedical knowledge graphs.31–35 Accessibility of digital chemical and biological data facilitates both ML model development and the active use of models to optimize and drive the experimental drug discovery process, as realized in the emerging self-driving laboratories (SDLs).36 There is no doubt that modern drug discovery can already be characterized as an information-driven technology, helping to transition the field to exponential growth.
Deception
In this stealth phase of technological development (preceding exponential growth), a field grows slowly in preparation for the explosive transition to an exponential trajectory. In fact, it might seem linear and could go unnoticed or dismissed. This phase is called deceptive for a reason as it can describe the initiation of any new technology that might or might not become transformative later. However, one of our key arguments in this Perspective is that drug discovery is on the verge of coming out of the deceptive phase because of the cumulative effect of digitization (discussed above) and several experimental and computational technologies that can help put many more drugs on the market.
One strong indication of transitioning from the deception phase in drug discovery to a phase of rapid growth is the increase in the size of purchasable chemical libraries. Indeed, ‘in stock’ and ‘made on demand’ virtual molecular libraries are used in virtual screening (VS) campaigns as an essential, key part of hit identification. Significant advances in computational planning, robotics, and AI-guided methods to automate organic synthesis have enabled continuous growth of the number of purchasable compounds over a long period of time. In the 15 years between 2004 and 2019, the size of the ZINC database (very broadly used for VS) has grown linearly from less than a million to 1.4 billion molecules. However, in the last 4 years we have observed beyond-exponential growth of the chemical space, recently called the ‘Big Bang of the Chemical Universe.’37 Indeed, the ZINC database, commonly used for VS, includes over 37 billion unique chemical entries in its latest 22nd release38 (Figure 3). This recent dramatic transition signifies the very beginning of a disruptive trend, with some commercial libraries such as Enamine REAL Space,28 eMolecules,38 and MolDB39 already entering the league of trillions of readily synthesizable compounds.40 Such explosive growth has been supported by the increased reliance on AI and robotic synthesis (rather than through the manual labor of experienced medicinal chemists). Concurrently, the cost of synthesizing novel chemical matter has been gradually decreasing.41
FIGURE 3.
The ‘Big Bang’ of the chemistry universe, reflected by the growth of the ZINC database. This figure is taken from our previous work.37 Note that the current size of the accessible REAL SPACE chemical collection from Enamine stands at over 70 billion compounds.38
Of course, mere growth in the number of accessible chemicals does not equate to the transformation of the entire field of drug discovery. However, there has been a progressive evolution of multiple key technologies that can collectively enable transition from the deceptive to the exponential phase. For instance, DNA-encoded library (DEL) screening (which supports effective and relatively inexpensive bioactivity testing of billions of molecules) was initially proposed over 20 years ago and remained predominantly a ‘boutique’ technology practiced by a few specialized companies.42 However, with the recent addition of ML analysis of DEL data, the DEL–ML approach43 is becoming increasingly successful for generating multiple viable hit compounds across numerous protein targets.44 There are also ongoing developments increasing the accuracy and throughput of binding affinity measurements for small molecules using techniques such as surface plasmon resonance (SPR)45 and affinity selection mass spectrometry (ASMS),46 among others. Another computational methodology, free energy perturbation (FEP), which for a long time (about 30 years) was predominantly practiced in academic labs, has recently come of age as a major commercial technology47,48 to support hit-to-lead optimization. The development of protein degraders represents another recent hit discovery method supported by computational tools.49 And of course, there are growing hopes for generative AI methods such as AlphaFold3,50 which can enable the reliable de novo design of synthetically feasible protein binders as viable starting points for hit-to-lead optimization. We also project that supplementing the convergence of multiple computational or experimental tools enriched by digital technologies that focus on early-phase drug discovery with multiple virtual filters for drug likeness51 could eliminate molecules unlikely to succeed in absorption, distribution, metabolism, excretion, and toxicity (ADME-Tox) studies from further expensive development. For instance, it is a well-known observation that the broad use of Lipinski rules-based filters has dramatically increased the number of molecules surviving drug development phases of the drug discovery pipeline.52
To summarize this section, we think we can express cautious optimism that the ‘deception’ phase of drug discovery will be ending in the very near future, if this transition has not started already. This notion is supported by the multiple developments in computational and computer-assisted experimental technologies briefly reviewed above, which are generating hit and lead molecules more rapidly and in larger quantities than ever before, and by the recent promising observation mentioned above that drug candidates developed with AI have historically record success rates in Phase I clinical trials.21 We expect that with this increase and the development of end-to-end drug discovery platforms, such as that supported by Insilico Medicine,53 this input will translate into an exponential increase in the number of drug candidates entering and eventually succeeding in clinical trials.
Disruption
According to the ‘Six Ds of Exponentials,’ digital technologies disrupt existing industries by making traditional approaches obsolete. Typical examples of digital disruption include the use of MP3 players and streaming videos instead of CDs and tapes, and the global replacement of landline phones with mobile devices.
Digital disruption of experimental drug discovery technologies has been evident since the early days of computer-aided drug design (CADD). Starting from the early days of initial quantitative structure–activity relationship (QSAR) modeling attempts more than 60 years ago, mathematical tools have increasingly become a common part of medicinal chemistry, disrupting conventional protocols by significantly reducing both synthetic and biological screening efforts, and in some cases, replacing them altogether. Actively developed chemical similarity searching and VS methods, both ligand-based54–56 and structure-based,57,58 have served as alternatives to physical screening of compounds. It has been acknowledged that the use of computational approaches significantly increased the discovery rate as compared with experimental screening technologies. For instance, a recent viewpoint publication placed the experimentally confirmed hit rate for six different targets at 9.1% to 75%, with a median of 44.4%, with a few dozen molecules nominated and tested in each campaign.59 This compares favorably to the standard hit rate of high-throughput screening (HTS), estimated to be as low as 0.03% in some cases.51 It is also well established that computational filters such as Lipinski’s rules have dramatically reduced the failure rates for drug candidates with promising bioactivity that advanced to ADME studies.60 Further development of advanced computational filters51,61,62 should continue to increase the success rate of the steps involved in the drug discovery pipeline following initial hit identification.
Nowadays, AI continues to significantly disrupt traditional phases of the entire drug discovery and development process, spanning from target identification and modeling using tools such as AlphaFold to clinical development.63 Digital disruption has even affected computational drug discovery tools; indeed, the emergence of generative AI, ML, and DL-accelerated VS methods such as Deep Docking64 and hit discovery using docking enriched by generative modeling (HIDDEN GEM)58 questioned the need for the standard brute-force docking of large chemical libraries. Even with the use of high-performance computing (HPC), traditional docking methods can typically handle millions of chemicals, with isolated cases of tackling 1 billion molecules using gigantic computational resources only accessible to a handful of labs worldwide.65 However, the capabilities of brute-force docking lag behind the current demand for VS campaigns against current ultra-large purchasable libraries counting dozens of billions of molecules.66
The increased use of Deep Docking methods has been complemented by similar trends in ligand-based drug discovery. In our recent publication,36 we reviewed key advances in the field that we now call deep QSAR, where we reflected on the use of deep generative reinforcement and modeling approaches in drug discovery that help improve hit rates in screening campaigns. Integration of these technologies within the modern design, make, test, analyze (DMTA) cycle, especially when built into SDLs, has an evident disrupting effect on medicinal chemistry by effectively eliminating humans from the process of chemical synthesis optimization.
In summary, computational methods have been disrupting experimental drug discovery for over 60 years. As discussed in this section, the recent proliferation of innovative AI and DL algorithms to support chemistry is bound to further increase the numbers of viable drug candidates. For instance, we expect that in the near future, the development of AI-based de novo ligand design tools, such as AlfaFold3, will further reduce the demand for both experimental and computational screening of chemicals. These technologies are expected to increase the number of experimentally validated hit compounds in early drug discovery. The development of robust ADME-Tox filters is also expected to further decrease the failure rate of ADME screening and increase the number of candidate molecules entering and succeeding in clinical trials.
At the same time, the natural excitement about the innovative power of AI methods should not translate into universal hype, as was intelligently alluded to by Bender and Cortés-Ciriano.67,68 It is indeed critical to remain objective in assessing areas where AI can support substantive technological advances or merely maintain the status quo. Nevertheless, the recent studies discussed above suggest that AI approaches are not only disruptive but also result in quantitative improvements in experimental hit rates. We express cautious optimism that digital technologies should progressively increase success rates in every stage of the drug discovery pipeline.
