A powerful, yet easy to use, computerized analysis of competitive protein binding and radioimmunoassay data

A computerized method has been developed and tested for the automatic computation of data obtained from competitive radio-binding assays, that is easily used by computer and non-computer oriented individuals. The program requires a series of only eleven parameter lines coded from the assay protocol, followed by the data exactly as it is produced from a standard sample counter. From the set of parameters the program is able to find standard curves at scattered locations through an assay, check their "log dose-logit response" least squares linear regression equations statistically for homogeneity of slopes and elevations, pool the standard curves and check the pooled least squares linear regression equation statistically for linearity and non-linearity. The results of the pooled standard curve is presented graphically and in tabular form. Using the linear equation for the pooled standard curves, the concentration for each unknown sample is predicted with its corresponding 95% confidence interval and presented in a table of unknowns. Also provided in the table of unknowns is a mean and standard error of the mean for all biological replicates, including footnote flags to warn the user when an unknown concentration: (i) is outside the 0-100% bound range; (ii) is estimated from a range outside that covered by the standards; or (iii) was estimated from the unusable tail regions of the standard curve.