First QSAR equation
This time: a historical MedChem milestone – the birth of QSAR in 1961, long before AI, a breakthrough that laid the foundation for today’s MedChem, CADD, and AI revolution
There are moments in medicinal chemistry where the field doesn’t just progress, it changes its language. In July 1961, Corwin Hansch, the father of the computer-assisted design, (with Toshio Fujita) formulated what is widely considered the first QSAR equation, a quantitative way to connect molecular properties to biological activity.
It emerged from more than a decade of frustration trying to rationalize structure-activity relationships (SAR) for plant growth regulators using classical physical-organic tools alone. Soon after 1961 the idea accelerated: activity could be modeled, compared, optimized, not only “explained after the fact”. Before QSAR, MedChem was already brilliant but often qualitative:
• “This substituent seems better”
• “Adding lipophilicity helps (sometimes)”
• “Electronics might matter”
Hansch’s shift was bold (at that time, no AI tools at all): Treat biological activity like a property you can regress against physicochemical descriptors.
Hydrophobicity, electronics, and steric – as measurable variables, not just intuition. This is the conceptual ancestor of Hansch analysis, and in a very real sense, a foundational piece of what we now call computer-aided drug design (CADD).
Who was Corwin Hansch? He was trained in organic chemistry, worked during WWII, and later spent his career at Pomona College, where he continued to push quantitative thinking into chemical biology. Before becoming the “QSAR guy,” he worked as a group leader on the Manhattan Project in WWII-era research. Hansch is one of those rare people whose name became a verb in the community: “Do a Hansch analysis” = classic QSAR regression thinking. That’s peak scientific immortality. The impact we’re still living with (and how it connects to AI drug discovery).
QSAR didn’t replace medicinal chemistry intuition; it made it measurable, testable, and improvable. It shaped the way we still work today:
• rational lead optimization
• property-driven design
• toxicity prediction and risk modeling
• chemoinformatics and descriptor-based thinking
Here’s the real bridge to modern AI-driven drug discovery: AI models didn’t appear out of thin air – they inherited QSAR’s core idea: biological activity can be learned from molecular features. We’ve moved from linear regressions and handcrafted descriptors to deep learning and foundation models but the mindset is the same: turn chemistry into signals, and signals into predictions.
Even classics like the hydrophobic substituent constant (π) helped cement the idea that properties drive potency – a principle still embedded in today’s predictive pipelines.
From “make-and-test” → “design-and-test” → “predict-and-design”
That’s the legacy.
Other Molport Chronicles posts- from Alchemy to Pharma:
- Everyone, calm down! Benzodiazepine history
- Eli Lilly: From Civil War Roots to Global Pharma Leader
- Toxicology and Preclinical Testing (Animal Models)
- Dose Makes The Poison
- Oldest Pharmacy
- From Apothecary to Alkaloid Pioneer: The Origins of Merck and Modern Drug Discovery
- GSK legacy shaped by Beecham’s Pills and Research Laboratories
- The journey of ASPIRIN
- The footprint of Bristol-Myers Squibb (BMS)
- BASF: From dyes to drugs
