Research — Jorgensen Research Group, Yale

Why it matters

Two campaigns.

The group's two largest campaigns took on the defining virus of two different decades — and each ended with molecules more potent than medicines already on the market.

HIV-1 · 2011

HIV-1 reverse transcriptase

SARS-CoV-2 · 2021

SARS-CoV-2 main protease

How it works

How a campaign runs

Synthesis is the slow, expensive step. Free-energy perturbation simulates each candidate inside the target's binding site and ranks them by how tightly they should bind, before they are synthesized.

1
Design

Model the inhibitor inside the target's binding site — docking a hit, or growing new groups in place with BOMB — and propose the next analogs.
2
Predict · FEP

Free-energy perturbation computes each analog's change in binding free energy (ΔΔG) by Monte Carlo simulation in BOSS / MCPRO — ranking them before one is made.
3
Make

Synthesize the highest-ranked compound in our own laboratory.
4
Test

Measure potency and solve the X-ray crystal structure with collaborators — confirming the prediction and guiding the next design round.

Each structure sharpens the next round — repeated until a micromolar hit becomes a low nanomolar lead.

The toolkit

The software behind

Every step above runs on code written in this group: BOSS and MCPRO for the Monte Carlo and free-energy perturbation, the OPLS force fields for the energetics, BOMB for growing analogs inside a binding site, and LigParGen for the parameters. see more here →

The signature result

The HIV‑1 campaign, compound by compound

Here is that HIV‑1 campaign as it actually ran — seven synthesized compounds, each one the molecule the free‑energy calculation ranked highest at that point. Every rung is a real, assayed compound, and the picomolar lead's binding was later confirmed by X‑ray crystallography.

Hover, tap or tab to any point to see the edit that produced it.

Each edit was chosen by predicting its effect on binding free energy before a single gram was synthesized. Anti‑HIV‑1 EC₅₀ values against wild‑type virus in infected MT‑2 cells, from Tables 2 and 4 of Bollini, Domaoal, Thakur, Gallardo‑Macias, Spasov, Anderson & Jorgensen, J. Med. Chem. 2011, 54, 8582 (doi:10.1021/jm201134m); the campaign made many more compounds than the seven milestones plotted here. The series was later extended to inhibitors that stay picomolar against drug‑resistant HIV (Lee et al., J. Am. Chem. Soc. 2013, 135, 16705) and advanced to a long‑acting preclinical candidate (Chan et al., PNAS 2018, 10.1073/pnas.1717932115).

The COVID-19 campaign

In response to COVID, from a weak hit to a preclinical candidate

The HIV campaign took years. When SARS‑CoV‑2 arrived, the same loop ran on a different clock: an approved epilepsy drug that barely bound the viral protease was reshaped into a nanomolar inhibitor in a matter of weeks, with crystallography confirming the predictions almost as fast as they were made.

2020 · spring

A virtual screen of known drugs against M^pro gives 42 consensus hits; 17 are bought and assayed, and 14 are active — the pool the optimization starts from.

ACS Med. Chem. Lett., published Oct 2020 ↗
2020 · June

Perampanel — an approved epilepsy drug and a very weak hit at 100–250 µM — is picked for optimization. The first FEP-guided designs come back at 10, 6.4 and 4.0 µM.
2020 · 2 September

The ninth compound made reaches an IC₅₀ of 140 nM. That is roughly a thousand-fold gain in a few weeks of design.
2020 · 4 October

A 1.6 Å crystal structure confirms the predicted binding mode exactly.

PDB 7L10 ↗
2021 · February

Published: 27 compounds, five crystal structures, IC₅₀ values around 20 nM — from weak hit to potent lead in a few months.

ACS Cent. Sci. ↗
2021 · July

Triarylpyridinones reach EC₅₀ ≈ 80 nM against infectious virus.

ACS Med. Chem. Lett. ↗
2024 · April

Mpro61 is profiled as a preclinical candidate: no CYP3A4 inhibition, so no ritonavir booster is needed, and it acts synergistically with molnupiravir.

