Attempts to predict protein secondary structural class using nonlinear statistical techniques showed that a level of 55% correct predictions is attainable. This result is substantiated using cross-validation techniques. Predicting the absolute level of alpha helix content could also be accomplished with substantial success. Inspection of the Brookhaven protein database did not support the claims for the existence of distinct structural classes. We have created a novel algorithm for aligning multiple protein sequences. This approach iteratively utilizes pairwise alignment of subgroups of the original group of proteins and is able to substantially reduce the computer space and time required to produce an optimal result. A technique for producing scientific posters on the Macintosh computer was worked out. This procedure utilizes the advanced graphic and page-layout capabilities of the Mac combined with commercially available high resolution printing to produce standard one-piece posters for presentation at conferences. The procedure potentially reduces time and costs associated with poster preparation.
