Season 06 · Episode 03 · Medicine · 50 min
AI in Medicine: The Protein Fold Dividend.
When AlphaFold solved protein structure, headlines promised a cure for everything. Dr. Samuel Okafor explains what the breakthrough actually bought, and why the bottleneck simply moved.
September 25, 2024·With Dr. Samuel Okafor
▶50:22§01Chapters
Introduction: Fifty Years in One Weekend
What a Fold Actually Tells You
The Bottleneck Moves Downstream
Designing Molecules That Never Existed
The Diseases Capital Forgot
A Decade of Deliverables
§02Show Notes
For fifty years, determining the shape of a single protein could consume a doctoral career. Then, effectively overnight, a model predicted nearly all of them. Dr. Samuel Okafor leads computational biology at Helix Therapeutics, and he opens this conversation with a correction to the mythology: the fold was never the cure, it was the map, and a map is only as valuable as the expeditions it enables.
The middle chapters are a masterclass in where drug discovery actually stalls. Knowing a target's structure compresses years into weeks at the front of the pipeline, but the pipeline still empties into the same clinical trials, the same biology that refuses to be simulated, the same nine-in-ten failure rate. Samuel's data on how his team's hit rate changed, and his honesty about what didn't change at all, is the antidote to a hundred breathless keynote slides.
The most affecting section concerns neglected disease. Structure prediction is nearly free, which means, for the first time, the economics of drug discovery for diseases of the poor have genuinely shifted. Samuel's team releases open structures and open leads for two parasitic diseases as a matter of policy, and his explanation of why a for-profit company does this, and why more should, closes the episode on unexpected moral high ground.
“The fold was never the cure. It was the map, and a map is only as valuable as the expeditions it enables.”
§03Transcript Extract
NISHANT:
You were finishing your PhD when AlphaFold's results came out. What did that weekend feel like inside the field?
DR. OKAFOR:
Like being a cartographer the week the satellite photos arrived. Some colleagues grieved, honestly, because a craft they had given decades to was suddenly a download. I felt the opposite. I never wanted to draw the map. I wanted to go where the map pointed, and overnight there were two hundred million new places to go.
NISHANT:
The headlines said this would cure cancer, Alzheimer's, everything. Four years on, score the promise.
DR. OKAFOR:
The front of the pipeline kept the promise. Target identification that took my predecessors three years takes my team three weeks. But a drug is not a shape, it is a decade of chemistry, toxicity, manufacturing and trials, and none of that got faster because a model knows the fold. The bottleneck didn't disappear. It moved downstream, where it is more expensive and less photogenic.
NISHANT:
So where is the dividend actually being spent?
DR. OKAFOR:
Two places. Molecules we would never have dared attempt, targets everyone called undruggable because nobody could see them properly. And diseases capital abandoned. A structure used to cost a million dollars of someone's grant. Now it costs nothing, which means leishmaniasis and Chagas get the same quality of map as the diseases of rich countries. We publish those leads openly. The fold was humanity's result. Some of the dividend should be too.
§ Keep listening
More on Medicine.
S07 · E02 · 47 min
The Diagnosis Revolution
How machine learning is detecting disease earlier than any human can — and what hospitals must do to keep up.
S07 · E06 · 46 min
Powering the Algorithmic Grid
AI is balancing the world's most complex machine — the electricity grid — in real time. What happens when it fails?
S07 · E05 · 55 min
The Death of the Script?
Generative tools are rewriting Hollywood's economics. We talk to the people on both sides of the picket line.