The development and comparative evaluation of a simple programme for processing data from competitive protein binding analysis and radioimmunoassay

Abstract

The function y = A/(C + xE) where A, C and E are constants and y the number of counts specifically bound to antibody at any concentration x, was adapted to a programmable calculator with tape drive. Approximations of constants were obtained by fitting a weighted least squares regression to the linear form of the function, viz., 1 / y = C / A + ( 1 / A ) x E The estimates of A, C and E obtained, were used as a first approximation; these approximations were improved using the interactive method described by Burger, Lee and Rennie (1972) to obtain exact solutions. Using linear regression analysis, values for the data of Burger et al. obtained by this procedure were compared with Rodbard's weighted logit transformation. The slope was significantly different from 1 and the intercept significantly different from zero; the relationship between the two methods was given by the equation R = 1.05081B – 0.548872 where: R = values obtained from weighted logit transformation B = values obtained from the function y = A/C (C + xE) As a consequence, comparison of values obtained from the same set of data using the two curve fitting techniques differ. The function y = A/(C + xE) where A, C and E are constants and y the number of counts specifically bound to antibody at any concentration x, was adapted to a programmable calculator with tape drive. Approximations of constants were obtained by fitting a weighted least squares regression to the linear form of the function, viz., 1 / y = C / A + ( 1 / A ) x E The estimates of A, C and E obtained, were used as a first approximation; these approximations were improved using the interactive method described by Burger, Lee and Rennie (1972) to obtain exact solutions. Using linear regression analysis, values for the data of Burger et al. obtained by this procedure were compared with Rodbard's weighted logit transformation. The slope was significantly different from 1 and the intercept significantly different from zero; the relationship between the two methods was given by the equation R = 1.05081B – 0.548872 where: R = values obtained from weighted logit transformation B = values obtained from the function y = A/C (C + xE) As a consequence, comparison of values obtained from the same set of data using the two curve fitting techniques differ.