A quote from around 18 minutes into the David Liu: A Master Class on the Future of Genome Editing podcast:

Virtually every substance we've ever put into a person, including just about every medicine we've ever put into a person, has off-target effects, meaning [the substance] modulates the function of biological molecules other than the intended target.

This seems strange, but the understanding how a drug works might come long after it is approved. As long as it works, and is safe, who cares?

There’s an argument that we should be deliberately exploiting more promiscuous molecules

Much of drug development is focused on making a drug do one very specific thing at one very specific place. It’s easier to tackle and safer today. And yet these molecules do have effects elsewhere (“off-target”).  

In Drug discovery for ageing: SIMPs, NEDs and screening challenges an alternative is to “engage multiple pathways and processes weakly”, but that’s hard to do. The authors suggest AI is needed.

Resources to help get there

There are existing data sets and approaches that look promising:

• Library of Integrated Network-Based Cellular Signatures (LINCS), which is a dataset of gene regulation before and after a treatment.
• Companies like Kantify are predicting multiple-target interactions.
• Astonishing work like BENEIN, which builds a network of interactions in how cells change over time, and where you might intervene.

There’s now a new open database showing the off-target effects of approved drugs.  I head about this, and it’s nicely described, in: Mapping the off-target effects of every FDA-approved drug in existence (EvE Bio).  

The point is made there that 20%+ of approved drugs end up being used for other conditions (“off-label”), which is a big hint that those off-target effects are (sometimes) handy. It “implies that we’re sitting on a vast, under-explored landscape of therapeutic potential, one that clinicians are already intuitively poking into”.

Why this is interesting

Repurposing existing drugs is a great idea, but what I really find interesting about this is:

• the possibility of uncovering networks (systems) in biology;
• being able to use that information to find drugs to nudge us into a healthy state via multiple small effects.

It seems like the pieces (data + computation) are being put in place to try that.