ROSFIT: an enzyme kinetics nonlinear regression curve fitting package for a microcomputer.

Abstract A nonlinear regression curve fitting package has been specifically developed for enzyme kinetic analyses for use on the Hewlett-Packard HP-85 microcomputer. Data are entered in a conversational manner. Data can be changed, deleted or added, and data sets can be stored and retrieved from a magnetic tape cassette. Data can be fit to any of nine models: the Michaelis-Menten equation, substrate inhibition, random bi bi, ordered bi bi, ping pong bi bi, competitive inhibition, classical noncompetitive inhibition, modern noncompetitive inhibition, or uncompetitive inhibition. The printout for each model consists of several goodness-of-fit statistics, the parameter estimates with 95% confidence intervals, the variance-covariance, and correlation matrices, a residual analysis, and graphs. For example, for competitive inhibition the graphs provided are v vs [S], v vs [I], 1 v vs 1 [S] , and 1 v vs [I] with up to four concentrations of the second compound plotted on each graph. The nonlinear regression algorithm in the package is that of Marquardt. The values determined by ROSFIT are essentially the same as those found using the BMDPAR, BMDP3R, and NONLIN programs on a Univac 90 60 mainframe computer. A typical run for competitive inhibition with 46 data points took a total time of about 18.5 min, not including dat entry time, 14 min for the graph generation, 2 min for the three required iterations, and 2.5 min for miscellaneous operations.