Advancing antibody cancer therapeutics with AI

Antiverse is an artificial intelligence (AI)-driven company that specialises in antibody design against difficult-to-drug targets, including G-protein coupled receptors (GPCRs) and ion channels. Murat Tunaboylu, Co-founder and CEO of Antiverse, speaks with Megan Thomas about how AI can advance therapeutic antibody design.

Megan T: How is Antiverse using AI in cancer therapy?

Murat T: Antiverse is using generative AI to design antibodies for challenging targets associated with more than 60 diseases, including oncology-based immune checkpoint targets. Using machine learning (ML), we design target-specific libraries against immune checkpoint targets, as well as many other target types. These libraries emphasise high diversity, explicitly tailored to the target of interest.

In screening a smaller subset of the antibody sequence space with high confidence against the desired target, we enhance the probability of finding functional binders; this saves time and resources. After several rounds of biopanning, we sequence the outputs which include both positive and negative binders. We then perform multi-parameter AI analysis to cluster the sequences and select diverse antibody candidates with the desired properties.

Megan T: What is Antiverse’s approach to therapeutic antibody design?

Murat T: Traditional antibody discovery involves screening of antibody ‘libraries’ containing random antibodies to find a ‘hit’, or an antibody which binds the target. Our algorithms design ‘target-specific’ libraries – designed specifically against the target of interest, significantly increasing the likelihood of finding functional binders.

This allows Antiverse to work with particularly challenging targets, where we strive to make the most impact.

In addition, especially in the case of GPCRs and other challenging targets, receptors are found in very low quantities. When screening against cell lines, many functional binders may be missed. To combat this, Antiverse has developed hyperexpressing cell lines that express over of a million receptors on the cell surface, compared to traditional methods which only express around 10,000. This ensures that most functional binders are identified through subsequent screening rounds.

Megan T: What challenges can this AI tech overcome?

Murat T: Antiverse is solving the challenge of difficult-to-drug targets, like GPCRs and ion channels. Functional antibodies against these targets are rare, meaning random-sampling methods, including screening antibody libraries, often fail. The result is that many therapeutic targets are left undrugged despite enormous research efforts. The problems with discovering antibodies against GPCRs can be broken down into two analogies:

  1. The iceberg problem, which represents the small portion of the exposed GPCR receptor available for antibody recognition.
  2. The forest problem illustrates the challenge of identifying a specific receptor within a dense and complex receptor landscape. We are tackling these barriers with a hybrid approach, optimising for antigen recognition. We use ML to design target-specific libraries, generating a diverse repertoire of antibody candidates targeting the exposed epitopes. We then screen these antibodies against hyper-expressing cell lines – converting the ‘forest’ into more of a dense monoculture to facilitate binding events, boosting the likelihood of discovering functional binders.
Megan T: When has Antiverse’s AI tech led to the development of novel therapies?

Murat T: Our current lead candidate antibodies have shown promising results against PAR1, a GCPR linked with thrombosis in disease settings. IC50 determination curves revealed six of our initial binders have nanomolar affinity. Then, our first round of ex-vivo thrombosis studies in mice showed efficacy equivalent to vorapaxar, an antiplatelet therapy used for thrombosis. Whilst our most developed asset is just reaching pre-clinical stages of development, we are confident our technology will impact many patient lives in the coming years, through a combination of our own in-house efforts and our ongoing partnerships with major pharma companies.

There are currently only three FDA-approved antibodies against GPCRs, which highlights a highly underserved therapeutic space many fear entering or re-entering because of a history of low success rates or because traditional techniques, up until recently, have been largely unsuccessful. AI offers an exciting new technology to augment the antibody discovery process, and Antiverse aims to engineer this technology into this highly neglected area and open up the antibody druggable space.

Megan T: Can Antiverse’s AI balance out the trade-off between specificity and affinity?

Murat T: The quality of the data input for training AI models directs the quality of the output. So far, traditional antibody design techniques have struggled to balance specificity and affinity. As a large subset of training is based on pre-existing antibody data, our AI-designed target- specific libraries face this same challenge. However, by utilising an innovative antibody library design process, we account for the entire antibody sequence space whilst only focusing our attention on evaluating binders with high confidence. In the same way, this method improves our likelihood of finding functional binders, we also improve the likelihood of discovering antibodies with an optimal balance of affinity and specificity.

Megan T: Can you elaborate on Antiverse’s recent collaboration with GlobalBio?

Murat T: Our co-development aims to continue to progress the most promising candidates further into pre-clinical studies. Our initial success came from combining our ML models and naive AI-generated antibody libraries with GlobalBio’s high-grade antibody libraries. In screening both libraries for candidates, we found a panel of highly specific anti-PD-1 antibodies. Whilst many therapeutic checkpoint inhibitors targeting PD-1 exist in Western markets, South America is a growing market where lots of value can be added. Antiverse focuses its efforts where the potential for impact is greatest, so upon learning about the opportunity from GlobalBio, our team was fully committed.

DDW Volume 25 – Issue 1, Winter 2023/2024 – Therapeutic Antibodies Guide


Murat TunaboyluMurat Tunaboylu is CEO and Co-Founder at Antiverse. With a background in software engineering and bioinformatics, Tunaboylu has built cell imaging software and lab robots to accelerate cancer research and automated Thermo Fisher Scientific’s gene synthesis workflows. Tunaboylu studied Computer Engineering at Bahçeşehir Üniversitesi and has a degree in Electrical Engineering from Yildiz Technical University, Turkey.

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