Improved, computer-generated system for pyocin typing of Pseudomonas aeruginosa.

We applied numerical clustering algorithms to the selection of a new indicator strain set for the pyocin typing of Pseudomonas aeruginosa. The new indicator set is composed of selected indicator strains from the sets described in 1966 by Gillies and Govan (J. Pathol. Bacteriol. 91:339-345) and in 1974 by Jones, Zakanycz, Thomas, and Farmer (Appl. Microbiol. 27:400-406) and is designated the G-F set. This indicator set consists of 14 indicator strains which typed 99.5% of 114 test cultures, has a high degree of discrimination (10 patterns encompass 50% of the test strains), and provides 62.3% reproducibility of the same typing pattern in duplicate tests done on different days. The G-F set of indicator strains provides slightly higher percentages of typable cultures than either of the other two sets, has greater discriminatory capability, and is more reproducible than they are. We recommend that the G-F set of indicator strains be used instead of the two other sets for pyocin typing of P. aeruginosa. We also tested a recently described overlay procedure for pyocin testing of P. aeruginosa and found it to be superior to previous methods in that it is easier to perform, it provides answers in only 24 h instead of 48 h, and it can be used to type mucoid strains (which previous techniques could not readily do). Thus, the application of numerical clustering algorithms and use of a revised typing procedure have produced an improved system for pyocin typing of P. aeruginosa. Similar procedures may be applicable to other typing systems.