Computerized estimation of size of nucleic acid fragments using the four-parameter logistic model.

We have developed improved methods for statistical estimation of the size of linear duplex DNA after continuous- or pulsed-field electrophoresis in agarose and polyacrylamide gels. We employ the four-parameter logistic model to describe the smooth, symmetrical, sigmoidal relationship between electrophoretic mobility (distance migrated) and log of molecular size (kb): distance = a - d/1 + (size/c)b + d. The four parameters (a,b,c,d) are estimated by nonlinear least-squares curve fitting using the Marquardt-Levenberg algorithm, where a represents an upper plateau, b a slope factor, c the midpoint, and d the lower plateau. Estimates of size for unknown species are accompanied by estimates of the standard error (SE) or coefficient of variation (%CV). A plot of SE and %CV versus size results in a "precision profile" which objectively defines the useful range of the calibration curve. This logistic relationship is more general than the rectangular hyperbola or linear methods, provides excellent goodness of fit, and can be used as a "global" method for the entire calibration curve, rather than as a "local" method for small segments of the curve.(ABSTRACT TRUNCATED AT 250 WORDS)