The next generation of AI


Reece Armstrong speaks to Sean McClain, Founder & CEO of Absci about the rise of generative AI in life sciences. The company has a biologics pipeline with a focus on differentiated antibodies targeting a range of indications. 

Last year, a lot of focus was placed on artificial intelligence (AI) both outside of the life sciences sector, with the rise of Chat GPT, and inside, following a growing interest in the use of generative AI for drug discovery and development. 

AI is undoubtedly one of the key technologies that is set to impact drug discovery and development practices. In the Wellcome “Unlocking the potential of AI in drug discovery,” report, 84% of current AI users within the industry, as well as 70% of non-users, stated that AI will drive significant impact within drug discovery in the next five years.

AI has seen major advances in recent years, with areas such as deep learning, language modelling, vertical AI, and of course, generative AI, offering users more options to apply to their research.

On that generative AI note, Sean McClain, Founder and CEO of Absci, says that what Chat GPT did for the industry was really put a “focus on the importance of generative AI.”

Absci has been working within this space for a while, having recently signed deals with AstraZeneca and Almirall to discover and develop targets within oncology and dermatology, using its generative AI platform and wet lab capabilities. The company has a pipeline of three drug candidates, with its lead compound ABS-101 having recently started IND-enabling studies. ABS-101 is the company’s lead asset developed using AI capabilities and is focused on cytokine biology for the treatment of inflammatory bowel disease. 

McClain says that now that the interest in generative AI has been established, companies will now be questioning how to integrate it into their own teams. 

“Do they build it out internally or do they partner?” he asks. 

Towards the end of last year, Merck partnered with BenevolentAI and Exscientia on two strategic drug discovery collaborations that will use AI design and discovery capabilities in an effort to generate drug candidates across oncology, neurology and immunology.

On the generative AI side, last year saw Genentech partner with NVIDIA to use the technology to discover and develop new therapeutics.

It’s obvious then that pharma is interested in how best to use AI and companies will be searching for any advantages that it can bring to discovery and development, where projects are often expensive, time-consuming and risky. 

For McClain, he believes that ultimately, data is the key if companies are to start getting the outputs they want. 

“If you look at Chat GPT, it was trained on the whole internet. And if you look at the publicly available data that exists for biological data, it’s very sparse, or it’s really a drop in the ocean compared to the overall data of the internet.” 

He says that it’s not enough to simply have an AI model to generate potential drug discovery routes, but that teams need a way to validate hypotheses in order to back-up the model’s theory, and also to alleviate time and cost pressures for the team exploring the option that’s been provided by the AI system. 

“The winners in the space are those that have figured out how to successfully integrate scalable wet lab data with generative AI,” he says. 

“It’s important we figure out how to get high quality and high throughput data to train your models. But not only training those models, but then going into the wet lab and validating the results to ensure that the model is doing what you say it’s going to do.” 

In fact, access to infrastructure was one of the key challenges raised in Wellcome’s report, that states that a lack of drug discovery capabilities such as supporting wet lab capabilities to test model hypotheses, is one of the major barriers to current AI-related drug discovery efforts.

McClain attributes this integration of generative AI and wet-lab capabilities as an important loop that the industry hasn’t been able to have until recently. 

“We’re using generative AI to design molecules that we can test the hypotheses that we have of biology. But we’ve never been able to be in that kind of design and test loop where it’s like, I have this hypothesis, I believe that if I design an antibody that has these attributes, it’s going to solve my biology. We’ve never had that ability to specifically design molecules the way we’ve wanted to, we’ve always had to rely on this trial-and-error process. And so I think this is a big step forward. I will say it is going to transform healthcare, faster than we anticipate.”

The use-case of AI is varied across pharma. Currently, the industry has a lot of companies targeting discovery and development with the technology, but as of writing this no AI-designed molecule has been brought to market. Most notably, last year saw Exscientia tighten its focus on precision medicine, ending development on its A2A receptor antagonist (EXS21546) which was targeting patients with immunotherapy relapsed or refractory renal cell carcinoma (RCC) and non-small cell lung cancer (NSCLC). 

BenevolentAI also released mixed results for its atopic dermatitis candidate, with the therpay BEN-2293 being safe and well tolerated, but ultimately failing to meet secondary efficacy endpoints of reducing itching and inflammation. 

