ANALYSIS OF PROTEIN TRANSMEMBRANE HELICAL REGIONS BY A NEURAL-NETWORK

Neural networks were used to generalize common themes found in transme mbrane-spanning protein helices. Various-sized databases were used con taining nonoverlapping sequences, each 25 amino acids long. Training c onsisted of sorting these sequences into 1 of 2 groups: transmembrane helical peptides or nontransmembrane peptides. Learning was measured u sing a test set 10% the size of the training set. As training set size increased from 214 sequences to 1,751 sequences, learning increased i n a nonlinear manner from 75% to a high of 98%, then declined to a low of 87%. The final training database consisted of roughly equal number s of transmembrane (928) and nontransmembrane (1,018) sequences. All t ransmembrane sequences were entered into the database with respect to their lipid membrane orientation: from inside the membrane to outside. Generalized transmembrane helix and nontransmembrane peptides were co nstructed from the maximally weighted connecting strengths of fully tr ained networks. Four generalized transmembrane helices were found to c ontain 9 consensus residues: a K-R-F triplet was found at the inside l ipid interface, 2 isoleucine and 2 other phenylalanine residues were p resent in the helical body, and 2 tryptophan residues were found near the outside lipid interface. As a test of the training method, bacteri orhodopsin was examined to determine the position of its 7 transmembra ne helices.