Demonetization
The essence of demonetization is that when technologies advance, the associated development costs decrease, and what was once expensive becomes cheap or free. The typical examples are multiple software products, such as cell phone apps, which are so inexpensive to develop that they are offered free or for a nominal charge. Such demonetization is still economically viable because demonetized components are still part of the overarching econosphere, which relies on these components to facilitate the path to economically tangible objectives.
As elaborated in the previous sections, the digitized experimental components of drug discovery, including computer-guided small molecules synthesis, automated assays and methods of structural biology, have evolved to become more efficient and cheaper with time (following or superseding Moore’s law). Many critical CADD resources also became freely available along the way, with some of them having a long-standing and remarkable history of open access and impact. Several notable examples of both small-molecule and macromolecular open-access databases support this notion.
On the small molecules side, one of the most impactful, freely accessible ZINC databases integrates data on commercially available chemicals that can be explored by various computational tools and nominated for biological testing. This resource served the scientific community for two decades.27 Numerous major depositories of drug-related data, including ChEMBL,30 PubChem,29 DrugBank,69 and DrugCentral,70 also provide valuable scientific information for free. Various algorithms have been developed to analyze these massive datasets to identify potential safety and efficacy issues early in the drug development process. For example, many ML approaches that have been used for years to build predictive QSAR models have been advanced to include DL and generative chemistry methods published as open-source tools. Similarly, generative adversarial networks (GANs) have been used to predict potential side effects of drug candidates and subsequent toxicities,71 including several that are open source.72–74
On the proteins side, the PDB was established in 1971 as a free public depository of carefully curated three-dimensional structures of biological macromolecules.26 The role of the PDB in drug development cannot be overestimated. Similarly, the Human Genome Project has transformed life sciences by providing the results of gene and protein sequencing to researchers worldwide since the 2000s.75 To accelerate and scale-up structure-based drug discovery efforts, the Structural Genomics Consortium (SGC) was founded in 2003 as a global public–private partnership (PPP) dedicated to open science.76 In fact, it was the public accessibility of both protein structures from the PDB and sequences from GenBank that enabled the development of revolutionary tools such as AlphaFold and RosettaFold, earning their developers a 2024 Nobel Prize in physiology and medicine.
The software part of the drug discovery workflow underwent especially significant demonetization over the years in both ligand- and structure-based domains. In cheminformatics and ligand-based research, the first software packages for calculating chemical descriptors, such as Molconn-Z77 and Dragon,78 were developed as commercial offerings; however, software such as PaDEL79 and the Chemistry Development Kit (CDK80) have emerged rapidly as free and powerful alternatives. Nowadays, many groups in both industry and academia rely on free cheminformatics resources, including multiple descriptors and ML tools offered by RDKit.81
Similar trends have been observed with in silico tools for protein structure prediction and structure-based drug discovery. Historically, academic software for homology model building, such as Modeller82 and Swiss-Model,83 has been offered for free to investigators in nonprofit organizations along with commercial licenses, whereas molecular modeling companies like MSI were developing and distributing commercial packages.84 Similarly, molecular docking software was provided historically through fee-based commercial licensing (GLIDE,85 ICM,86 FRED57) or as academia-restricted free programs like DOCK87 and AutoDock.88 Over time, it has become increasingly common to provide docking software under open-access conditions (AutoDock Vina,89 GPU-AutoDock,90 DiffDock,91 EquiBind,92 CDOCKER,93 MEGADOCK-GPU,94 Open Drug Discovery Toolkit95). This trend became particularly noticeable with the emergence of AI-driven drug discovery approaches, when it became conventional to deposit source codes and software for CADD methods employing ML acceleration (such as Deep Docking,64,96 and DeepMol97) and methods of generative chemistry (REIVNVENT,98 HIDDEN GEM,58 RELATION99) into GitHub and other free exchange platforms. At the interface between computational and experimental chemistry, a similar trend can be observed for the algorithms used to create robotic chemistry laboratories and SDLs.36
Traditionally, basic research has been one of the most demonetized areas of human creativity, exemplified in recent decades, for instance, by the Human Genome Project and the international agreement to freely release genome sequences. This open science initiative and unrestricted accessibility of biological data have led to and catalyzed the development of the entire field of bioinformatics and enabled many impactful scientific discoveries. Chemical research, including medicinal chemistry, stayed way behind the biomedical field in terms of openness and the sheer amount of data available for modeling and ML exploration because of heightened intellectual property (IP) protection. This restriction inhibited the development of cheminformatics, also keeping it behind bioinformatics in terms of breadth and impact.
However, the recent data revolution in chemistry has led to the rapid expansion of both chemical bioactivity and screening libraries in the public domain and growing availability of free cheminformatics and computational chemistry software. The ongoing demonetization of chemical and bioactivity data, as well as software, further supports the notion of a new era of exponential growth of drug discovery technologies.
Dematerialization
As physical products and services become digital, they increasingly dematerialize and become smaller in size or completely virtual, gaining much increased operability and accessibility. A common example is the evolution of cell phones, which progressively acquired multiple functions on top of their original use as a means of mobile communications.
Digital representation of chemical and biological information has enabled efficient implementation of all stages of the DMTA cycle of drug development100 mentioned above and emerging trends in the field, such as the merger of wet- and dry-lab operations36 and hybridization of the hardware and software components of CADD.36,101 The reality of laboratory science is also being drastically transformed by the use of robotic and autonomous chemical synthesis and biological activity testing technologies. This concept relies on the idea of SDLs,102 which realize the power of predictive and generative AI by combining it with robotics and advanced computing components. In addition to the disruption of medicinal chemistry by effectively eliminating human chemists, as mentioned in one of the preceding sections, SDLs dematerialize the planning, execution, and analysis of laboratory experiments and enable much more efficient discovery of novel drug candidates at drastically reduced time and cost. The creation of continuous manufacturing pipelines supported by digitalized self-driving chemical labs is an example of physical miniaturization with improved process efficiency.103,104
Dematerialization of the computer hardware used in CADD has been highly visible as mainframes have been gradually replaced by distributed computing and storage. The aforementioned evolution of hardware toward hybridization with software101 is another notable trend that aligns with the dematerialization of computational resources, as is the proliferation of novel computational principles such as quantum computing. One can envision that extremely specific technological requirements for operational qubits will make the corresponding CADD hardware infrastructure even more elusive.70 Yet, thanks to the growing availability of distributed HPC and advances in ML and AI-accelerated algorithms, this dematerialization of computer hardware is coupled with the increased productivity of CADD methods. Modern VS pipelines, including VirtualFlow,105 V-Synthes,106 DeepDocking,96,107 HIDDEN GEM,58 and other AI-enabled tools, can be regarded as the poster children of these trends.
The process of miniaturization has touched not only CADD hardware. In traditional medicinal chemistry, the processes of chemical synthesis, biological screening and ML/AI modeling have been conceptually and functionally linked but disconnected in time and space. However, there is a growing trend toward physical integration of all key components of the DMTA cycle as the major design principle of SDLs, which are rapidly evolving in the direction of autonomation and dematerialization. It is feasible to speculate that, at some point in the near future, the entire drug discovery process (or at least its ‘hit-discovery’ and ‘hit-to-lead’ optimization components) could go exclusively ‘dry,’ integrated in the form of a ‘digital twin’ of the current physical drug discovery platforms. We envision that through this digital twin transformation, computational tools will continue to support further physical dematerialization of drug discovery. We expect that as computational tools become more powerful and models built with these tools more reliable, the demand for physical equipment will evolve from supporting the synthesis and testing of large chemical libraries as part of the discovery process to validation of small numbers of virtual hit compounds nominated by efficient computational models. Thus the ongoing dematerialization of both computational and experimental drug discovery technologies considered in this section lends another level of support to our observation that drug discovery is becoming aligned with the ‘Six Ds’ concept.
