The Research

I ran four computational studies — each one asking a different question about coffee at the molecular level. Here's what I found, explained the way I'd explain it to a friend over coffee.

Research Studies

Diterpene-Receptor Binding

Cafestol and kahweol bind FXR at −10.1 kcal/mol — rivaling pharmaceutical agonists. Using AutoDock Vina molecular docking, we mapped how coffee’s cholesterol-raising diterpenes interact with nuclear receptors.

Read Paper ↓

Multi-Target Network Analysis

6 coffee compounds predicted to interact with 10 protein targets across 4 disease pathways. Network pharmacology reveals coffee isn’t a single drug — it’s a multi-target therapeutic system.

Read Paper ↓

Maillard Reaction Modeling

A 75–125× kinetic bottleneck in the Maillard reaction explains roasting’s development phase. DFT quantum chemistry reveals what happens at the electron level during roasting.

Read Paper ↓

ADMET Profiling

10 of 15 coffee bioactives are predicted to cross the blood-brain barrier. We applied pharmaceutical absorption/distribution/metabolism/excretion/toxicity screening to coffee compounds.

Read Paper ↓

Our Publications

All papers are open access. Download the PDFs or read the book-length treatment in The Science Inside Your Cup.

Paper 1 — Molecular Docking

Molecular Docking of Coffee Diterpenes Cafestol and Kahweol Against Cholesterol Metabolism Targets: A Computational Docking Study

Enrique Zueco — AIXC Research

2026 · Preprint · 4 targets, AutoDock Vina

Show Abstract

Over 2.25 billion cups of coffee are consumed daily worldwide, yet the molecular basis by which unfiltered coffee raises cholesterol remains poorly understood at the atomic level. Here we present the first systematic molecular docking study of cafestol (C<sub>20</sub>H<sub>28</sub>O<sub>3</sub>) and kahweol against four key cholesterol metabolism proteins: LXR-β, HMG-CoA reductase, CYP7A1, and FXR. Both diterpenes showed strongest binding to FXR (cafestol: −10.06 kcal/mol; kahweol: −10.11 kcal/mol), exceeding typical drug-like thresholds and rivaling obeticholic acid, a clinically approved FXR agonist. Our results reveal a target selectivity hierarchy (FXR ≫ LXR-β ≈ HMGCR > CYP7A1) that reconciles previously disconnected experimental observations.

cafestol kahweol molecular docking FXR cholesterol nuclear receptors
Paper 2 — Network Pharmacology

Integrative Network Pharmacology Reveals Multi-Target Cardiometabolic Mechanisms of Coffee Bioactives

Enrique Zueco — AIXC Research

2026 · Preprint · STRING v12.0, 16 nodes, 36 edges

Show Abstract

Coffee is one of the most widely consumed beverages worldwide, and epidemiological evidence consistently associates moderate consumption (3–5 cups/day) with reduced risk of type 2 diabetes, cardiovascular disease, and neurodegenerative disorders. Here, we apply an integrative network pharmacology approach to systematically map the target space of six major coffee bioactives—caffeine, cafestol, kahweol, 5-CQA, trigonelline, and caffeic acid—against 10 validated human protein targets. Network topology analysis identified NFE2L2 (Nrf2), PTGS2 (COX-2), and PPARγ as hub targets with the highest degree centrality. Pathway enrichment analysis revealed convergent modulation of lipid metabolism, neuroinflammation, oxidative stress defense, and xenobiotic metabolism pathways.

network pharmacology multi-target Nrf2 PPARγ COX-2 STRING
Paper 3 — Quantum Chemistry

From Maillard to Melanoidins: Computational Modeling of Coffee Roasting Chemistry

Enrique Zueco — AIXC Research

2026 · Preprint · DFT + Molecular Dynamics, Maillard Development Index

Show Abstract

The Maillard reaction is the dominant chemical transformation during coffee roasting, producing melanoidins that constitute up to 25% of brewed coffee dry weight. Here we present an integrated multi-scale computational framework connecting quantum-level reaction energetics to macroscopic roasting outcomes. DFT-computed activation energies translated via Eyring transition-state theory reveal a persistent kinetic bottleneck—the 1,2-enolization step is 75–125× slower than the Amadori rearrangement. We introduce the Maillard Development Index (MDI), the first roast-quality metric grounded in quantum chemical rate constants. Original molecular dynamics simulations characterize melanoidin self-aggregation energetics: π–π stacking contributes −35 kJ/mol per aromatic contact, hydrogen bonding −25 kJ/mol per pair, and Fe<sup>3+</sup> binding free energy of −48 kJ/mol.

melanoidins Maillard reaction DFT molecular dynamics roasting kinetic bottleneck
Paper 4 — ADMET

Computational Pharmacokinetic Profiling of Fifteen Coffee Bioactives: Absorption, Distribution, and Bioavailability Across Brewing Methods

Enrique Zueco — AIXC Research

2026 · Preprint · SwissADME + pkCSM, 15 compounds, 5 brewing methods

Show Abstract

No systematic in silico ADMET analysis exists for coffee’s complete bioactive panel across brewing methods. Here, we present the first comprehensive computational pharmacokinetic profiling of 15 major coffee bioactives using SwissADME and pkCSM predictive models. All 15 compounds satisfy Lipinski’s Rule of Five. Ten compounds are predicted to cross the blood–brain barrier, including caffeine, trigonelline, and the diterpenes. We construct an “effective bioavailability index” revealing that espresso delivers the highest neuroprotective SBI per unit volume, but filter coffee offers the most favorable risk–benefit profile by minimizing diterpene exposure while still providing moderate neuroprotective delivery.

ADMET pharmacokinetics blood-brain barrier brewing methods bioavailability SwissADME
Conference Abstract

World of Coffee Geneva — Conference Abstract Submission

Enrique Zueco — AIXC Research / Coffee Science Lab

2026 · World of Coffee, Geneva · Abstract

Want the Full Story?

We turned these four studies into a book — written for anyone who's curious, not just scientists. Each chapter starts with a cup of coffee and ends with a molecule.

Read About the Book