Giving Failed Drugs a Second Chance: How Ignota Labs Is Using AI to Transform Drug Discovery
Could the future of drug discovery lie in revisiting medicines that have previously been abandoned?
Drug discovery is a complex, expensive and high-risk process. Despite significant investment in scientific research and development, approximately 90% of drug candidates never reach patients.1
Ignota Labs also estimates that failed drug programmes represent a substantial loss of value across the pharmaceutical industry, with safety failures of small molecule drugs contributing to more than $400 billion annual loss.
But what if some of these failures represented opportunities rather than the end of the road?
At ELRIG’s Drug Discovery event, we spoke to Dr Layla Hosseini-Gerami, Co-Founder and Chief Data Science Officer at Ignota Labs, about how artificial intelligence (AI) could:
help rescue failed drugs,
address toxicity challenges, and
accelerate the development of safer medicines.
Rethinking Drug Discovery: Why Failed Drugs Deserve a Second Chance
Rather than starting the drug discovery process from scratch, Ignota Labs focuses on identifying promising drug candidates that have been previously discontinued due to safety concerns and investigating whether these drugs could be chemically modified to overcome these challenges.
As Layla explains, the opportunity lies in unlocking one of the pharmaceutical industry's most underutilised resources: the scientific data generated by failed drug development programmes.
These abandoned assets have often already benefited from substantial investments of time, funding and scientific expertise. Importantly, many have demonstrated promising properties that enabled them to progress through clinical development. By revisiting these candidates, Ignota Labs builds on existing scientific knowledge rather than allowing potentially valuable research to be lost.
Ignota Labs describes this approach as drug re-engineering rather than simply drug repurposing. Its model involves:
identifying abandoned assets,
developing improved chemical structures to address their limitations,
and progressing them towards clinical development.
From Scientific Frustration to Innovation: The Story Behind Ignota Labs
Ignota Labs was co-founded by three individuals bringing together expertise in drug discovery, commercial development, and artificial intelligence.
The idea originated with Dr Jordan Lane, the company's Chief Scientific Officer, who had witnessed numerous promising drug discovery programmes fail despite having strong underlying science.
Convinced that some of these candidates deserved another opportunity, Jordan approached Sam Windsor, MMath MBA, a university friend with a business and commercial background in healthcare and biotech sectors, to explore establishing a company focused on reviving failed medicines.
Recognising that AI and machine learning could provide the technological foundation for this approach, they subsequently approached Layla, who was completing her PhD at the time.
What followed was a period of building the company alongside existing commitments, with evenings and weekends dedicated to fundraising and establishing the business.
Today, the three co-founders spearhead Ignota Labs: Sam Windsor, MMath MBA, as Chief Executive Officer, Dr Jordan Lane as Chief Scientific Officer, and Dr Layla Hosseini-Gerami as Chief Data Science Officer.
Why Do Promising Drugs Fail? Understanding Drug Safety and Toxicity
One of the most significant challenges in drug development is understanding why candidates that demonstrate therapeutic potential in clinical trials subsequently fail.
According to Ignota Labs' analysis, safety-related issues account for approximately 56% of small molecule drug development failures.
However, the relationship between safety and efficacy is not always straightforward.
Image source: Ignota Labs
The Challenge of Balancing Therapeutic Benefit and Drug Toxicity
A drug may appear to fail because it cannot achieve the desired therapeutic effect. Yet, in some cases, the underlying problem is that the dose required to produce a meaningful clinical benefit also causes unacceptable toxicity.
This creates a narrow therapeutic window, limiting the ability to increase the dose sufficiently to achieve the intended effect.
Overcoming this challenge may require improving how effectively a drug interacts with its intended biological target or reducing its interactions with unintended targets.
How Off-Target Effects Contribute to Drug Failure
Drugs are generally designed to interact with specific proteins associated with disease. However, they may also interact with other proteins, potentially causing unexpected adverse effects. These unintended interactions, known as off-target effects, can contribute to toxicity and ultimately prevent a drug from progressing through development.
Although laboratory safety panels can investigate potential off-target interactions, testing every possible protein interaction would be prohibitively expensive. Consequently, some safety concerns only become apparent when drug candidates enter preclinical studies or early clinical trials.
Ignota Labs typically investigates candidates that have encountered safety issues during preclinical development or Phase 1 trials, when unexpected adverse events may first emerge in animal or human studies.
Liver toxicity was highlighted as one particularly important example.
How Ignota Labs Uses AI to Understand and Address Drug Toxicity
Central to Ignota Labs' approach is the application of AI to investigate the biological mechanisms responsible for drug toxicity.
Ignota Labs calls its proprietary AI-platform SAFEPATH. According to the company, SAFEPATH combines deep learning, cheminformatics, bioinformatics and multiple types of biological data to investigate the mechanisms of drug toxicity and identify potential strategies for overcoming them.
Rather than simply identifying that a drug causes an adverse effect, the aim is to understand why that effect occurs and whether the underlying problem can be resolved.
Using Cheminformatics to Predict Off-Target Drug Interactions
One component of the Ignota platform uses cheminformatics, which involves analysing chemical structure information to make predictions about how drugs may behave.
The platform uses machine learning models trained on existing experimental data describing interactions between drugs and proteins. When a new drug candidate is analysed, these models generate a ranked list of proteins with which the compound may interact. This provides an initial set of hypotheses about potential off-target effects.
The underlying models use neural networks rather than simple decision trees, allowing them to identify complex relationships within the data. Importantly, the technology can also help researchers investigate which parts of a molecule's chemical structure are contributing to particular predictions.