Democratization
The sixth D, democratization, implies that at certain points of their evolution, exponential technologies become ubiquitous and accessible to everyone, not only governments, big companies or wealthy individuals. Accessibility increases the number of technology users and practitioners, which empowers entrepreneurship and innovation. A well-known example of democratization in the software industry is the advent of Linux. Historically, the UNIX operating systems (OSs) were costly and available to big businesses as proprietary software. Linux was created to provide a freely available kernel and OS available for all to learn. This democratization stimulated innovation in the IT industry, allowed organizations to save on OS costs and establishing an ecosystem of free applications and tools.108
Both computational- and experimental technologies that support drug discovery have become considerably more accessible over the years. As mentioned above, in the early days of CADD, most molecular modeling programs were developed by specialized companies such as Tripos, Biosym, CCG, and others, and licensed to the pharmaceutical industry for significant fees. Progressively, software developed in academia such as MD simulation packages (AMBER,109 CHARMM,110 GROMOS111) and docking tools (DOCK87 and AutoDOCK88) have been made freely accessible to academics and even nonacademic users.
It is also possible to anticipate that, as more versatile and powerful CADD resources become public and freely available, we will witness broader involvement of ‘community scientists’ in drug discovery, as has been exemplified by crowd-sourcing initiatives such as folding@home112 and docking@home.113 These community platforms, developed in the earlier days of CADD, were resurrected during the coronavirus disease 2019 (COVID-19) pandemic along with novel crowdsourcing and crowdfunding mechanisms.114 We would point out, however, that the pandemic also revealed a somewhat unexpected consequence of the democratization of CADD, when thousands of preprint papers with loud claims of not just discovering new COVID-19 drugs but doing this quickly were published. Most of these studies employed free CADD resources and did not provide any experimental verification, which led to an eventual ban being placed on such submissions even to preprint servers114 along with the harsh criticism of such predictions unsupported by experimental data.115
Although there are clear signs of democratization of the in silico component of drug discovery, the industry as a whole is still confounded within big pharma. We could, however, speculate that CADD being ahead of experimental drug discovery research on the ‘Six Ds’ trajectory could catalyze the overarching transformation of the pharmaceutical industry toward open science.44 For example, the global initiative dubbed Target 2035 supports drug discovery by identifying and providing freely available chemical probes for all human proteins.116 Multiple open science projects that support drug discovery, such as OPenMM,117 Conscience,118 Open Molecular Software,119 Open Free Energy,120 the multiple open source ML tools listed above, and many other developments in the open science community, continuously contribute to drug discovery democratization.
We anticipate that these recent advances in democratized AI-accelerated CADD should lead to digital disruption within the actual physical processes of experimental drug discovery, resulting in a surge of better and safer therapeutics. Thus the digital transformation of drug discovery could eventually change the entire regulatory landscape of the industry and bring down the time and cost of therapeutic development, helping to turn the ‘Eroom’ trend around.
Concluding remarks
Undoubtedly, there continues to be a strong demand for novel therapeutic agents. Despite progressively growing costs in the pharmaceutical industry over decades, the pace of introducing approved drugs on the market has remained steady and low. Hence, without significant breakthroughs in drug discovery and development technologies, we cannot address the urgent societal demand for novel medications.
In this Perspective, we have reviewed the state of the art in modern drug discovery through the prism of the ‘Six Ds of Exponentials’ concept developed by Peter Diamandis to explain the mechanisms that drove multiple transformative technologies to rapid and impactful economic and functional success. According to this concept, the first phase, digitization, lies at the core of such technologies, which then proliferate throughout the remaining phases of deception, disruption, demonetization, dematerialization, and democratization, enabling exponential growth. Herein we have contemplated and explored the critical question of whether this inspirational concept can be applied to and support the exponential acceleration of drug discovery. We summarize our observations in a compact form in Table 1.
TABLE 1.
Translation of the ‘Six Ds of Exponentials’ to pharmaceutical development
| The ‘6-Ds of exponentials’ as defined by Peter Diamandis | Original definition | Examples of respective factors and trends in pharmaceutical development |
|---|---|---|
| Digitization | Transfer of physical information into digital form | Growth of diverse biomedical databases; emergence of knowledge graphs to represent and facilitate mining of biomedical knowledgebase; Large Language Models |
| Deception | The stealth phase of technology development preceding the exponential growth phase | Evolution of chemical synthesis (expansion of purchasable chemical libraries) and high-throughput screening technologies [e.g., DNA encoded libraries (DEL)] |
| Disruption | Digital or digitized technologies that make traditional approaches obsolete | Traditional (QSAR modeling, docking) and recently emerged AI methods (generative chemistry, Deep Docking) that reduce the need for chemical synthesis and experimental testing |
| Demonetization | Cost reduction of goods due to technological advances | Reduction in cost per molecular entity of gene sequencing, robotic synthesis, biological screening; information sharing as part of open science movement |
| Dematerialization | Size reduction of physical products or services due to digital technologies | Reduction in size of instruments for chemical synthesis and biological screening, self-driving laboratories, virtual screening technologies |
| Democratization | Broad accessibility of advanced technologies | Open access or open-source CADD, ML, AI software; open science crowd-sourcing projects |
We have observed that, starting with computational tools and hardware used to support drug discovery, digitization has already affected all key processes including, in some cases, target discovery, chemical synthesis (retro- and forward planning, automation, robotic synthesis), testing (automated and miniaturized HTS), free data and software exchange, leading to a progressive reduction of experimental effort by using computational discovery tools. At the same time, we acknowledge that digitization has been most effective to support early-phase hit discovery and hit-to-lead optimization once the biological target associated with the disease has been identified. The great current challenge is to extend the ‘Six Ds’ principles to accelerate target discovery, as we continue to lack substantial knowledge and data concerning the mechanisms of chemical effects on biological processes in the human body. The inadequacy of many current animal models, let alone assays of lower biological complexity, has been acknowledged. Hopefully, this challenge will be cogently addressed through studies supported by the recently announced Complement-ARIE program at the National Institutes of Health (NIH),121 with supporting in silico developments stimulated by the FDA Modernization Act 2.0,122 and by recent FDA guidance on the ‘Use of Artificial Intelligence to Support Regulatory Decision-Making.’123
There is growing hope that AI-driven methods are beginning to accelerate drug discovery.124 Several companies such as Recursion (and Exscientia, which was acquired by Recursion) and Insilico Medicine have reported early successes in developing preclinical drug candidates and pushing these candidates into clinical trials on a 12–18-month time scale, at a speed previously unseen in the industry. Although it is too early to claim unequivocal success, and where cautionary notes have been published very recently, careful optimism has been expressed concurrently with both companies reporting positive Phase II results.125,126
With the expectation that AI acceleration will make drug discovery more time- and cost-efficient, we will increasingly witness its expansion into smaller companies, academic groups or even ‘community science,’ where any individual could design a high-value, high-probability drug candidate with just a few clicks. When better and safer drug candidates start emerging in large numbers, inevitable changes will also occur in the research, industry, and regulation landscapes, leading to the closure of a shocking gap between approved and needed drugs.
We would like to note that there are many emerging, but not yet explicit, trends in drug discovery that reflect the field’s readiness to adopt the ‘Six Ds’ transformational stages. Therefore once we have rationalized these trends, we could start building upon them and use them to drive digital transformation of pharmaceutical discovery and development. We see the following key areas of investment as likely having rapid returns in the form of transitioning into the exponential growth phase.
Massive high-quality chemical bioactivity data generation to train models
As discussed below, this objective can be achieved by supporting massive DEL screening followed by ML modeling, as well as by VS of ultra-large purchasable libraries with the methods of deep QSAR36 and Deep Docking64 and reliable hit validation using the methods of ASMS and SPR.
Development of chemical robotics, automation, and SDLs
As we discussed at the beginning of this Perspective, the entire concept of ‘Six Ds’ emerged by analyzing the burgeoning evolution of computer hardware. The federal support for this evolution continues today, as illustrated by a recent notice of funding opportunity (NOFO) for activities that will use cutting-edge AI and autonomous experimentation (AE) technologies to support the long-term viability of next-generation semiconductor manufacturing.127 Similar investments are required to promote automation in drug discovery.
