It was a Tuesday evening in the lab — one of those sessions where you tell yourself you’ll run just one more simulation before heading home, and then three hours vanish. I was sitting in front of my workstation, watching AutoDock Vina finish a molecular docking run on cafestol, a diterpene compound found in unfiltered coffee. The molecule had been positioned inside a protein pocket, and the software was calculating how tightly it could bind.
The result came back: -10.06 kcal/mol.
To put that in context: many pharmaceutical compounds in clinical development show binding affinities in the range of -7 to -9 kcal/mol. Here was a molecule that millions of people consume every morning — a compound that seeps into your cup when you brew with a French press or a Turkish cezve — and our models were predicting a binding interaction that rivaled designed drugs.
I want to be careful here. A strong predicted binding affinity does not mean cafestol is a drug. It does not mean coffee treats or prevents any disease. Computational predictions require experimental validation, and the journey from a docking score to a clinical outcome is long, uncertain, and full of caveats that fill entire textbooks. But what that number told me was something I had been suspecting for years: we have barely begun to understand what is happening inside a cup of coffee, and the computational tools to investigate it already exist — we just haven’t been using them.
My path to this book was not a straight line. I trained as a physicist. My doctoral work was in computational methods — the kind of research where you spend your days building mathematical models of physical systems and your nights debugging code that refuses to converge. After my PhD, I moved into computational chemistry, drawn by the extraordinary power of molecular simulation. The idea that you could place a molecule inside a virtual protein, calculate the forces between every atom, and predict whether binding would occur — it felt like scientific sorcery. Except it was real, it was rigorous, and it was transforming pharmaceutical research.
For years, I worked at the intersection of computation and molecular science. I watched as molecular docking, network pharmacology, quantum mechanical calculations, and ADMET profiling became standard tools in drug discovery pipelines around the world. These methods were helping researchers identify drug candidates faster, understand mechanisms of action at atomic resolution, and predict how molecules would behave in the human body before ever running a clinical trial.
And then one morning, over a cup of coffee — naturally — I asked myself a question that seems obvious in retrospect: Why aren’t we doing this with food?
More specifically: why aren’t we doing this with coffee?
Here is a beverage consumed at a staggering scale — approximately 2.25 billion cups every day worldwide. It contains over 1,000 identified chemical compounds, many of them biologically active. Decades of epidemiological research suggest associations between coffee consumption and various health outcomes, though the mechanisms remain poorly understood. Coffee is, by any measure, one of the most chemically complex substances in the human diet.
And yet, when I searched the literature for molecular docking studies on coffee diterpenes, I found almost nothing. When I looked for network pharmacology analyses mapping coffee compounds to biological pathways, the cupboard was nearly bare. Quantum chemistry of the Maillard reaction during roasting? A handful of papers, mostly focused on small model systems that bore little resemblance to the actual chemistry occurring inside a coffee roaster.
The gap was staggering. The tools existed. The questions were waiting. Almost nobody was connecting the two.
That realization led me to found the Coffee Science Lab, now part of AIXC Research — a computational research initiative dedicated to applying the same rigorous methods used in drug discovery to the study of coffee chemistry. Not because coffee is medicine, but because coffee is chemistry, and chemistry deserves to be understood.
Let me be honest about what already exists and what does not.
There are excellent books about coffee science. Britta Folmer’s The Craft and Science of Coffee is a comprehensive overview of the coffee chain from agronomy to the cup. Victor Preedy’s reference volumes compile vast amounts of analytical and nutritional data. James Hoffmann’s The World Atlas of Coffee is a beautifully accessible guide to origins, processing, and brewing. These are valuable works, and I have learned from all of them.
But none of them takes the reader inside a molecular simulation.
None of them explains what happens when you position a cafestol molecule at the entrance of a nuclear receptor’s ligand-binding domain and ask a physics-based algorithm to predict whether it fits. None of them maps the network of protein targets that chlorogenic acid appears to interact with, or calculates the quantum mechanical energy barriers of the Maillard reaction, or runs an ADMET profile to predict how a coffee compound might be absorbed, distributed, metabolized, and excreted by the human body.
This is not a criticism of those books. Computational chemistry applied to coffee is a genuinely new field. When we began our research at the Coffee Science Lab, we were working in territory with very few footprints. The methods we use — AutoDock Vina for molecular docking, STRING and KEGG databases for network pharmacology, density functional theory for quantum calculations, SwissADME and pkCSM for pharmacokinetic prediction — these are well-established in pharmaceutical research. Applying them systematically to coffee compounds is what’s new.
This book is my attempt to share what we have found so far, and more importantly, to share how we found it.
The Science Inside Your Cup is built around four research studies conducted at the Coffee Science Lab. Each study applies a different computational method to a different aspect of coffee chemistry:
Diterpene Binding — We used molecular docking to predict how cafestol and kahweol, the oily compounds in unfiltered coffee, interact with nuclear receptors involved in lipid metabolism. Our models predict binding affinities that suggest these interactions deserve serious experimental attention.
