A method to determine the filter matrix in four-dye fluorescence-based DNA sequencing.

In a previous paper (Yin et al., Electrophoresis 1996, 17, 1143-1150), an automated method for matrix determination in four-dye fluorescence-based DNA sequencing was presented. As a continuation of that work, we have developed an alternative method to estimate the matrix from raw sequence data. The method uses an iterative clustering technique to associate each 4 x 1 data vector with one column of the desired filter matrix, using Kullback's I-divergence as a distance measure. The method requires less preprocessing of the data and less computation than the approach described by Yin et al. (Electrophoresis 1996, 17, 1143-1150). An example demonstrating applicability of the proposed method to Applied Biosystems sequencer data is given.