Much higher level of data integration across the entire drug discovery pipeline
Historically, the use of information about later phases of drug discovery has been instrumental in reducing the number of compounds designed in early phases that would likely fail in preclinical trials. A better understanding of the mechanisms of drug action and analysis of real-world data on drug efficacy and adverse effects can be enabled by emerging data science, AI, and technologies including biomedical knowledge graphs,31,33–35 as illustrated by the successful repurposing of the rheumatoid arthritis drug baricitinib as an FDA-approved therapy for COVID-19.128,129
In addition to these technical developments, which require focused investments, we foresee that the following strategic objectives could further facilitate the digital disruption of drug discovery:
Greater integration of the research community
The recent COVID-19 pandemic made a significant contribution to the demonetization and democratization of drug discovery, driven by the assumption that openness could accelerate the development of much-needed antiviral therapies. As described in the 2021 review coauthored by many leaders in the field,130 in efforts to bypass traditional scientific burdens (the need to patent, secure research funding, or boost academic profiles), several demonetization paradigms have been promoted.
Open access
Broad dissemination of peer-reviewed research is crucial for rapid drug discovery. In the wake of the global pandemic, all major publishers made COVID-19-related papers freely available, in an unprecedented move. Moreover, the ever-increasing popularity of preprint services has also significantly contributed to open-access COVID-19 research.
Open data
The deposition of open-license datasets has been crucial for science. The value of open data has been clearly seen through numerous COVID-19-devoted initiatives, including the community-based drug development project Moonshot and the experimental platform for SARS-CoV-2 target screens of the National Center for Advancing Translational Sciences (NCATS) (with the results publicly disclosed in a real time) as well as the open aggregation of tools and data through platforms such as CanSAR, CORD-1915, and Nextstrain, among others.
Open-source CADD
Even before the pandemic, open source had been implemented in many areas of drug research, including the discovery and synthesis of therapeutics for neglected and tropical diseases, sharing physical samples for biomedical research and collective optimization of drug candidates. A generalized open science consensus was agreed upon by a number of global and national coalitions during the COVID-19 pandemic, when many research, business and regulatory consortia, for instance the COVID-19 Moonshot project,131 committed to supporting open science and collectively contributed to the resolution of the deadly pandemic and raising awareness and preparedness for future infectious threats.
The open science concept has recently been realized in the Critical Assessment of Computation Hit-finding Experiments (CACHE) global challenges, which expanded community-based drug discovery to many other orphan- or otherwise critically unaddressed diseases including Parkinson’s.132 CACHE aims to provide free high-quality experimental feedback on CADD predictions from qualified participants around the world, with a new critical drug target every 4 months and with all CACHE predictions and experimental results becoming public.
Greater engagement between science and society
As we have repeatedly noted throughout this Perspective, the COVID-19 crisis played a pivotal role in the disruption, demonetization, and democratization of drug research. In a time of dire need for an entirely novel antiviral drug, which had to be developed in a matter of months rather than years (if not decades), many novel mechanisms of interaction between science and society emerged. Some of those mechanisms have proven to be very effective, involving aspects of community research, crowdfunding, and educating the public.
The reputation of scientists and science surged during the pandemic. Nowadays, the dialogue between science and society should not only widen and deepen but should start including certain sensitive subjects including the use of patient data, access to biological materials, stem cell research, and animal testing, among others. It is fair to state that many more interactions are needed between the research community and society. People should be better informed about the life-saving work of scientists.133 In turn, scientists should better communicate their needs to society, articulating their needs for support and recognition to bring much-needed scientific discoveries.
Stronger government support for science
Government support of fundamental science is imperative; most academic research is funded through government grants, fellowships, and contracts. There are many positive examples of government regulations, including open policies on genomic information, gene and drug target patenting, government-funded programs on the democratization of predictive toxicology, chemical genomics, and many other critical areas.
These days, we need extensive government support for cloud technologies and HPC platforms playing a paramount role in the demonetization, dematerialization, and democratization of science. Government support is much needed for technologies with significant disruptive potential, including cryo-microscopy, quantum computing, robotics, nano-engineering, and AI. Strategic government investments are needed to ensure the much-needed transition of drug discovery into exponential growth.
Scientists should also closely work with governments on the strategic positioning of AI and CADD technologies to address pressing drug development issues, including reducing failure rates for investigatory new drugs. If (when) those efforts are successful and AI-designed drugs are indeed proven to be better and safer, then the stage will be set for lowering regulatory burdens and further drastic reduction of drug development costs.
Increase in PPPs
One of the indicative signs of democratization is the growing trend of early drug discovery being shifted from industry to academia. Many early-clinical or even preclinical candidates are being acquired by big pharma from universities. The emergence of novel models of interaction with the industry in the form of PPPs, various models of industry funded research, and philanthropy is also being seen.
With the above models being on a rise, there is a need for entirely new financial models involving drug industry that will work in the context of the ‘Six Ds’ when they become fully realized. Ultimately, drug discovery, while being broadly disseminated and democratized, should still be profitable, like all other exponential technologies. Yet, the appropriate mechanisms remain to be developed, and this Perspective could be considered as an invitation to start the corresponding discussions.
Outlook
As was noted even by New York Times,134 last year’s Nobel prize awards in both physics and chemistry went to computer science for developing foundational AI algorithms and applying them to outstanding problems in protein folding and design. This recognition of the pivotal role that computing and AI have already played in modern research also underscores the hope for even greater achievements of computing and digitization in improving life in our society. A significant part of our wellbeing is accounted for by healthy living, which requires effective treatment of known and emerging diseases using powerful pharmacotherapies. This societal demand remains unaddressed, however, and as we argue in this Perspective, the already-ongoing digitization of pharmaceutical discovery and development has distinct potential to reverse the current ‘Eroom’ trend and enable exponential growth of drug discovery. We posit that this transformation is already on the horizon if we consciously adopt the ‘Six Ds’ of exponentials as the driving concepts for all key elements of drug discovery.
We need to acknowledge that the different phases of drug development might not be equally amenable to implementation of the ‘Six Ds’ concept. Early drug discovery has been the most affected so far, particularly concerning hit discovery and lead optimization. Digitization, automation and cloud-based data analysis have dramatically accelerated these processes, allowing for rapid identification of promising drug candidates at reduced costs. Open-access databases and AI tools have also democratized drug discovery, making it more accessible beyond large pharmaceutical companies. In contrast, clinical trials and regulatory approval remain as significant bottlenecks because of biological variability, ethical concerns and regulatory constraints, particularly in clinical trials where human testing is required and where regulatory agencies naturally prioritize safety over speed.
However, we expect that thoughtful application of the ‘Six Ds’ principles in all phases of therapeutics development will lead to a much higher number of molecules entering and successfully progressing through the drug discovery pipeline toward approved medicines. As discussed in this work, AI-driven technologies are beginning to proliferate into the regulatory science space, illustrated by the growing push by the FDA and similar agencies elsewhere to rely on NAMs22 and support the use of AI123 for regulatory decision-making. We anticipate that transformation of the drug discovery pipeline, driven by the ‘Six Ds of Exponentials,’ and modernization of regulatory drug approval will soon enable rapid, exponential growth of the pharmaceutical industry, eradicating the ominous problem of previously uncurable diseases.
Highlights.
Drug discovery has been marked by rising costs and longer development times, with a modest increase, if any, in the number of annual drug approvals.
In this Perspective, we argue that similarly to other successful sectors of the economy, the transformative acceleration of drug discovery, coupled with significant cost reduction, can also be achieved through digital disruption brought by maturing AI, increased automation of experiments, public–private partnerships, and open science initiatives.
Through the analysis of recent trends and advancements in pharmaceutical research, we detected the emergence of all ‘Six Ds of Exponentials’ as part of modern drug discovery and discuss how each of the ‘Six Ds’ can further empower the field.
Teaser.
Digital disruption characterized by the ‘Six Ds of Exponentials’ has emerged in modern drug discovery and is poised to empower the field. Rational adoption of this paradigm can catalyze the transformative change needed for novel, potent, affordable, and accessible medicines.
Acknowledgments
AC and AT wish to thank Professor Alexandre Varnek from the University of Strasbourg, who organized the Strasbourg Summer School in Chemoinformatics – 2024, providing the venue where this Perspective was conceptualized. All authors express sincere gratitude to Dr Jack Scannell for careful reading of our manuscript and many critical suggestions that helped improve this Perspective. AT acknowledges support from NIH (grant GM140154).
