DNA sequencing and parametric deconvolution

One of the key practices of the Human genome project is Sanger DNA sequencing. Its data analysis part is called base-calling, which attempts to reconstruct target DNA sequences from fluorescence intensities generated by sequencing machines. In this paper, we present our modeling framework of DNA sequencing, in which a base-calling scheme arises naturally. A large portion of DNA sequencing errors come from the diffusion effect in electrophoresis, and deconvolution is the tool to solve this problem. We present a new version of the parametric deconvolution which is motivated by the spike-convolution model, and some recently obtained results regarding its asymptotics. One application of the asymptotics is to look at the resolution issue from the perspective of confidence intervals. We also report on an empirical study of the progressiveness of electrophoretic diffusion by way of estimating the slowly-changing width parameter in the spike-convolution model. Furthermore, we include an example of complete preprocessing of DNA sequencing data. Running title: DNA sequencing and parametric deconvolution