Estimation of restriction maps with known site order using a generalized linear model.

A generalized linear model with Gamma errors is used to estimate the coordinates of a restriction map when the site order is known. This can be conveniently programmed in a wide range of statistical packages (e.g. Genstat 5, Minitab, SAS), and gives maximum likelihood estimates with their associated optimal properties. Regression diagnostics allow the checking of assumptions and help to identify mis-specified, influential or discordant fragment lengths. A specific diagnostic for identifying fragment lengths causing reversal of restriction site order is derived. Exact 'fragment' lengths from DNA sequencing can be conveniently included in an approximate manner by giving them a larger weight than observed restriction fragment lengths. Two examples and the Genstat 5 codes used in their analysis are presented.