Artificial Intelligence

Artificial Intelligence

Paving the Way for AI Drug Discovery

Paving the Way for AI Drug Discovery

Paving the Way for AI Drug Discovery

Aug 29, 2024

|

7

min read

Teal Flower
Teal Flower
Teal Flower

The Problem

Each year thousands if not millions of drugs are abandoned from discovery pipelines and millions of people remain suffering from cancer. For each successful FDA approved drug, there is a success rate of less than 15% and costs that range from seven years and one billion dollars to fifteen years and three billion dollars. As mentioned in previous articles, many drugs fail due to low efficacy, safety, lack of synchronization with business strategy, costs, and time and even repurposed drugs that have existing data struggle with time and money. Even after the drug passes the FDA approval it is under surveillance in the market by doctors and patients with various methods to report any adverse events after taking the medicine. Every 10 years (for the past 50 years) there has been a halving of pharmaceutical output per billion dollars spent on R&D, a term coined ‘Eroom’s Law’ or the inverse of ‘Moore’s Law’. This is an inverse relationship that expresses that as technology is looking to be doubled every year, Pharma is experiencing the opposite trends. However, with proper consolidation of technology and biopharma, there is potential to prevent Eroom’s Law from creating an irreversible bottleneck in the industry. Year after year, pharmaceutical companies are abandoning drug pipelines even after billions of dollars have been poured in, and because of the high costs of re-evaluation, these drugs are not usually investigated again. However, with the advent of AI-capable data and computer infrastructure, Revilico has leveraged AI modeling and physical simulations to reduce costs and time of re-evaluation. This allows for abandoned drugs to be revitalized to their full potential, something not previously possible.


What inspired Revilico?

Revilico was founded as an AI-driven biotechnology consulting company after the founders recognized an opportunity to streamline the commercialization of technologies being developed at academic labs at UCLA. The company's vision was to leverage AI to automate essential but time-consuming processes such as grant writing, developing commercialization strategies, patent documentation, and building the necessary team and infrastructure for manufacturing. By integrating AI into these critical steps, Revilico, previously Nexabio Venture Solutions, aimed to bring innovative technologies to market more quickly and efficiently - though this business model was set to pivot quite dramatically. Now, 8 months into Revilico’s development, our team of engineers has pushed the boundaries of what is possible in AI driven drug discovery and development with the conception of our first proprietary generative chemistry and screening engine - creating the capabilities of screening almost 2 million molecules for one disease target alone. Given that the possible set of compounds that can physically exist spans to a space of 10^60,  and that traditional pharmaceutical discovery has trouble investigating this wide ocean of molecules that could potentially become life saving compounds - Revilico sees the opportunity to pave the way for the new era of pharmaceutical development. We do this by diving deeper into what is possible, uncovering solutions to diseases still left uncured, with molecules still yet to be uncovered.

Revilico Inc.

Revilico Inc. is a start-up company entering the pharmaceutical industry, leveraging AI algorithms to repurpose abandoned small molecules into potent cancer therapeutics. Aiming to create a solution to minimize wasted drugs, with the goal of reducing the time required for drug commercialization by 3-6 years and costs by $300 million to $500 million in R&D costs, Revilico uses physics driven AI models to accomplish this goal. Our timeline shows the difference between the standard process, showing two stages that are now nonessential, with the assistance of AI throughout the process. We currently have a small molecule development focus with potential for expansion into gene therapies and biologics to push the whole therapeutic space forward.  Our primary product, a small molecule engine, is currently in finalization stages and currently will generate and screen compounds down to components that can be synthesized to ensure computational design translates to the wet lab, a huge differentiator of our product.


Using datasets across 8 million chemical samples, we are able to screen through millions of drug candidates to identify possible drug prospects for synthesis and wet lab testing on particular protein targets. This assists in the early stages of traditional drug discovery, from generating candidates to lead optimizations. Moreover, our company prides itself on the duality of disease modeling and chemical analysis with a proprietary disease model that has been completed as well. Inputting millions of cell samples across different ‘omics’ layers, our team has developed an interface to run automated disease data analysis to characterize the leading causes of diseases in a quantifiable and engineerable way. This allows for a target selection criteria to be developed for the sake of disease target identification, like a gene overly expressed in cancer, which will subsequently be passed into the small molecule model to generate compounds that can effectively halt disease progression.


We estimate that by using our product, the time and costs spent within the first two stages of drug discovery can be cut by approximately 40% and 30% respectively as expensive wet lab tests that lead to failure can be replaced with in-silico simulations before hand. The product will always eventually need wet lab confirmation and analysis, but instead of ‘looking for a needle in a haystack’ , we provide the alternative pathway of ‘generating a brand new needle’ to allow for more directed wet lab approaches. These physics driven algorithms are all proprietary and span a wide array of computational models ranging from predictive and generative AI, to in-house developed mathematical equations for the purpose of simulating the biological microenvironment of tumors, to drug-disease interactions and fluid dynamic simulations. We gather our data using four methods: data mining, in house data collection, synthetic data generation, and acquired data from partners - which is still in development.

What makes Revilico Different?

Compared to our competitors, Insilico, Atomwise, Roivant, BenevolentAI, Recursion, and other large pharmaceuticals, we boast cancer and repurposing specialization, de novo discovery, and we pride ourselves on being able to translate computational design to the wet lab. Because of the difficulty of generating a symbiosis between biology, chemistry and computational sciences - our team is finely crafted to tackle not only the computational developments but their translations to the wet lab, cutting costs and time at all facets. Revilico boasts a diverse and highly skilled team, including AI engineers, data engineers, research analysts, and industry experts. We are supported by a board of advisors and partners with extensive experience in medical informatics, pharmaceutical development, chemical engineering, and established business consultants.  The market we are targeting for our Total Addressable Market (TAM) is the Drug Discovery AI market valued at $1.2 billion with a 29.6% Compound Annual Growth Rate (CAGR) according to Grand View Research. We narrowed this number down to about $400 million for our Serviceable Addressable Market (SAM) and our Service Obtainable Market (SOM) is estimated at $120 million. However, to expand this market further we are looking to not only penetrate into the AI driven discovery ecosystem, but the pharmaceutical and biotechnology industry both towering at a $1.56 Trillion and $1.77 Trillion valuation respectively. In terms of our business model, we are looking for strategic partners for development and operate functionally in two ways: 1) licensing for technology usage and 2) milestone payments to incentivize success from both parties involved. We are currently looking to secure strategic partners who are looking to develop drugs with the supplement of Revilico’s AI drug discovery tools and our team of internal engineers with experience in building customized AI models for predictive or generative analytics.


What’s Next For Revilico

Revilico is looking to expand and iterate upon its already established computational molecular screening platform by validating our modeling capabilities in the wet lab. Moreover, after completion of the full model, customer insight will be driving an iterative process amongst our internal engineers. In parallel, expansive R&D will be continued into the biologics and gene therapeutic market prospectively for generative and predictive applications. As our business model drives revenue forward through a systematic software integration to pharmaceutical developers, eventually Revilico will expand into developing an internal drug pipeline for the sake of targeting triple negative breast cancer and other disease indications within the cancer umbrella. The possibilities for expansion and automation of our disease model pipeline for target selection are truly limitless up to the point of data quality and availability. As Revilico drives the success of pharmaceutical programs, cutting costs and time of development, we pave the way for industry expansion and a boast competitive advantage over traditional pharmaceutical development. This expansionary process also leads to the  democratization of the industry, cutting down therapeutic costs for those suffering with cancer and other diseases. Software screening allows for a more accessible route to pharmaceutical discovery and development for companies in the space, creating more room for innovation to flourish.