Prediction by a neural network of outer membrane β‐strand protein topology

Abstract

An artificial neural network (NN) was trained to predict the topology of bacterial outer membrane (OM) beta-strand proteins. Specifically, the NN predicts the z-coordinate of Calpha atoms in a coordinate frame with the outer membrane in the xy-plane, such that low z-values indicate periplasmic turns, medium z-values indicate transmembrane beta-strands, and high z-values indicate extracellular loops. To obtain a training set, seven OM proteins (porins) with structures known to high resolution were aligned with their pores along the z-axis. The relationship between Calpha z-values and topology was thereby established. To predict the topology of other OM proteins, all seven porins were used for the training set. Z-values (topologies) were predicted for two porins with hitherto unknown structure and for OM proteins not belonging to the porin family, all with insignificant sequence homology to the training set. The results of topology prediction compare favorably with experimental topology data.

The Fusobacterium nucleatum porin FomA possesses the general topology of the non-specific porins

“…Kleivdal et al (1999) recently identified positively charged amino acids in the FomA protein that were found to be important to pore function, which supported the topology model, where these amino acids are predicted to be located in transmembrane segments 5 and 6 and contribute to a hydrophilic pore lumen. In addition, when the FomA protein sequences were subjected to the recently published neural network topology prediction method (Diederichs et al, 1998), a virtually identical topology model was predicted (Fig. 1b), in spite of the fact that the FomA proteins have no sequence similarity to the proteins used to train the neural network.…”

Applications of Artificial Neural Networks in Chemical Problems

“…Neural networks have been largely employed in biochemistry and correlated research fields such as protein, DNA/RNA and molecular biology sciences [121][122][123][124][125][126][127].…”

Frequent-subsequence-based prediction of outer membrane proteins

“…In fact, even for the predictive methods that are capable of identifying OMPs, precision remains poor [4,8,14,19,25,28]. Furthermore, the datasets used to train and evaluate these existing methods are often small and not manually curated.…”

“…Scientists have previously used neural network-based methods [4,8], hydrophobicity analysis [19], and combinations of methods, including homology analysis and amino acid abundance [25,28], to varying degrees of success. The most recent approach, reported by Martelli et al [14] is, to date, the most successful attempt at OMP classification.…”