Skip to content
Learn Motion
ExploreHow it worksMembership
Log in
Learn Motion

Can AI Discover New Drugs? A High-Level Overview

1Why Drug Discovery Is Hard, and Where AI Fits2How AI Learns From Molecules and Proteins3Finding and Validating a Biological Target4Designing Molecules: Generative AI and Virtual Screening5From Hit to Lead: Optimizing Properties With AI6What AI Still Cannot Do7Judging the Claims: Real Successes, Failures, and Open Questions
From Hit to Lead: Optimizing Properties With AI

One Molecule, Many Requirements

1 / 4
Think of the radar chart as a report card with several independent subjects. Each spoke is one requirement: how strongly it binds, how well it is absorbed, how long it survives the liver, how selectively it hits only the intended protein, how safe it is, and how easily it can be made. A compound is not judged by its best subject. It is judged by whether every spoke reaches the minimum acceptable ring. A brilliant binder that is destroyed by the liver in minutes still fails, and that is why chemists call this multi-parameter optimization: the target is a shape that clears every axis, not a spike on one.
0:00 / 0:00

A compound that binds its target tightly can still fail as a drug. Binding is only the first of several requirements, and they are largely independent of one another.

A candidate must reach its target in the body, which means surviving the journey from the gut into the bloodstream (absorption), avoiding rapid breakdown by liver enzymes (metabolism), and being cleared at a rate that keeps useful concentrations for hours rather than minutes (excretion). It must act on the intended protein without strongly engaging unrelated proteins (selectivity), because off-target activity is a common source of side effects. It must not be toxic on its own or through the breakdown products it forms. And it must be practical to manufacture at scale.

These requirements are not aligned. A molecule can be an excellent binder and still be destroyed by the liver within minutes, or bind its target and a heart-ion-channel protein with similar strength. Chemists summarize the situation as multi-parameter optimization: the goal is not the best value on any single axis but a set of values that is acceptable on all of them at once. A radar plot makes this visible — each axis is one property, and a compound is only viable if its shape stays outside the minimum acceptable ring on every axis.

Previous1 / 4Next

Learn Motion

Generate a course. Learn it properly.

Operated by Wuhan Daoyin Technology Co., Ltd.

Contact: [email protected]
Privacy PolicyTerms of Service

© 2026 Learn Motion