AI for selectivity and polypharmacology of small molecules

Off-target effects can be an essential component of a small molecule's clinical success or failure. With AI, these relationships can be identified and leveraged

In the world of drug discovery and development, the quest for compounds with high selectivity is a paramount challenge. Selectivity refers to a compound's ability to target a specific biological entity, typically a protein or receptor, while minimizing interactions with unintended off-targets.

Before we dive into the intricacies of selectivity, it is crucial to understand why it is such a critical factor in drug development.

Selective compounds are believed to offer several advantages:

Minimized Side Effects: When a compound interacts only with the intended target, the likelihood of adverse effects on other biological processes is significantly reduced, leading to a safer drug profile.

Reduced Cost and Time: Developing a selective compound can save time and resources by avoiding costly clinical trials and safety issues associated with non-selective compounds.

Off-target effects : friends or foe?

Off-target effects occur when a compound interacts with unintended biological targets, leading to unpredictable and - potentially - harmful consequences. These effects can manifest as adverse reactions, toxicity, or ineffectiveness of the drug. Several factors contribute to off-target effects, such as structural similarities between targets, and the inherent complexity of biological systems about which we have a very partial understanding.

Addressing off-target effects is a formidable challenge in drug development. Traditional methods involve extensive experimental screening, which is both time-consuming and costly. Moreover, they may not uncover all off-target interactions nor provide a comprehensive understanding of the compound's selectivity.

In some cases, off-target effects can actually be desirable.

The phenomenon of polypharmacology, where a drug interacts with multiple targets, is a prime example of this. Polypharmacology is the concept that a single compound can have therapeutic effects by interacting with multiple biological targets simultaneously. In some instances, the serendipitous discovery of off-target effects has led to groundbreaking pharmaceuticals. For example, Sorafenib is a multikinase inhibitor that targets both Raf and VEGF and PDGF receptor tyrosine kinase signaling.

The role of AI in compound selectivity and polypharmacology optimization

Artificial Intelligence, particularly machine learning and deep learning algorithms, has emerged as a new promising tool in addressing compound selectivity and off-target effects, as well as in uncovering polypharmacological profiles in drugs.

Some AI models can identify, with various levels of performance, the off-target hits of known or new drugs. This is a very important value driver of drug discovery, as it unlocks our understanding of how a drug will interact with different proteins in the body. Knowing the possible targets of a drug helps to identify potential new uses, identify molecules that act on multiple targets or disease pathways, and also identify the possible toxic effects of a drug.

In reverse, AI models can be used to optimize the selectivity of a compound. For some projects, we will allow our AI technology to select more or less selective compounds, depending on the target.

How to choose the right AI approach

Multiple AI approaches exist and, unsurprisingly, each one comes with its own benefits and challenges. Some of the observed challenges of AI based methodologies in determining the off target effects of drugs are:

The difficulty to make predictions actionable, as the AI generates many hypotheses which are difficult to rank and understand. This can be for example the case in network based approaches of -omics data analysis.

More importantly, the complexity to establish accurate predictions when it comes to new compounds or targets for which no data is available.

At Kantify, we have developed an AI technology to predict the off-target hits of new molecular entities, or of known drugs whose targets are not always initially known. This technology provides predictions that help to make the drug discovery work actionable.

For example, we recently discovered how a drug currently being validated for an indication in oncology is a potent inhibitor of an ion channel.

At Kantify we use AI in order to:

Discover new promising drugs where we can optimize the possible off-target effects and identify possible polypharmacological compounds.

Repurpose or reposition existing drugs, for which we can identify the potential of existing drugs for combinations or new indications.

More selectivity or less selectivity?

Off-target effects, whether we know them or not, are a core part of how a drug works, even if we first think of them as something to hold with caution due to their potential for adverse reactions. Thanks to AI, it is now possible to treat off-target effects, not only as a constraint, but as a true opportunity to optimize new small molecules or even expand therapeutic horizons of existing small molecules.

Do you wish to learn more about how to use Kantify’s technology? Get in touch via the below contact form.

Get in touch !