Topology prediction for helical transmembrane proteins at 86% accuracy–Topology prediction at 86% accuracy

Abstract

Previously, we introduced a neural network system predicting locations of transmembrane helices (HTMs) based on evolutionary profiles (PHDhtm, Rost B, Casadio R, Fariselli P, Sander C, 1995, Protein Sci 4:521-533). Here, we describe an improvement and an extension of that system. The improvement is achieved by a dynamic programming-like algorithm that optimizes helices compatible with the neural network output. The extension is the prediction of topology (orientation of first loop region with respect to membrane) by applying to the refined prediction the observation that positively charged residues are more abundant in extra-cytoplasmic regions. Furthermore, we introduce a method to reduce the number of false positives, i.e., proteins falsely predicted with membrane helices. The evaluation of prediction accuracy is based on a cross-validation and a double-blind test set (in total 131 proteins). The final method appears to be more accurate than other methods published: (1) For almost 89% (+.3%) of the test proteins, all HTMs are predicted correctly. (2) For more than 86% (*3%) of the proteins, topology is predicted correctly. (3) We define reliability indices that correlate with prediction accuracy: for one half of the proteins, segment accuracy raises to 98%; and for two-thirds, accuracy of topology prediction is 95%. (4) The rate of proteins for which HTMs are predicted falsely is below 2% (k 1070). Finally, the method is applied to 1,616 sequences of Haemophilus influenzae. We predict 19% of the genome sequences to contain one or moreHTMs. This appears to be lower than what we predicted previously for the yeast VI11 chromosome (about 25%).Keywords: .dynamic programming; genome analysis; Haemophilus influenzae; postprocessing neural network output; secondary structure prediction; structure prediction for integral membrane proteins; topology prediction for helical transmembrane proteins Integral membrane proteins comprise an important class of proteins for which experimental techniques for 3D structure determination are often not applicable. Fortunately, theoretical prediction of structural aspects is simpler for membrane proteins than for globular proteins because the lipid bilayer imposes strong constraints on the degrees of freedom for the 3D struc-

Functional Domains of NosR, a Novel Transmembrane Iron-Sulfur Flavoprotein Necessary for Nitrous Oxide Respiration

“…It predicts a cytoplasmic location for the C-terminal Fe-S domain (3,13,53). The same topology was obtained from neural network modeling based on multiple sequence alignment (37). The hydrophobic region TM6 is problematic since it is variably predicted as one or two helices (the latter variant would result in the periplasmic location of the Fe-S domain).…”

A key role for the peroxisomalABCD2transporter in fatty acid homeostasis

“…Secondary structure and transmembrane predictions of ABCD1 and ABCD2 were performed using PredictProtein (www.predictprotein.org; Ref. 50) and PHDhtm, a program to predict the location and topology of transmembrane helices from multiple sequence alignments (49).…”

USH3A transcripts encode clarin-1, a four-transmembrane-domain protein with a possible role in sensory synapses

“…Sequences were aligned using the BlockMaker 34 and MACAW 35 programs; PHYLIP ProtDist 36 PHYLIP -Phylogeny Inference Package (Version 3.2 Cladistics 5: 164 -166) and CLUSTAL W programs 37 were used to compute trees and their significance (bootstrap) values. TM regions were predicted using the PHDhtm 38 and TMHMM programs. 39 Primers were designed using the Oligo primer analysis software (http://medprobe.com/is/oligo/html).…”