PNAS 2024, 121, e2320713121 ↗

Beyond the antivirals

Other targets

Immunology · Inflammation

Macrophage Migration Inhibitory Factor

screening hit low nM

MIF is a cytokine implicated in inflammation, autoimmune disease, and cancer. Virtual screening and FEP-guided optimization delivered low-nanomolar inhibitors of its tautomerase activity, against human MIF and the malaria-parasite ortholog.

Also in progress

Further targets

hDM2 and hDMX
Tdp1
FGFR1 kinase
JAK2 kinase
DNMT

Campaigns at earlier stages, run through the same design–make–test loop — spanning oncology, inflammation and epigenetics.

A catechol diether inhibitor, shown as gold sticks, bound in the non-nucleoside inhibitor pocket of HIV-1 reverse transcriptase, whose molecular surface encloses it. — Designed inhibitors in their targets, rendered from the group's own deposited crystal structures — the experimental confirmation that closes each design cycle.

A biaryltriazole inhibitor, shown as gold sticks, bound in the tautomerase active site of macrophage migration inhibitory factor, whose molecular surface encloses it. — Designed inhibitors in their targets, rendered from the group's own deposited crystal structures — the experimental confirmation that closes each design cycle.

Right now

Current work

Current work pushes the SARS‑CoV‑2 main-protease candidate Mpro61 against emerging resistance and into combination regimens; examines a dual-function Gag‑Pol dimerizer that recruits the CARD8 inflammasome to clear HIV-infected cells; and continues the methodological work every campaign rests on — new OPLS parameters, and Monte Carlo free energies of hydration and solvation.

All 483 publications → The people behind the work → Get in touch →

Most recent papers

2026
Free Energies of Solvation in Cyclohexane and Water: United-Atom vs All-Atom
J. Chem. Theory Comput. · 22, 7581-7593
2025
Polycyclic Aromatic Hydrocarbons: Solvation, Solubility, and Hydrophobic Effects from Monte Carlo Simulations
J. Phys. Chem. B · 129, 10467-10477
2025
Mechanistic basis for a novel dual-function Gag-Pol dimerizer potentiating CARD8 inflammasome activation and clearance of HIV-infected cells
npj Drug Discov. · 2, 22
2025
The need to implement FAIR principles in biomolecular simulations
Nat. Methods · 22, 641-645
2025
Free Energies of Solvation in Benzene and Hexafluorobenzene: Is Explicit Polarization Needed?
J. Phys. Chem. B · 129, 6574-6583
2025
Exploring Possible Drug-Resistant Variants of SARS-CoV-2 Main Protease (Mpro) with Noncovalent Preclinical Candidate, Mpro61
ACS Bio & Med Chem Au · 5, 215-226

The programme

Four goals, fifty years.

I

Characterize intermolecular interactions in solution

How molecules actually behave in a solvent, rather than in a vacuum. This is where the OPLS force fields and the TIP3P/TIP4P water models came from — the potentials that made everything after it quantitative.

OPLS · TIPnP · BOSS →
II

Model solvent effects on reaction rates

Reaction mechanisms, medium effects, and the influence of site-specific mutations on enzymatic reactions — treated with QM/MM: the reacting atoms quantum-mechanically, the surrounding solvent by molecular mechanics, all sampled by Monte Carlo.

QM/MM · PDDG/PM3
III

Prove free-energy calculations can optimize a drug lead

First retrospectively, against series whose activities were already known; then prospectively, choosing which molecule to make next. That test is what the HIV and SARS‑CoV‑2 campaigns above are.

MCPRO · BOMB · two preclinical candidates
IV

Perfect pure-liquid and solution properties

The accuracy ceiling on every prediction above is the force field, so the force field keeps being rebuilt. Recent work reaches 0.2 kcal/mol on heats of vaporization and 0.01 g·cm⁻³ on densities — and resolves the long-standing problem of excess hydrocarbon hydrophobicity.

OPLS/2020 · OPLS-AA/M →