Identification and grouping of bacteria by numerical analysis of their electrophoretic protein patterns.

Improved methods for the identification and grouping of bacteria by polyacrylamide gel electrophoresis of soluble proteins are described. Electrophoretic protein patterns were obtained in rigorously standardized comditions. The results were much more reproducible than any described previously. Some of the factors affecting reproducibility were; growth conditions, time and speed of centrifugation of extracts, and conditions of gel electrophoresis. Protein patterns were compared by computing correlation coefficients from normalized densitometric tracings and clustering the strains by the unweighted average pair group method. As model systems, both Agrobacterium and Zymomonas were used because of differences in the sharpness of the peaks. The methodwas applied to 42 Agrobacterium strains. The agreement with the results of clustering by either phenotypic tests or DNA:DNA hybridization was excellent. Computerized comparisons of electrophoretic protein patterns can be a fast, easy and powerful tool for classification and identification of bacteria.