Network Pharmacology — We mapped the known and predicted protein targets of major coffee compounds — caffeine, chlorogenic acid, trigonelline, and others — to build a systems-level picture of how coffee may interact with human biology. The resulting network is far more complex than the single-target thinking that dominates popular discussion of coffee and health.
Maillard Reaction Modeling — We applied quantum chemical methods to study key steps in the Maillard reaction, the cascade of chemical transformations that occurs during roasting and produces much of coffee’s flavor, aroma, and color. Research suggests this reaction is far more energetically nuanced than simple descriptions imply.
ADMET Profiling — We ran pharmacokinetic predictions on major coffee bioactives to estimate how they might be absorbed, where they might go in the body, how they might be metabolized, and how they might be eliminated. These profiles help explain why different compounds may have very different biological fates, even when consumed in the same cup.
Each chapter translates these studies for a general audience. You do not need a background in chemistry or computation to follow along. I have done my best to explain every concept from the ground up, using analogies, visualizations, and plain language wherever possible. But I have also included the actual numbers, the real parameters, and the genuine uncertainties. This is not a book that hides behind vague hand-waving. When our models predict something, I will tell you what they predict, how confident we are, and what we do not yet know.
I wrote this book for several kinds of readers, and I hope you will find yourself among them.
If you are a science-curious coffee drinker — someone who reads the back of the bag, wonders what “natural process” really means at a chemical level, or has ever Googled whether coffee is “good” or “bad” for you — this book will give you a deeper, more nuanced understanding of what is actually in your cup and how modern science investigates it.
If you are a barista or coffee professional — someone who already knows that paper filters remove oils and that roast level affects acidity — this book will show you why at the molecular level. When research suggests that paper filtration removes greater than 95% of diterpenes, there is a specific physical and chemical explanation for that, and understanding it changes how you think about brewing.
If you are a computational chemist or bioinformatician — someone familiar with docking protocols and network databases but looking for a real-world case study outside traditional drug discovery — coffee offers a remarkably rich and underexplored system. Over 1,000 compounds, dozens of biological targets, and an enormous population of daily consumers make it a compelling model for food-focused computational research.
If you are a food scientist — someone who works with analytical chemistry, sensory science, or nutrition but hasn’t yet seen what computational tools can bring to the table — this book is an invitation to explore methods that could complement and extend your existing approaches.
I need to be equally clear about what this book does not do.
This book is not medical advice. I am a computational researcher, not a physician. Nothing in these pages should be interpreted as a recommendation to drink more coffee, less coffee, or any specific type of coffee for health purposes. If you have questions about coffee and your health, please talk to your doctor.
This book is not a brewing guide. I will not tell you the ideal water temperature or the perfect grind size. There are wonderful books and resources for that. What I will tell you is what our computational models predict about the molecules that end up in your cup depending on how you brew — and I hope that adds a new dimension to your appreciation of the craft.
This book does not claim that coffee cures, treats, or prevents any disease. Computational predictions are hypotheses. They are starting points for experimental investigation, not endpoints. When our docking simulations predict that a coffee compound binds tightly to a particular receptor, that is an interesting finding that merits further study — in vitro, in vivo, and eventually in clinical research. It is not proof of a health effect. I will remind you of this throughout the book, because I believe that scientific honesty is not a limitation — it is what makes the science trustworthy.
This book is not the final word. Computational coffee chemistry is a young field. Our models will be refined. Some of our predictions will be confirmed by experiment; others will be revised or overturned. That is how science works, and I find it exciting rather than discouraging. What we present here is the best our current tools and methods can produce, offered with full transparency about their strengths and limitations.
Every morning, roughly a third of the world’s population performs an act of extraordinary chemistry without giving it a second thought. They pour hot water over ground, roasted seeds and extract a solution containing hundreds of biologically active compounds — alkaloids, polyphenols, diterpenes, melanoidins, volatile aromatics — in concentrations that would be considered pharmacologically relevant if they appeared in any other context.
And then they drink it. Usually while checking email.
I wrote this book because I believe that hidden complexity deserves to be revealed. Not to make coffee intimidating or to strip it of its simple pleasures — I assure you, understanding the Maillard reaction has not diminished my enjoyment of a well-roasted Ethiopian Yirgacheffe — but because the science itself is beautiful, and because the computational tools that make it visible are among the most powerful investigative methods our species has ever developed.
So here is my invitation: grab your favorite cup. It doesn’t matter if it’s a pour-over or an espresso, a light roast or a dark one, caffeinated or decaf. Hold it for a moment and consider that you are holding a solution of over 1,000 chemical compounds, each with its own shape, its own charge distribution, its own potential to interact with the proteins in your body in ways we are only beginning to map.
Now let’s explore what’s really inside it — molecule by molecule.
Coffee Science Lab Coffee Science Lab, AIXC Research March 2026