Limits on α‐helix prediction with neural network models

Using a backpropagation neural network model we have found a limit for secondary structure prediction from local sequence. By including only sequences from whole α‐helix and non‐α‐helixstructures in our training and test sets—sequences spanning boundaries between these two structures were excluded—it was possible to investigate directly the relationship between sequence and structure for α‐helix. A group of non‐α‐helix sequences, that was disrupting overall prediction success, was indistinguishable to the network from α‐helix sequences. These sequences were found to occur at regions adjacent to the termini of α‐helices with statistical significance, suggesting that potentially longer α‐helices are disrupted by global constraints. Some of these regions spanned more than 20 residues. On these whole structure sequences, 10 residues in length, a comparatively high prediction success of 78% with a correlation coefficient of 0.52 was achieved. In addition, the structure of the input space, the distribution of β‐sheet in this space, and the effect of segment length were also investigated. © 1992 Wiley‐Liss, Inc.