Biography
Dr. Artem Cherkasov

Dr. Artem Cherkasov is a Professor in the Department of Urologic Sciences, Faculty of Medicine at the University of British Columbia (previously – at UBC Infections Diseases) and the Head of Therapeutics Development at the Vancouver Prostate Centre. Research interests include computer-aided drug discovery (CADD), structure-activity modeling, drug repurposing, personalized drug discovery and development of novel CADD approaches and tools.
Dr. Cherkasov has co-authored more than 200 research papers, 80 patent filings and several prominent book chapters. During his tenure at UBC Dr. Cherkasov generated more than 100M in research funding (as PI or co-PI).
Together with UBC he has successfully commercialized 12 technologies, including 6 drug candidates that have been licensed to Big Pharma (Roche, Astra-Zeneca), a major VC fund (5AM Venture) and several national and international start-up companies (INDEL Therapeutics, LAST Innovation, ABT Innovations, MAIA Pharmaceuticals).
Among those, the androgen receptor inhibitor developed by Dr. Cherkasov was licensed by la Hoffman-Roche in 2015 for 142M USD – a record amount for Canadian academic institutions.
In 2017, Dr. Cherkasov led a group of 10 prominent PIs in a successful 24M dollar application to the Canadian Foundation of Innovation to create a complete ADDUCT (Accelerated Drug Discovery Using Clinical Translation) infrastructure at UBC. The grant completes the UBC cross-discipline drug discovery platform with the latest cryo-EM technology, enhanced protein production and purification capabilities and substantial computational infrastructure, providing cutting-edge support for personalized drug discovery.
In response to the global COVID19 pandemic, Dr. Cherkasov’s team immediately mobilized the ADDUCT drug discovery pipeline. The team utilized their novel AI-based platform, Deep Docking (PMID: 32607441), with the recently published structure of the COVID19 3CL Mpro target, applying DD to process 1.3 billion compounds; within a week, they had identified the top 1000 potential ligands which they shared with the scientific community in a scientific paper published as early as mid-February (PMID: 32162456) providing a significant head-start in developing drug against COVID19. Dr. Cherkasov led a group of UBC scientists in a series of successful COVID-19 grant applications securing substantial support from Canadian and international funding entities and entered research collaborations with NVIDIA, Dell and AMD.
The translational nature of Dr. Cherkasov’s research laid a foundation for the development of the ‘Best Practices of Computer-Aided Drug Discovery’ (https://pubs.acs.org/doi/abs/10.1021/acs.jcim.7b00137) that are being adopted world-wide.
In 2014 Dr. Cherkasov, together with Dr. Tropsha co-led a group of 20 international PIs in writing a program review of the field of QSAR (Quantitative Structure-Activity Relationship) modeling (PMID: 24351051). That paper where Dr. Cherkasov is the first author was cited more than 900 times, and for the last 5 years was listed among top 1% of all citations in chemistry.
Last year Drs. Cherkasov and Tropsha co-led another program publication from an international consortium of 19 leaders where we formulated Best Practices in Big Data processing and AI drug modeling. The review, where Dr. Cherkasov is a co-senior author is published in Chem Soc Rev (https://pubs.rsc.org/en/content/articlelanding/cs/2020/d0cs00098a).
Footnotes
Declarations of interest
No interests are declared.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Hinkson IV, Madej B, Stahlberg EA. Accelerating therapeutics for opportunities in medicine: a paradigm shift in drug discovery. Front Pharmacol. 2020;11:770. 10.3389/fphar.2020.00770. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Sanghvi R. Top 15+ Fastest Growing Industries in the US. Bitcot Website. https://www.bitcot.com/fastest-growing-industries/#:~:text=To%20summarize%2C%20the%20fastest%2Dgrowing,healthcare%2C%20and%20renewable%20energy%20sector. Published January 28, 2025. Accessed March 3, 2025. [Google Scholar]
- 3.Micro Processor Cost per Transistor Cycle. Singularity.com Website. https://www.singularity.com/charts/page62.html. Accessed March 3, 2025.
- 4.Transistors per Microprocessor. Singularity.com Website. https://www.singularity.com/charts/page63.html. Accessed March 3, 2025.
- 5.Pandey M et al. The transformational role of GPU computing and deep learning in drug discovery. Nat Mach Intell. 2022;4(3):211–221. 10.1038/s42256-022-00463-x. [DOI] [Google Scholar]
- 6.Wouters OJ, McKee M, Luyten J. Estimated research and development investment needed to bring a new medicine to market, 2009–2018. JAMA. 2020;323(9):844–853. 10.1001/jama.2020.1166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Munos B Lessons from 60 years of pharmaceutical innovation. Nat Rev Drug Discov. 2009;8(12):959–968. 10.1038/nrd2961. [DOI] [PubMed] [Google Scholar]
- 8.Hall J, Matos S, Gold S, Severino LS. The paradox of sustainable innovation: the ‘Eroom’ effect (Moore’s law backwards). J Clean Prod. 2018;172:3487–3497. 10.1016/j.jclepro.2017.07.162. [DOI] [Google Scholar]
- 9.Scannell JW, Blanckley A, Boldon H, Warrington B. Diagnosing the decline in pharmaceutical R&D efficiency. Nat Rev Drug Discov. 2012;11(3):191–200. 10.1038/nrd3681. [DOI] [PubMed] [Google Scholar]
- 10.CURE Drug Repurposing Collaboratory. C-Path Website. https://c-path.org/c-path-launches-cure-drug-repurposing-collaboratory-to-accelerate-identification-of-new-uses-of-existing-drugs-to-treat-infectious-diseases-including-covid-19/. Published June 23, 2020. Accessed March 3, 2025. [Google Scholar]
- 11.Fermaglich LJ, Miller KL. A comprehensive study of the rare diseases and conditions targeted by orphan drug designations and approvals over the forty years of the Orphan Drug Act. Orphanet J Rare Dis. 2023;18(1):163. 10.1186/s13023-023-02790-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ringel MS, Scannell JW, Baedeker M, Schulze U. Breaking Eroom’s law. Nat Rev Drug Discov. 2020;19(12):833–834. 10.1038/d41573-020-00059-3. [DOI] [PubMed] [Google Scholar]
- 13.Research and Development Continues Long after a Medicine is Initially Approved. PhRMA Website. https://phrma.org/blog/research-and-development-continues-long-after-a-medicine-is-initially-approved. Published February 1, 2022. Accessed March 3, 2025. [Google Scholar]
- 14.de la Torre BG, Albericio F. The pharmaceutical industry in 2023: an analysis of FDA drug approvals from the perspective of molecules. Molecules. 2024;29(3):585. 10.3390/molecules29030585. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Diamandis PH, Kotler S. Bold: How to Go Big, Create Wealth and Impact the World. Reprint Edition. New York, NY: Simon & Schuster; 2016. [Google Scholar]
- 16.Kurzweil R The Age of Spiritual Machines: When Computers Exceed Human Intelligence. City of Westminster, London: Penguin Books; 2000. [Google Scholar]
- 17.Rojas MA. Exponential Technologies in a Nutshell. Transformation Architect Website. https://transformation-architect.com/exponential-technologies-in-a-nutshell/. Accessed March 3, 2025. [Google Scholar]
- 18.Levy N. Amazon’s $2.5M ‘Alexa Prize’ seeks chatbot that can converse intelligently for 20 minutes. GeekWire Website. https://www.geekwire.com/2016/amazon-award-2-5m-quest-alexa-chatbot-can-converse-intelligently-20-minutes/#:~:text=The%20winning%20team%20will%20receive,converse%20coherently%20for%2020%20minutes. Published September 29, 2016. Accessed March 3, 2025. [Google Scholar]
- 19.Mortenson O. How Many Users Does ChatGPT Have? Statistics & Facts. https://seo.ai/blog/how-many-users-does-chatgpt-have#:~:text=The%20most%20recent%20data%20from,website%20attracted%201.6%20billion%20visits. Accessed March 3, 2025.
