Pauling.AI Unveils Autonomous Membrane Construction: Asymmetric Membranes Built From Scratch and Validated Against Experiment
Pauling.AI introduces a protocol that assembles complete protein-membrane systems with per-leaflet lipid control,
Press Release Disclaimer: This is a press release distributed through the XPR Media network. It has not been independently verified by our newsroom.

Boston, MA / Storyteller / Nov 18, 2026 /
Pauling.AI today announced the launch of its Membrane Construction protocol: a fully in-house workflow that builds a complete, simulation-ready protein-membrane system from nothing but a protein structure, with no external service in the path and no specialist in the loop.

Membrane proteins are among the most important classes of drug targets in pharmacology: GPCRs alone are the targets of about 36% of approved drugs, and ion channels, transporters, and receptors of every kind carry out their biological functions at a lipid interface. Simulating any of them requires something no experiment provides: a complete atomistic assembly of the protein embedded in a lipid bilayer. Crystallography removes the membrane, and putting it back has traditionally required expert input, specialized tools, and careful quality control. A single misplaced lipid, a protein tilted a few degrees off its correct orientation, or a protonation state assigned without accounting for ligands can silently compromise months of simulation work.
Physics Places the Lipids, Not a Packing Algorithm
The Membrane Construction protocol is built on a simple principle: the membrane should assemble itself. The prepared protein is oriented in the membrane frame using the community-standard OPM database or a transfer-energy fit, then embedded in a coarse-grained bilayer built with the Martini 3 force field and the 2025 Martini lipidome. Through tens of nanoseconds of coarse-grained dynamics, the lipids find their own contacts with the protein, including annular shells, cholesterol-binding grooves, and local thinning. The equilibrated assembly is then reconstructed at full atomic resolution using a conformer library of more than 83,000 all-atom lipid conformers spanning 82 lipids, generated from Pauling.AI‘s own simulations.
Around that core, the protocol automates the tasks a specialist would otherwise perform by hand: pH-correct, ligand-aware protonation; disulfide bonding; covalent cofactors, including palmitoylated cysteines and the retinal Schiff base of rhodopsins; explicit ligand parameterization; correct ion counting; and a battery of ten structural quality gates. These gates assess each build before it is exported, checking for issues ranging from ring penetrations to leaflet composition and core vacuum. A typical GPCR-sized system builds end to end in about an hour, reproducibly, with a full QC report.
Asymmetry as a First-Class Citizen
Real membranes are asymmetric, and the protocol treats them that way. Each leaflet gets its own lipid composition, specified in mole percent, balanced by area, and preserved through assembly rather than scrambled by it. Biologically grounded presets for the plasma membrane, neuronal synapse, endoplasmic reticulum, Golgi, mitochondrial inner membrane, and lysosome are selected automatically based on the protein’s annotated localization. Every value remains editable per leaflet before the build runs. Downstream analyses can distinguish the two leaflets by construction, which is the central purpose of building them separately.
Validated Against Experiment, Not Against Another Simulation
The builder is evaluated against physical principles and published experimental findings. Built bilayers reproduce the experimental area per lipid of POPC to within about 1% of X-ray and neutron scattering values, reproduce cholesterol’s condensing effect on membrane thickness and order, and match NMR deuterium order parameters. The exported energetics match the AMBER reference engine term by term to within hundredths of a kilojoule per mole.
A standing benchmark suite also compares every new system class against published experimental results, including scattering measurements of pure bilayers, solid-state NMR studies of designed transmembrane peptides, and the crystallographic cholesterol-binding sites of the beta-2 adrenergic receptor. In the latter benchmark, the built membrane’s cholesterol must recover the sites after they have been deleted.
“Membrane targets are where the unexplored biology lives, and they have always been the hardest to simulate,” said Javier Tordable, CEO of Pauling.AI. “With Membrane Construction, we collapsed the last manual, specialist-dependent step between a membrane protein structure and a production-quality simulation. The system builds itself, checks itself against experiment, and hands our agents a membrane they can trust. Nobody else offers that autonomously, end to end, with this level of validation.”
Membrane Construction is available today on the Pauling.AI platform, alongside the platform’s membrane equilibration, ligand insertion, production dynamics, and analysis workflows. Full methodology and references are published at www.pauling.ai.
About Pauling.AI
Pauling.AI is a Boston-based computational drug discovery company developing what it describes as the first fully autonomous platform for molecular discovery. Its technology integrates AI agents, physics-based computational chemistry, and enterprise-scale cloud infrastructure to automate workflows including structure preparation and repair, large-scale virtual screening, molecular docking, molecular dynamics, binding-affinity analysis, ADMET prediction, and candidate prioritization.
Pauling.AI’s mission is to make sophisticated AI drug discovery faster, more accessible, and immediately actionable for medicinal chemists, biotechnology companies, and pharmaceutical research teams.
For more information, visit Pauling.AI.
Media Contacts
Pauling.AI
Media Relations
info@pauling.ai
206-434-9648
www.pauling.ai
Social Media:
LinkedIn: https://www.linkedin.com/company/pauling-ai/posts/