These failures were perhaps given more attention considering the talk surrounding AI’s potential, but it must be said that drug discovery is a very risky endeavour, and with the majority of drug candidates failing to make it market, it shouldn’t come as a surprise when results aren’t quite as positive as they could have been. 

For McClain, he looks at what’s happening within the industry through a collective lens.  

“I’m rooting for the whole cohort of AI drug discovery companies. I want every single one of them to be successful. It’s a little bit like we all win and fail together, because I think things can very rapidly change if you have multiple programmes that fail in the clinic.”

McClain is right to be aware of the failures in the market, especially when companies are all vying for funding in a highly contested space. AI is currently a hot topic and investment has been strong within the last decade. Wellcome’s report states that there has been a total of $18 billion invested in AI-first biotechs within the last 10 years, however 60% of this investment has been directed at the top 20 companies within the sector. With investors ultimately wanting to see results from where their money is going, the pressure is on for the AI sector to deliver molecules to the market. 

Last year, McClain was invited to the Second Bipartisan Senate Forum on Artificial Intelligence. There, a group of experts discussed the various facets relating to AI, including policy making decisions for the US government, and the opportunities that AI presents to various industries. 

He says that for those invited, they all agreed on crucial aspects such as talent and infrastructure in bolstering AI, but McClain argued that the other piece of the AI puzzle is data and figuring out how to scale biological data specifically.

One of the biggest challenges highlighted in Wellcome’s report was the lack of access that researchers currently have to raw datasets that are appropriate to run and train models. Experts in that report stated the need for deep, broad datasets that are focused on a disease level, as well being able to access data that can be used to train AI models. 

There are initiatives out there that are aimed at enabling greater access to data. The Wellcome-Sanger African Genome Variation Project is laying the foundations for generating high-quality genomic datasets in Africa and grants from the US National Institutes of Health can give teams access to standardised existing datasets that can advance their machine   learning models. 

However, according to McClain, these open source data platforms aren’t enough.

At the Senate Forum he argued for the support of the biotechnology community through pooled proprietary datasets that will help “set the US apart when it comes to scaling biological data for AI use.” 

He gave the example of Absci’s own work in building on public data sources through experimental wet lab data using the company’s synthetic biology platform. That data is used to design and run experiments in silico in an effort to speed up the discovery process. 

That initiative to speed up the discovery process is key to AI’s current capabilities in the pharmaceutical sector. It’s thought that when using AI, the time and cost savings could be as high as 50% compared to the current economic impact of standard drug discovery practices1.

There’s a lot to consider going forward with AI but in the time speaking to McClain it’s clear that there is excitement within the sector, particularly regarding the potentials of generative AI and drug discovery. 

“We’re understanding that generative AI is going to have a huge impact but it’s going to take time. It’s not going to be a one-size-fits-all scenario that fixes every problem, but it’s going to make a big impact in drug discovery.” 

For Absci, the company recently presented pre-clinical data at J.P. Morgan for its ABS-101 anti-TL1A antibody programme. 

“We were able to show in a 14-month period we could get to in vivo validation and get to a drug candidate,” McClain explains, highlighting the potential of a generative AI platform in drug discovery. 

It was also the first time a company had shown it could create a differentiated antibody using generative AI and whilst McClain states there’s still a long way to go, he’s excited by the prospects that this represents. 

This pipeline data, coupled with a partnership with AstraZeneca and Almirall, hopefully bear success and drive forward AI’s potential in drug discovery.

DDW Volume 25 – Issue 2, Spring 2024



Sean McClainBiography:

Sean McClain is the Founder and CEO of Absci, a generative AI drug creation company on a mission to create better biologics for patients, faster. McClain embodies the entrepreneurial spirit, having built Absci from a basement lab to a publicly traded company and continuing to strive towards his vision to create breakthrough therapeutics at the click of a button. 

Related Articles

Join FREE today and become a member
of Drug Discovery World

Membership includes:

  • Full access to the website including free and gated premium content in news, articles, business, regulatory, cancer research, intelligence and more.
  • Unlimited App access: current and archived digital issues of DDW magazine with search functionality, special in App only content and links to the latest industry news and information.
  • Weekly e-newsletter, a round-up of the most interesting and pertinent industry news and developments.
  • Whitepapers, eBooks and information from trusted third parties.
Join For Free