Our experienced scientists and engineers optimize all phases of drug-discovery, including early hit identification, hit-to-lead, lead optimization, patent strategy, and preparation for IND filing, yielding shorter timelines to develop the highest-quality drug candidates.
We are using AI-driven drug discovery to ensure better molecules and faster cures. Our mission is to discover, develop and deliver to patients novel, small molecule drugs that were previously undruggable or undiscovered until our technology came along.
How DeepCure finds drug candidates among trillions of molecules
10^18 synthesizable molecules
Co-optimization of 50 drug properties for every entry in MolDB
5000 molecules synthesized & 100,000 datapoints generated every month
Discover candidates that don’t currently exist
DeepCure’s proprietary molecular database (MolDB), is the world’s largest medicinal chemistry database — that covers over one trillion compounds and is expanded on demand to one Quintillion molecules. These compounds are highly diverse and span most of the chemical space currently accessible to chemists. The trillion compounds cover more than 75% of the existing medicinal chemistry compounds, including drugs and current clinical candidates in development, synthesized or reported by medicinal chemists.
In addition, a suite of generative models can expand each selected molecule to a million analogs – expanding the search even further to one Quintillion (10^18) molecules.
MolDB enables the discovery of candidates that don’t exist in traditional screening libraries.
MolGen: A Machine Learning Algorithm Trained to Learn Organic Chemistry
MolGen is the first algorithm to fully account for synthesizability while supporting the creation of chemical space in the trillions scale.
Trained on a proprietary dataset of chemical reactions, MolGen learned the organic chemistry fundamentals that make MolDB, and enable DeepCure to explore almost every possible drug candidate and select the optimal ones.
DeepCure is charting a new path in drug development using AI-based drug design and discovery.
With each of our discovery programs, we scan our entire database for optimal candidates for our pipeline. This is made possible using our machine learning algorithms and our generative expansion that expands the effective search to one Quintillion molecules – orders of magnitude beyond what’s currently possible.
Advantages to Exploring New Areas of Chemical Space
In more than 100 years of drug development, the pharmaceutical industry has developed just over 2600 approved-small molecule drugs.
In 3 years, DeepCure has built the technology to discover the next 1000 and developed more than 30 lead compounds for multiple high-value therapeutic targets.
DeepCure's AI Engine
Our AI Engine is a platform of predictive models, optimization frameworks and an active learning workflow. In simple terms, we use AI to predict and optimize a drug’s most important properties, such as potency, selectivity, toxicity, to name a few.
Discovering the Unknown. Accuracy. Scalability.
Unlike other approaches, our AI engine has the proven ability to generalize to truly novel chemical spaces.
Our state-of-the art machine learning (ML) models were designed to overcome the most challenging task for ML in drug discovery – generalizability (aka discovering the unknown). Traditional methods and most algorithms using machine learning can only accurately predict properties of molecules similar to known entities – which of course, have little to no use in discovering novel drugs.
In contrast to traditional approaches that struggle to accurately predict properties of molecules different than the known data — our models leverage a proprietary molecular representation to demonstrate superior performance and an unprecedented ability to generalize to previously unseen and unexplored parts of chemical space. This is possible due to DeepCure’s unique ML algorithms we have pioneered and developed in-house.
Value In, Value Out
Our AI team has designed a set of statistical methods to clean the data, including a process to normalize and filter data points from the different sources by synthesizing and testing compounds in our internal assets. This effort transforms the low-quality public data into a high-quality, proprietary dataset at DeepCure.
The proprietary representation enables the generalizability and accuracy of our predictions, as well as the ability to search trillions of molecules for each target of interest.
Target and Indication Selection
Our technology is indication-agnostic, which gives us the unique ability of identifying targets with the best fit based on unmet medical need, market size, biological validation and technological fit.
Every target and candidates we develop has been selected based on their potential to improve patient outcomes compared to currently available therapies on the market. We focus on unmet medical need, and prioritize candidate selection based on this criteria. Our “super power” with discovering the unknown, allows us to discover drug candidates with optimal profiles with complete freedom to operate.
Leveraging cutting-edge robotics coupled with our leading software platform, our Molecular Foundry allows synthesis and biological testing of thousands of novel and diverse drug candidates.
At DeepCure, we’re doing things differently – we’re building the lab of the future. Our data generation lab leverages automated synthesis and robotic testing to create a continuous feedback loop between the AI and the wet lab. The scale, quick turnaround time, and high-level data quality that we require, can only be achieved by creating our own fully-automated, robotic in-house wet lab powered by AI.
This allows us to develop our pipeline of novel candidates in the most cost-effective and faster way, with continuous improvement of our ML pipeline.
Using AI and cutting-edge automation, we can reduce cost and cut cycle times. The chemistry is based on robotic synthesizers, MS-triggered purification systems, and liquid-handling systems, while automated screening and in-vitro testing generates up to a 100,000 datapoints per month.
This AI-to-Bench integrated workflow will ensure that novel in-silico target discoveries are immediately and directly validated in-vitro and fed back into the AI pipeline to optimize for success. This streamlined process shortens cycle times, increases efficiency and reduces costs.