Several facets of small molecule drug design have already been altered by artificial intelligence (AI) and deep learning. More difficult tasks, like the design and optimization of antibodies, are now now being impacted by recent advancements in the field. As a result, a breakthrough occurred in 2022 when the first antibody to be computationally developed entered a clinical trial and piqued the interest of biopharmaceutical companies. The antibody was created by Israel-based Biolojic Design.
Using AI to speed up antibody discovery
In the past, experimental methods like hybridoma screening or high throughput platforms like yeast or phage display were used to find the first mAbs (hit identification, or HI). To enhance the binding and other features, the technique is followed by computationally guided mutagenesis or evolution of the antibody sequence (hit to lead, or H2L, optimisation).
Although the field is still in its infancy, numerous AI companies have achieved significant advancements in recent years. The advantages of AI-guided evolution of hit for improvements to binding, solubility, yield, immunogenicity, etc. are currently sufficiently demonstrated by case studies. As a result, start-ups are more likely to offer lead optimization or molecular evolution services in the space. A small number of businesses are also working on de novo design of the candidates just from the target/antigen, omitting the requirement for time-consuming and expensive experimental approaches, which might be a game-changer.
Here are a few instances of how businesses are utilising AI in various ways to speed up the discovery of antibodies:
1) Generate Biomedicine: Founded in 2018, this Boston-based pioneering firm has already raised $420 million. De novo design of a range of proteins, including antibodies, is done by Generate Bio using a combination of sequence- and structure-based methods. The business also has its own automation platform to quickly synthesise and test antibodies in iterations, in addition to its unique generative AI.
2) BigHat Biosciences: The California-based business, founded in 2019, has raised around $100M to date. They have created an integrated platform that integrates cell-free high throughput synthesis with antibody testing proposed by their AI platform for antibody optimisation.
BigHat employs a sequence-based methodology that needs hit data or the original sequence of antibodies found using other techniques. After that, the initial antibody sequence is run through machine learning (ML) algorithms to optimise aspects like binding to target, solubility, immunogenicity, yield, and other factors. Each test cycle’s data is used to update the AI/ML models for the following iteration.
In 2022, BigHat purchased Frugi Biotechnology, a business that was creating cell-free protein synthesis (CFPS) technology that was both affordable and of excellent quality. They now have five internal and joint programmes in the works.
3) Biolojic Design: The Israeli startup, founded in 2009, employs a structure-based method to find a template antibody against the desired target from the current human antibody pool. Their model is trained on millions of Antibody-Antigen pairs. The discovered template is then subjected to guided evolution using a different machine learning model to anticipate mutations and enhance affinity and other biophysical features.
The three businesses mentioned above are among the most advanced, although the market is expanding quickly. Here is a lengthy list of biotech companies in the AI antibody market:
The application of AI to antibodies is gaining ground.
Biopharma businesses have begun investigating partnerships with a small number of companies as well as developing internal skills after realising the potential of these platforms. Amgen and Generate Bio collaborated on the discovery of five targets, with the potential value of the contract being up to $1.9 billion plus royalties. Chugai released encouraging findings from their in-house MALEXA-LI AI platform for antibody discovery. Genentech bought Prescient Design in August 2021. Along with Amazon Web Services, Inc. (AWS) and the Israel Biotech Fund (IBF), four major pharmaceutical companies—AstraZeneca, Merck, Pfizer, and Teva—have also formed AION laboratories, an incubator to speed up AI-driven antibody discovery.
Not just biopharmaceutical companies are closely monitoring this market. Large AI businesses and CROs have already begun to engage in M&A activity in the field, with numerous early-stage start-ups being acquired. 2019 saw the $90 million acquisition by Evotec of the AI/ML antibody discovery start-up Just Biotherapeutics. AbSci, a platform for experimental antibody discovery, purchased Denovium Inc., a deep learning startup formed in 2018, and Totient Bio, a firm founded in 2017, both of which had big datasets for training AI/ML models.
AI-small molecule discovery spaces versus AI-biologics
This sector is developing differently from the small molecule AI drug discovery environment. As opposed to simply in-silico platform companies, “closed-loop” businesses that combine high throughput experimentation with AI technology are more common. The majority of these firms are now concentrating on internal pipeline as a result of learning from the development of the small molecule AI field and realising the enormous advantages those assets may provide over a services-only strategy.
Furthermore, given the difficulty of developing AI models for complicated compounds like antibodies, the number of players is probably substantially lower than in the small molecule AI field. These “closed-loop” businesses are all the more appealing due to the lack of datasets because they continuously produce their own data to enhance the forecasts.
Recent achievements include a number of sizable acquisitions involving pharma and AI businesses, and exit possibilities through M&A have encouraged investors to learn more about prospective investment prospects in the sector. But it will take three to four years for the AI-biologics discovery field to reach the same degree of maturity as the AI-small molecule discovery space. However, these are exciting times for biopharma since these platforms are changing the status quo and how we find new medications.