Combining AI Predictions with Biological Data
Predicting potential protein interactions is only the first stage.
Ignota Labs then integrates additional information about the adverse event observed during drug development. For example, if a compound has caused cholestasis, a form of liver toxicity associated with disrupted bile flow, this information can help narrow the range of possible biological mechanisms.
The company can also incorporate RNA sequencing data to examine changes in gene expression following exposure to a drug. By comparing predicted protein interactions with observed biological responses, researchers can determine which potential targets are most consistent with the available evidence.
This process allows the platform to refine an initial list of potential off-target interactions into a smaller number of higher-confidence hypotheses.
Image source: Ignota Labs
Why Human Expertise Remains Essential in AI Drug Discovery
Although AI can accelerate the identification of potential toxicity mechanisms, experimental validation remains a critical part of the process.
Once the platform has identified the most plausible explanations for a drug's adverse effects, scientists conduct laboratory experiments to investigate and validate these predictions.
As Layla describes it, the platform essentially acts as a hypothesis-generation tool, helping researchers prioritise the most promising explanations rather than investigating every possible mechanism experimentally.
Could AI Drug Rescue Reduce the Cost, Time and Environmental Impact of Medicine Development?
Rescuing failed drugs could have significant implications for the cost, time, and sustainability of pharmaceutical research. Drug development requires substantial financial investment, extensive experimentation and considerable use of resources. When a programme fails, much of that investment may no longer contribute to the development of a treatment.
Addressing Unmet Medical Needs: The Potential for Rare Disease Drug Development
Although Ignota Labs is not focused exclusively on one therapeutic area, the company places particular importance on identifying conditions with significant unmet medical needs.
Commercial considerations also influence programme selection, including whether there is a viable opportunity to partner or sell a drug candidate following further development.
However, Layla highlighted rare and orphan diseases as areas where the approach could have a particularly meaningful impact. For some rare diseases, there are no approved treatments. In other cases, available therapies may have limited effectiveness or significant adverse effects that patients must tolerate because few alternatives exist.
The ability to revisit promising drug candidates and address their safety limitations could therefore provide additional therapeutic opportunities for patients with limited treatment options. This is particularly important where developing entirely new medicines may be challenging because of the small patient populations involved.
What is Ignota Labs Business Model?
Ignota Labs' strategy involves acquiring or accessing promising drug candidates that have been discontinued, identifying the root cause of failure and developing new and improved versions.
Some of these assets originate from biotechnology companies that have ceased operating following unsuccessful development programmes.
Once a candidate has been re-engineered, the company aims to advance it to an appropriate clinical development milestone before selling or partnering the asset.
The intention is not necessarily to take every medicine through to commercialisation independently.
Instead, Ignota Labs seeks to create value by addressing the original development challenge and progressing the drug to its first clinical inflection point, before selling or partnering with another organisation, to take it forward. Depending on the programme, this may involve reaching early Phase 1 clinical development.
This model also creates opportunities for value to return to previous stakeholders through arrangements linked to milestones or subsequent asset sales.
Ignota Labs describes its operating model in three stages:
Identify, involving the assessment of failed or shelved assets;
Solve, involving the development of new and improved chemistry and intellectual property
Develop & Sell, involving progression to a key development milestone before sale or partnership.
The Future of Ignota Labs: Scaling AI-Driven Drug Development
Looking ahead over the next five to ten years, Ignota Labs aims to expand its drug rescue pipeline and demonstrate the effectiveness of its model at greater scale.
A key ambition discussed during our interview with Layla was that the team is aiming to complete an end-to-end drug turnaround, from acquiring a failed asset and addressing its limitations to successfully selling or partnering the improved candidate. Achieving this would validate the company's approach and support its plans to expand the number of programmes progressing through development.
Ultimately, the ambition is to establish a repeatable process for identifying promising failed drugs, resolving their development challenges and creating new opportunities for patients to access effective medicines.
Changing How the Pharmaceutical Industry Thinks About Failure
Perhaps the most important message from the interview was not about AI or machine learning, but about how the pharmaceutical industry approaches failure.
Drug development is inherently challenging, and failed programmes are an unavoidable part of scientific progress.
Yet failure does not necessarily mean that the underlying science lacks value.
For Ignota Labs, abandoned drug candidates represent the most valuable untapped dataset in biotech and an opportunity to learn from previous research, investigate the causes of unsuccessful development and potentially create better medicines.
Rather than treating failed programmes as the end of the discovery process, the company believes the industry should recognise the value of the knowledge they generate.
As Layla concluded, the industry needs to “stop viewing failure purely as a negative outcome, accept how common it is, and learn to work with it rather than avoid it.”
By combining AI-driven insights with scientific expertise and existing drug development data, Ignota Labs is exploring an alternative route to bringing promising medicines to patients.
Together, the future of drug discovery may depend not only on discovering new medicines, but also on recognising the untapped potential of those that have already been developed.
Zeeks is a science marketing company and media partner of ELRIG.
If you want to learn more about Zeeks media partnership, please email us at info[at]zeeks-artforgeeks.com.
References
Sun, Duxin, et al. ‘Why 90% of Clinical Drug Development Fails and How to Improve It?’ Acta Pharmaceutica Sinica B, vol. 12, no. 7, July 2022, pp. 3049–62. DOI.org (Crossref). Available from: https://doi.org/10.1016/j.apsb.2022.02.002.
Comments