- 20.Exponential Growth in Technology & 5 Examples of Exponential Technologies. Rehab Agency Website. https://www.rehabagency.ai/insights/exponential-technology-growth. Published November 8, 2022. Accessed March 3, 2025. [Google Scholar]
- 21.Kp Jayatunga MKP, Ayers M, Bruens L, Jayanth D, Meier C. How successful are AI-discovered drugs in clinical trials? A first analysis and emerging lessons. Drug Discov Today. 2024;29(6):104009. 10.1016/j.drudis.2024.104009. [DOI] [PubMed] [Google Scholar]
- 22.Potential Approaches to Drive Future Integration of New Alternative Methods for Regulatory Decision-Making. FDA Website. https://www.fda.gov/media/182478/download. Published October, 2024. Accessed March 3, 2025. [Google Scholar]
- 23.Abernethy A et al. The promise of digital health: then, now, and the future. NAM Perspect. 2022;6(22). 10.31478/202206e. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Clark K, Karsch-Mizrachi I, Lipman DJ, Ostell J, Sayers EW. GenBank. Nucleic Acids Res. 2016;44(D1):D67–D72. 10.1093/nar/gkv1276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Pruitt KD, Tatusova T, Maglott DR. NCBI Reference Sequence (RefSeq): a curated non-redundant sequence database of genomes, transcripts and proteins. Nucleic Acids Res. 2005;33(Database issue):D501–D504. 10.1093/nar/gki025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Berman HM et al. The protein data bank. Nucleic Acids Res. 2000;28(1):235–242. 10.1093/nar/28.1.235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Irwin JJ, Shoichet BK. ZINC – a free database of commercially available compounds for virtual screening. J Chem Inf Model. 2005;45(1):177–182. 10.1021/ci049714. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.REAL Database. Enamine Website. https://enamine.net/compound-collections/real-compounds/real-database. Accessed March 3, 2025. [Google Scholar]
- 29.Kim S et al. PubChem 2023 update. Nucleic Acids Res. 2023;51(D1):D1373–D1380. 10.1093/nar/gkac956. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Zdrazil B et al. The ChEMBL Database in 2023: a drug discovery platform spanning multiple bioactivity data types and time periods. Nucleic Acids Res. 2024;52(D1):D1180–D1192. 10.1093/nar/gkad1004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Fecho K et al. A biomedical knowledge graph system to propose mechanistic hypotheses for real-world environmental health observations: cohort study and informatics application. JMIR Med Inform. 2021;9(7):e26714. 10.2196/26714. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Womack F, McClelland J, Koslicki D. Leveraging distributed biomedical knowledge sources to discover novel uses for known drugs. bioRxiv. 2019:765305. 10.1101/765305. [DOI] [Google Scholar]
- 33.Bizon C et al. ROBOKOP KG and KGB: integrated knowledge graphs from federated sources. J Chem Inf Model. 2019;59(12):4968–4973. 10.1021/acs.jcim.9b00683. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Morton K et al. ROBOKOP: an abstraction layer and user interface for knowledge graphs to support question answering. Bioinformatics. 2019;35(24):5382–5384. 10.1093/bioinformatics/btz604. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Beasley JMT et al. ExEmPLAR (Extracting, Exploring, and Embedding Pathways Leading to Actionable Research): a user-friendly interface for knowledge graph mining. Bioinformatics. 2024;40(1):btad779. 10.1093/bioinformatics/btad779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Tropsha A, Isayev O, Varnek A, Schneider G, Cherkasov A. Integrating QSAR modelling and deep learning in drug discovery: the emergence of deep QSAR. Nat Rev Drug Discov. 2024;23(2):141–155. 10.1038/s41573-023-00832-0. [DOI] [PubMed] [Google Scholar]
- 37.Cherkasov A. The ‘Big Bang’ of the chemical universe. Nat Chem Biol. 2023;19(6):667–668. 10.1038/s41589-022-01233-x. [DOI] [PubMed] [Google Scholar]
- 38.eXplore The World’s Largest Commercially Available Chemical Space. eMolecules Website. https://www.emolecules.com/explore. Accessed March 3, 2025. [Google Scholar]
- 39.Home Page. MolDB Website. https://www.moldb.com. Accessed March 3, 2025. [Google Scholar]
- 40.Warr WA, Nicklaus MC, Nicolaou CA, Rarey M. Exploration of ultralarge compound collections for drug discovery. J Chem Inf Model. 2022;62(9):2021–2034. 10.1021/acs.jcim.2c00224. [DOI] [PubMed] [Google Scholar]
- 41.Fernandez M et al. Quantitative structure–price relationship (QS$R) modeling and the development of economically feasible drug discovery projects. J Chem Inf Model. 2019;59(4):1306–1313. 10.1021/acs.jcim.8b00747. [DOI] [PubMed] [Google Scholar]
- 42.Brenner S, Lerner RA. Encoded combinatorial chemistry. Proc Natl Acad Sci U S A. 1992;89(12):5381–5383. 10.1073/pnas.89.12.5381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.McCloskey K et al. Machine learning on DNA-encoded libraries: a new paradigm for hit finding. J Med Chem. 2020;63(16):8857–8866. 10.1021/acs.jmedchem.0c00452. [DOI] [PubMed] [Google Scholar]
- 44.Edfeldt K et al. A data science roadmap for open science organizations engaged in early-stage drug discovery. Nat Commun. 2024;15(1):5640. 10.1038/s41467-024-49777-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Neumann T, Junker HD, Schmidt K, Sekul R. SPR-based fragment screening: advantages and applications. Curr Top Med Chem. 2007;7(16):1630–1642. 10.2174/156802607782341073. [DOI] [PubMed] [Google Scholar]
- 46.Prudent R, Lemoine H, Walsh J, Roche D. Affinity selection mass spectrometry speeding drug discovery. Drug Discov Today. 2023;28(11):103760. 10.1016/j.drudis.2023.103760. [DOI] [PubMed] [Google Scholar]
- 47.Wang L, Chambers J, Abel R. Protein–ligand binding free energy calculations with FEP+. 2019:201–232. 10.1007/978-1-4939-9608-7_9. [DOI] [PubMed] [Google Scholar]
- 48.FEP+. Schrodinger Website. https://www.schrodinger.com/platform/products/fep/. Accessed March 3, 2025. [Google Scholar]
- 49.Mostofian B et al. Targeted protein degradation: advances, challenges, and prospects for computational methods. J Chem Inf Model. 2023;63(17):5408–5432. 10.1021/acs.jcim.3c00603. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Abramson J et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature. 2024;630(8016):493–500. 10.1038/s41586-024-07487-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Wellnitz J et al. STOPLIGHT: a hit scoring calculator. J Chem Inf Model. 2024;64(11):4387–4391. 10.1021/acs.jcim.4c00412. [DOI] [PubMed] [Google Scholar]
- 52.Oprea TI. Virtual screening in lead discovery: a viewpoint. Molecules. 2002;7(1):51–62. 10.3390/70100051. [DOI] [Google Scholar]
- 53.Artificial Intelligence for Every Step of Pharmaceutical Research and Development. Insilico Medicine Website. https://insilico.com. Accessed March 3, 2025. [Google Scholar]
- 54.Luo M, Wang XS, Tropsha A. Comparative analysis of QSAR-based vs. chemical similarity based predictors of GPCRs binding affinity. Mol Inform. 2016;35(1):36–41. 10.1002/minf.201500038. [DOI] [PubMed] [Google Scholar]
- 55.Wegscheid-Gerlach C. Chemoinformatics approaches to virtual screening. Edited by varnek alexandre and tropsha alexander. ChemMedChem. 2009;4(9):1546. 10.1002/cmdc.200900223. [DOI] [Google Scholar]
- 56.Yan X, Liao C, Liu Z, Hagler AT, Gu Q, Xu J. Chemical structure similarity search for ligand-based virtual screening: methods and computational resources. Curr Drug Targets. 2016;17(14):1580–1585. 10.2174/1389450116666151102095555. [DOI] [PubMed] [Google Scholar]
- 57.McGann M. FRED pose prediction and virtual screening accuracy. J Chem Inf Model. 2011;51(3):578–596. 10.1021/ci100436p. [DOI] [PubMed] [Google Scholar]
