In the world of artificial intelligence and machine learning, Google Deepmind’s AlphaGo program beats world champion Lee Sedol in the game go March 19, 2016, was a watershed. It shows, among other things, that an artificial intelligence algorithm can beat the most advanced neural network: the human brain, if it is trained on a large number of data points and iteratively improves.
Biotech startup Anagex promises to bring to the world of small molecule drug discovery massive in-house developed compound datasets that can be queried with AI. By combining large-scale laboratory experiments with computational tools, it hopes to upend the drug discovery game that requires infinite patience and even more dollars. The Boston-based company announced Wednesday that it has raised $30 million in a Series A round to move closer to its ultimate goal of developing life-saving medicines for previously undruggable targets. Catalio Capital Management co-led the round with Lux Capital, Khosla Ventures, Obvious Ventures, Airstreet Capital and Menlo Ventures. The previous seed round was $7.2 million.
“We saw a lot of platform technology but were struck by the potential of Anagex to fundamentally reshape the way small molecule drug discovery is done,” Catalio general partner George Petrocheilos said in a press release announcing the fundraising. “Pursuing goals that have disappointed the industry for decades is always a risky business. The power and efficiency of the Anagex platform makes that risk bearable, especially given the potential rewards.”
What are the powerful features of this platform?
In a lengthy email response to questions, Anagex CEO Nicolas Tilmans argued that other companies calling themselves AI drug discovery companies have only a fraction of the datasets they can use to train their AIs. Furthermore, they rely on external datasets, often from poor-quality clinical research organizations. Tilmans claims that “even though they are of high quality, traditional drug discovery datasets are 100-1000 times smaller than what is required to leverage modern ML methods,” while Anagex can “generate appropriately sized datasets (billions of data points) .”
Another key difference, he believes, is how the company tests initial populations of compounds and feeds the information from those tests back to the AI engine to learn from it.
“We can iterate cheaply by rapidly synthesizing and testing millions of compounds in parallel to further update our ML models. This iterative process between predictions and real-world measurements, also known as active learning, is similar to DeepMind’s The method used to train the AlphaGo model that beat Lee Sedol at Go,” explained Tilmans. “Our lab was built from the ground up to make compounds faster and cheaper than any competing technology, while improving data quality.”
In announcing the investment round, the company noted that since the start of operations in the fall of 2020, Anagex has established a special biochemical laboratory that can simultaneously use technologies such as DNA Encoding Libraries (DEL) and Affinity selection mass spectrometry. DEL is a technique used by the pharmaceutical industry to discover small molecules that can affect biologically relevant targets. Affinity selection mass spectrometry is a method for screening large numbers of compounds.
By definition, AI/ML algorithms are designed to answer questions. If you show an AI algorithm enough pictures of cows, eventually when you show a picture of an animal, it can tell if it’s a cow. Anagenex’s artificial intelligence is also designed to answer the question, albeit a more complex one. Tilmans again, explaining what Anageneex’s AI is predicting and what’s next:
“Does the compound bind to the target, but not to a closely related target that we don’t want to interfere with? After subsequent experiments, update the model to see which compounds have biochemical effects. The model attempts to follow the order from best to worst Arrange compounds, not just learn binary “binding” vs. “not binding”.
Tilmans knows all too well its competitors in the space, big and small.He lists some of them, some of them are Teri Therapeutics, all drug, 1859 Company , relay therapy, introduce — and in any case, he is unsurprisingly convinced that Anagex is better.
However, in response to another question, he did acknowledge the strength of the two public companies.
“Very, very few small-molecule drug discovery companies have large-scale proprietary datasets combined with the computational talent needed to make the most of those datasets. Among the public companies, there are only Recursion and Exscientia. [Chris Gibson, the CEO of Recursion and Richard Law, chief business officer, Exscientia are both taking part in a panel discussion about the state of AI in drug discovery moderated by MedCity News’s senior biopharma reporter Frank Vinluan at our INVEST PharmaTech virtual conference available on-demand on July 26]
recursive clinical pipeline Includes clinical-stage drug candidates for the neurological diseases cerebral cavernous hemangioma and neurofibromatosis type 2. The Salt Lake City, Utah-based company is also developing preclinical programs for other diseases: Batten disease, Tay-sachs disease and hereditary hemorrhagic telangiectasia.
The UK-based Exscientia platform has produced seven drug candidates, three of which are in Phase 1 testing. One of its most advanced internally developed drug candidates, EXS21546, is an immuno-oncology drug.
Anagenenex also has “several initial programs focused on precision oncology and cardiovascular disease,” Tilmans said.
Like many AI discovery companies leveraging machine learning, the startup is looking for proteins/targets that have been challenging for drugs in the past but have a high disease burden.
Anagenenx is still in the early stages of the drug discovery cycle, and it may be several years before a drug reaches the clinic, but Tilmans is confident of future success.
“Our platform is a unique synthesis of innovative large-scale laboratory experiments and machine learning,” he said. “Lab experiments alone are slow and expensive, but successfully bring drugs to market. Machine learning is fast and cheap, but difficult to bring drugs to market. By combining the strengths of each, we create a unique cost-effective Beneficial platform to find new medicines.
A current investor was sold based on this concept.
“At Lux, we have repeatedly seen how combining state-of-the-art computational tools with custom laboratory operations can transform drug discovery and bring innovative medicines to patients,” said Zavain Dar, venture partner at Lux Capital. “However, few companies have blended the two as seamlessly as Anagex, and we’re happy to have been there from the start.”
Clearly, investors in Anagex sold off on its vision. Whether it will deliver will only be clear on the long road to commercialization. The first compounds are expected to enter clinical trials within two years.
