Figure 2

Prediction of pore-lining helices from sequence, mapped onto native crystal structures. Homotetrameric potassium channel Kir3.1 (left, 2QKS); both helices in each subunit are correctly classified. Monomeric 12 transmembrane helix multiple-drug resistance transporter NorM (right, 3MKT); 10 out of 12 helices are correctly classified–one false positive and one false negative are predicted. The pore identified by Pore-Walker is shown in yellow spheres, while helices correctly predicted as pore-lining are shown in green (true positives), incorrectly predicted as pore-lining are shown in orange (false positives), correctly predicted as non-pore-lining are shown in grey (true negatives) and incorrectly predicted as non-pore-lining are shown in magenta (false negatives). In both cases, all transmembrane helices were correctly predicted by MEMSAT-SVM.