June 2026
Model Performance Series
Astra AI on Transporters
Membrane transporters move ions, nutrients, and drugs across the lipid bilayer — among the most under-exploited drug-target classes, and among the hardest to characterize: folds vary widely, many shift between conformations, and a sizable subset are themselves enzymes. This report shows how Orbion's Astra AI Suite characterizes them from sequence alone — function, topology, PTM sites, binding pockets, and thermostability — measured against public experimental ground truth across 6,203 transport-associated proteins.
0.95 AUROC Topology
56% Ligand Recall
ρ=0.59 ΔTm (SERT)
97% Recognition Across the Canonical Transporter Families
SLC carriers 97%, ABC transporters and aquaporins 100% recognized as transporters. The broad keyword also captures enzymatic ATPases and trafficking GTPases, which the model routes to their true function rather than averaging away — and residue-level topology stays strong across every family (AUROC 0.94–0.98).
F1 Up to 0.88 on PTM Site Prediction
Per-residue sites across 39 classes; strongest on disulfide bonds (0.88) and N-linked glycosylation (0.83) — the modifications that govern folding and surface trafficking of multi-pass carriers.
ρ=0.59 ΔTm Ranking on SERT
On 197 mutations of the serotonin transporter, the suite ranks by predicted ΔTm at ρ 0.59, MAE 2.13 °C, 81% directional on strong stabilizers. It compresses extreme-magnitude effects (shown honestly) — the value is correct ranking before a thermal-shift screen.
Have a question?
Whether you want to understand our models, test a specific protein, or just ask how Orbion applies to your research — send us your question and we'll get back to you. Or, if you'd prefer a live walkthrough, book a demo.