- 58.Popov KI, Wellnitz J, Maxfield T, Tropsha A. HIt Discovery using docking ENriched by GEnerative Modeling (HIDDEN GEM): a novel computational workflow for accelerated virtual screening of ultra-large chemical libraries. Mol Inform. 2024;43(1):e202300207. 10.1002/minf.202300207. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Zhao H. The science and art of structure-based virtual screening. ACS Med Chem Lett. 2024;15(4):436–440. 10.1021/acsmedchemlett.4c00093. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60.Waring MJ et al. An analysis of the attrition of drug candidates from four major pharmaceutical companies. Nat Rev Drug Discov. 2015;14(7):475–486. 10.1038/nrd4609. [DOI] [PubMed] [Google Scholar]
- 61.Li B et al. DrugMetric: quantitative drug-likeness scoring based on chemical space distance. Brief Bioinform. 2024;25(4):bbae321. 10.1093/bib/bbae321. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 62.Kralj S, Jukič M, Bren U. Molecular filters in medicinal chemistry. Encyclopedia. 2023;3(2):501–511. 10.3390/encyclopedia3020035. [DOI] [Google Scholar]
- 63.Vijayan RSK, Kihlberg J, Cross JB, Poongavanam V. Enhancing preclinical drug discovery with artificial intelligence. Drug Discov Today. 2022;27(4):967–984. 10.1016/j.drudis.2021.11.023. [DOI] [PubMed] [Google Scholar]
- 64.Gentile F et al. Deep docking: a deep learning platform for augmentation of structure based drug discovery. ACS Cent Sci. 2020;6(6):939–949. 10.1021/acscentsci.0c00229. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.Lyu J et al. Ultra-large library docking for discovering new chemotypes. Nature. 2019;566(7743):224–229. 10.1038/s41586-019-0917-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 66.Kuan J, Radaeva M, Avenido A, Cherkasov A, Gentile F. Keeping pace with the explosive growth of chemical libraries with structure-based virtual screening. WIREs Comput Mol Sci. 2023;13(6). 10.1002/wcms.1678. [DOI] [Google Scholar]
- 67.Bender A, Cortés-Ciriano I. Artificial intelligence in drug discovery: what is realistic, what are illusions? Part 1: ways to make an impact, and why we are not there yet. Drug Discov Today. 2021;26(2):511–524. 10.1016/j.drudis.2020.12.009. [DOI] [PubMed] [Google Scholar]
- 68.Bender A, Cortes-Ciriano I. Artificial intelligence in drug discovery: what is realistic, what are illusions? Part 2: a discussion of chemical and biological data. Drug Discov Today. 2021;26(4):1040–1052. 10.1016/j.drudis.2020.11.037. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 69.Knox C et al. DrugBank 6.0: the DrugBank Knowledgebase for 2024. Nucleic Acids Res. 2024;52(D1):D1265–D1275. 10.1093/nar/gkad976. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 70.Ursu O et al. Drug central: online drug compendium. Nucleic Acids Res. 2017;45(D1):D932. 10.1093/nar/gkw993. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 71.Gangwal A et al. Generative artificial intelligence in drug discovery: basic framework, recent advances, challenges, and opportunities. Front Pharmacol. 2024;15:1331062. 10.3389/fphar.2024.1331062. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 72.Chen X, Roberts R, Liu Z, Tong W. A generative adversarial network model alternative to animal studies for clinical pathology assessment. Nat Commun. 2023;14(1):7141. 10.1038/s41467-023-42933-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 73.Abbasi M et al. Designing optimized drug candidates with Generative Adversarial Network. J Cheminform. 2022;14(1):40. 10.1186/s13321-022-00623-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 74.Chen X, Roberts R, Tong W, Liu Z. Tox-GAN: an artificial intelligence approach alternative to animal studies—a case study with toxicogenomics. Toxicol Sci. 2022;186(2):242–259. 10.1093/toxsci/kfab157. [DOI] [PubMed] [Google Scholar]
- 75.Collins FS, Fink L. The human genome project. Alcohol Health Res World. 1995;19(3):190–195. [PMC free article] [PubMed] [Google Scholar]
- 76.Structural Genomics Consortium Home Page. SGC Website. https://www.thesgc.org. Accessed March 3, 2025. [Google Scholar]
- 77.Molconn Products. Molconn Website. https://www.molconn.com/products.html. Accessed March 3, 2025. [Google Scholar]
- 78.DRAGON 7.0. Kode Chemoinformatics Website. https://chm.kode-solutions.net/pf/dragon-7-0/. Accessed March 3, 2025. [Google Scholar]
- 79.Yap CW. PaDEL-descriptor: an open source software to calculate molecular descriptors and fingerprints. J Comput Chem. 2011;32(7):1466–1474. 10.1002/jcc.21707. [DOI] [PubMed] [Google Scholar]
- 80.Willighagen EL et al. The Chemistry Development Kit (CDK) v2.0: atom typing, depiction, molecular formulas, and substructure searching. J Cheminform. 2017;9(1):33. 10.1186/s13321-017-0220-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 81.Landrum G. RDKit. RDKit Website. https://www.rdkit.org. Accessed March 3, 2025. [Google Scholar]
- 82.Sali A, Blundell TL. Comparative protein modelling by satisfaction of spatial restraints. J Mol Biol. 1993;234(3):779–815. 10.1006/jmbi.1993.1626. [DOI] [PubMed] [Google Scholar]
- 83.Waterhouse A et al. SWISS-MODEL: homology modelling of protein structures and complexes. Nucleic Acids Res. 2018;46(W1):W296–W303. 10.1093/nar/gky427. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 84.MSI Offers Free Software Developer’s Kit to Encourage Unified Drug Interface. Laboratory Network Website. https://www.laboratorynetwork.com/doc/msi-offers-free-software-developers-kit-to-en-0001. Published December 28, 1998. Accessed March 3, 2025. [Google Scholar]
- 85.Friesner RA et al. Glide: a new approach for rapid, accurate docking and scoring. 1. Method and assessment of docking accuracy. J Med Chem. 2004;47(7):1739–1749. 10.1021/jm0306430. [DOI] [PubMed] [Google Scholar]
- 86.Neves MAC, Totrov M, Abagyan R. Docking and scoring with ICM: the benchmarking results and strategies for improvement. J Comput Aided Mol Des. 2012;26(6):675–686. 10.1007/s10822-012-9547-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 87.Allen WJ et al. DOCK 6: Impact of new features and current docking performance. J Comput Chem. 2015;36(15):1132–1156. 10.1002/jcc.23905. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 88.Goodsell DS, Olson AJ. Automated docking of substrates to proteins by simulated annealing. Proteins. 1990;8(3):195–202. 10.1002/prot.340080302. [DOI] [PubMed] [Google Scholar]
- 89.Trott O, Olson AJ. AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem. 2010;31(2):455–461. 10.1002/jcc.21334. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 90.Tang S et al. Accelerating AutoDock Vina with GPUs. Molecules. 2022;27(9):3041. 10.3390/molecules27093041. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 91.Corso G, Stärk H, Jing B, Barzilay R, Jaakkola T. Diffdock: diffusion steps, twists, and turns for molecular docking. arXiv:2210.01776. 10.48550/arXiv.2210.01776. [DOI] [Google Scholar]
- 92.Li Y, Li L, Wang S, Tang X. EQUIBIND: a geometric deep learning-based protein-ligand binding prediction method. Drug Discov Ther. 2023;17(5):363–364. 10.5582/ddt.2023.01063. [DOI] [PubMed] [Google Scholar]
- 93.Wu Y, Brooks CL III. Flexible CDOCKER: hybrid searching algorithm and scoring function with side chain conformational entropy. J Chem Inf Model. 2021;61(11):5535–5549. 10.1021/acs.jcim.1c01078. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 94.Shimoda T, Ishida T, Suzuki S, Ohue M, Akiyama Y. MEGADOCK-GPU: acceleration of protein-protein docking calculation on GPUs. In: Proceedings of the International Conference on Bioinformatics, Computational Biology and Biomedical Informatics. BCB’13. Association for Computing Machinery; 2013:883–889. 10.1145/2506583.2506693. [DOI] [Google Scholar]
- 95.Wójcikowski M, Zielenkiewicz P, Siedlecki P. Open Drug Discovery Toolkit (ODDT): a new open-source player in the drug discovery field. J Cheminform. 2015;7(1):26. 10.1186/s13321-015-0078-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 96.Gentile F et al. Artificial intelligence–enabled virtual screening of ultra-large chemical libraries with deep docking. Nat Protoc. 2022;17(3):672–697. 10.1038/s41596-021-00659-2. [DOI] [PubMed] [Google Scholar]
