Interlaboratory comparison of autoradiographic DNA profiling measurements. 2. Measurement uncertainty and its propagation.

Identifying the intrinsic sources of measurement uncertainty greatly facilitates control and further optimization of a measurement system. We have developed a model which quantitatively describes the observed interlaboratory variability of autoradiographic DNA band sizing. The model focuses on optical imaging measurements of band position and the calibration techniques used to convert measured band position to reported band size. The imaging component of measurement variability is described as a 0.05-0.2% standard deviation in determining the relative location of sample and calibration bands on a given film image. While developed solely with optical imaging information, the model is consistent with interlaboratory band sizing measurement variability observed with pristine samples. This interlaboratory variability can be modeled as a 0.2-0.4% standard deviation in the relative positions of sample and calibration bands across different electrophoretic gels. Further band sizing protocol standardization among laboratories would thus be expected to achieve at best a 2-fold reduction in interlaboratory band sizing variability.