In a world where malaria still claims over 600,000 lives annually, the race to develop effective antimalarial treatments has never been more critical. For Medicines for Malaria Venture (MMV), a Swiss-based non-profit at the forefront of antimalarial drug research, the challenge is clear: find a way to expedite the discovery of potent and safe drugs to combat this persistent global threat.

The Challenge: A Deadly Stalemate

Despite decades of progress in malaria prevention and treatment, recent years have seen a plateau in advancements. Yet the situation remains dire:

• Malaria poses a risk to nearly half of the world's population.
• An estimated 608,000 people died from malaria in 2022 alone.
• Young children in sub-Saharan Africa are the most vulnerable.
• Emerging resistance to artemisinin, the core compound of the best available drugs, has hampered efforts to contain the disease.

MMV's Malaria Libre program, an open-source drug discovery initiative, identified a promising antimalarial compound: aryl piperazine. However, while demonstrating good inhibitory activity against red blood cells infected with Plasmodium falciparum (the most deadly malaria strain, also called 3D7 strain), aryl piperazine exhibited a high likelihood of inducing drug-drug interactions (DDI) as the lead compound had high affinity against CYP450 with an inhibitory activity of up to 84% (against the 3A4 isoform) at a concentration of 10 µM (see SAR analysis here). DDIs are a common issue for drug discovery programs and many assets have been abandoned due to high likelihood of DDIs.

The Solution: Identifying novel and better compounds faster

When we met the MMV team, DeepMirror proposed that we could generate a molecule that is less active against CYP450, yet has a similar potency against malaria as the lead molecule in the aryl piperazine series. The DeepMirror platform learnt the relationship between the 175 compound structures from MMV and their 3D7 lethality to suggest novel compounds that could still be potent, while using the global model of CYP450 activity on DeepMirror to prioritise the ones that could reduce CYP450 activity – all within an hour.

Here's how:

1. Data Ingestion: We downloaded the aryl piperazine dataset from the Malaria Libre data repository and uploaded the molecules to DeepMirror.
2. AI-Powered Generation: We generated >1,000 novel compounds with our generative AI capabilities and the goal of similar antimalarial activity and lower CYP450 activity (see video below).  
3. Rapid Iteration: In just one hour, DeepMirror identified a novel compound (DM1133) with predicted CYP450 activity above 10 µM, a potential improvement of ~10x over the existing top candidate (see Figure).
   

Scrolling through a subset of the generated molecules. Note how the chemical space of the generated molecules covers a large and diverse area.

Results: Lab-testing the generated molecules

We showed MMV the molecule and tested it for antimalarial and CYP450 activity against 3A4 and other isoforms with MMV testing entities.

Remarkably, the lab tests confirmed the predictions as DM1133 did decrease CYP450 activity against all tested isoforms by ~10x (Figure below). Further medicinal chemistry knowledge driven changes such as the replacement of the aminal group with a pyrrolidine may further improve the molecule:

The Future: Accelerating Hope

The collaboration between MMV and DeepMirror underscores the potential of AI to accelerate drug discovery by aiding in the development of testable molecules with optimal properties. We hope that DeepMirror's software will empower chemists worldwide to discover the next breakthrough molecules more efficiently, helping to bring life-saving treatments for malaria and other diseases to the clinic faster than ever before.

‍

About DeepMirror

DeepMirror is the AI drug design platform that empowers chemists to focus on only the most promising drug molecules to accelerate the time to the clinic. Simple enough for anyone to use, powerful enough to transform drug discovery, and already deployed in biotechs worldwide.