- 97.Correia J, Capela J, Rocha M. DeepMol: an automated machine and deep learning framework for computational chemistry. J Cheminform. 2024;16:136. 10.1186/s13321-024-00937-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 98.Loeffler HH et al. Reinvent 4: Modern AI–driven generative molecule design. J Cheminform. 2024;16(1):20. 10.1186/s13321-024-00812-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 99.Wang M et al. RELATION: a deep generative model for structure-based de novo drug design. J Med Chem. 2022;65(13):9478–9492. 10.1021/acs.jmedchem.2c00732. [DOI] [PubMed] [Google Scholar]
- 100.Patronov A, Papadopoulos K, Engkvist O. Has artificial intelligence impacted drug discovery? Methods Mol Biol. 2022;2390:153–176. 10.1007/978-1-0716-1787-8_6. [DOI] [PubMed] [Google Scholar]
- 101.Pandey M et al. The transformational role of GPU computing and deep learning in drug discovery. Nat Mach Intell. 2022;4(3):211–221. 10.1038/s42256-022-00463-x. [DOI] [Google Scholar]
- 102.Martin HG et al. Perspectives for self-driving labs in synthetic biology. Curr Opin Biotechnol. 2023;79:102881. 10.1016/j.copbio.2022.102881. [DOI] [PubMed] [Google Scholar]
- 103.Konagurthu S. Accelerating Small Molecule Drug Development with In Silico Modeling: The Future is Now. Pantheon Website. https://www.patheon.com/us/en/insights-resources/blog/accelerating-drug-development-with-in-silico-modeling.html. Published September 27, 2023. Accessed March 3, 2025. [Google Scholar]
- 104.An G, Bartels J, Vodovotz Y. In silico augmentation of the drug development pipeline: examples from the study of acute inflammation. Drug Dev Res. 2011;72(2):187–200. 10.1002/ddr.20415. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 105.Gorgulla C et al. An open-source drug discovery platform enables ultra-large virtual screens. Nature. 2020;580(7805):663–668. 10.1038/s41586-020-2117-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 106.Sadybekov AA et al. Synthon-based ligand discovery in virtual libraries of over 11 billion compounds. Nature. 2022;601(7893):452–459. 10.1038/s41586-021-04220-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 107.Gentile F et al. Automated discovery of noncovalent inhibitors of SARS-CoV-2 main protease by consensus Deep Docking of 40 billion small molecules. Chem Sci. 2021;12(48):15960–15974. 10.1039/D1SC05579H. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 108.Petrecolla D. Linux: The Open Source Revolution and Its Impact on the Lives of Developers. Code Motion Website. https://www.codemotion.com/magazine/dev-life/linux-the-open-source-revolution-and-its-impact-on-the-lives-of-developers/. Published September 18, 2023. Accessed March 3, 2025. [Google Scholar]
- 109.Ponder JW, Case DA. Force fields for protein simulations. Adv Protein Chem. 2003;66:27–85. 10.1016/s0065-3233(03)66002-x. [DOI] [PubMed] [Google Scholar]
- 110.Brooks BR et al. CHARMM: the biomolecular simulation program. J Comput Chem. 2009;30(10):1545–1614. 10.1002/jcc.21287. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 111.Scott WRP et al. The GROMOS biomolecular simulation program package. J Phys Chem A. 1999;103(19):3596–3607. 10.1021/jp984217f. [DOI] [Google Scholar]
- 112.Folding @ Home. Folding @ Home Website. https://foldingathome.org. Accessed March 3, 2025. [Google Scholar]
- 113.Docking @ Home. Docking @ Home Website. http://docking.cis.udel.edu. Accessed March 3, 2025. [Google Scholar]
- 114.Muratov EN et al. A critical overview of computational approaches employed for COVID-19 drug discovery. Chem Soc Rev. 2021;50(16):9121–9151. 10.1039/D0CS01065K. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 115.Gentile F, Oprea TI, Tropsha A, Cherkasov A. Surely you are joking, Mr Docking! Chem Soc Rev. 2023;52(3):872–878. 10.1039/D2CS00948J. [DOI] [PubMed] [Google Scholar]
- 116.Target 2035. Target 2035 Website. https://www.target2035.net/contributors. Accessed March 3, 2025. [Google Scholar]
- 117.Eastman P et al. OpenMM 8: molecular dynamics simulation with machine learning potentials. J Phys Chem B. 2024;128(1):109–116. 10.1021/acs.jpcb.3c06662. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 118.Developing Medicines through Open Science (DMOS). Conscience Website. https://conscience.ca/developing-medicines-through-open-science-dmos/. Accessed March 3, 2025. [Google Scholar]
- 119.Open Molecular Software: Making Bonds. Open Molecular Software Foundation Website. https://omsf.io. Accessed March 3, 2025. [Google Scholar]
- 120.Open Free Energy. Open Free Energy Initiative Website. https://openfree.energy. Accessed March 3, 2025. [Google Scholar]
- 121.Complement Animal Research in Experimentation (Complement-ARIE) Program. NIH Website. https://commonfund.nih.gov/complementarie. Published February 14, 2025. Accessed March 3, 2025. [Google Scholar]
- 122.FDA Modernization Act 2.0. Congress.gov Website. https://www.congress.gov/bill/117th-congress/senate-bill/5002. Published September 29, 2022. Accessed March 3, 2025. [Google Scholar]
- 123.Considerations for the Use of Artificial Intelligence to Support Regulatory Decision-Making for Drug and Biological Products. FDA Website. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-use-artificial-intelligence-support-regulatory-decision-making-drug-and-biological. Published January 6, 2025. Accessed March 3, 2025. [Google Scholar]
- 124.Jayatunga MKP, Xie W, Ruder L, Schulze U, Meier C. AI in small-molecule drug discovery: a coming wave? Nat Rev Drug Discov. 2022;21(3):175–176. 10.1038/d41573-022-00025-1. [DOI] [PubMed] [Google Scholar]
- 125.Hu J. Insilico Medicine Reports Positive Phase IIa Results for ISM001–055, a Novel First-in-class Drug Treatment for Idiopathic Pulmonary Fibrosis (IPF) Designed Using Generative AI. Eurekalert Website. https://www.eurekalert.org/news-releases/1058384. Accessed March 3, 2025. [Google Scholar]
- 126.Recursion Announces First Patient Dosed in Phase 2 Clinical Study of REC-3964, a Potential First-in-class, Oral, Non-antibiotic Small Molecule for Recurrent Clostridioides Difficile Infection. Recursion Website. https://ir.recursion.com/news-releases/news-release-details/recursion-announces-first-patient-dosed-phase-2-clinical-study. Published October 22, 2024. Accessed March 3, 2025. [Google Scholar]
- 127.Notice of Funding Opportunity: CARISSMA. Chips for America Website. https://www.nist.gov/chips/notice-funding-opportunity-carissma. Published October 29, 2024. Accessed March 3, 2025. [Google Scholar]
- 128.FDA Approves Lilly and Incyte’s OLUMIANT® (Baricitinib) for the Treatment of Certain Hospitalized Patients with COVID-19. Lilly Investors Website. https://investor.lilly.com/news-releases/news-release-details/fda-approves-lilly-and-incytes-olumiantr-baricitinib-treatment. Published May 11, 2022. Accessed March 3, 2025. [Google Scholar]
- 129.Smith DP, Oechsle O, Rawling MJ, Savory E, Lacoste AMB, Richardson PJ. Expert-augmented computational drug repurposing identified baricitinib as a treatment for COVID-19. Front Pharmacol. 2021;12:709856. 10.3389/fphar.2021.709856. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 130.COVID Moonshot. DNDi Website. https://dndi.org/research-development/portfolio/covid-moonshot/. Published February 26, 2025. Accessed March 3, 2025. [Google Scholar]
- 131.Ackloo S et al. CACHE (Critical Assessment of Computational Hit-finding Experiments): a public–private partnership benchmarking initiative to enable the development of computational methods for hit-finding. Nat Rev Chem. 2022;6(4):287–295. 10.1038/s41570-022-00363-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 132.Weaver DF. Chemists invent drugs and drugs save lives. ChemMedChem. 2024;19(12):e202400074. 10.1002/cmdc.202400074. [DOI] [PubMed] [Google Scholar]
- 133.Nobel Prize in Chemistry Goes to 3 Scientists for Predicting and Creating Proteins. NY Times Website. https://www.nytimes.com/2024/10/09/science/nobel-prize-chemistry.html. Accessed March 3, 2025. [Google Scholar]

