A method for processing signals from non-stationary spectrometric processes in DNA sequencing using inverse functions

The need to analyze non-stationary processes arises in many spectrometry applications. Non-stationary processes are those in which some signal parameters change over time. For a wide variety of devices, time can be understood conditionally. For example, in spectrophotometry, it is the photon wavelength; in mass spectrometry, the molecular mass; in photoelectron spectrometry, the electron binding energy in an atom, etc. As a measure of time in most devices, one can choose a physical characteristic, according to which the spectrum scale — the abscissa axis of the graph, is calibrated. This paper discusses algorithms for processing signals from a non-stationary process, which is a set of signals from Sanger DNA sequencing analysis using capillary electrophoresis. Ideally, in such devices, the width of the spectral peaks and the time interval between the peaks of adjacent nucleotides (in the DNA strand) are the same for all channels. However, in real devices, when analyzing long DNA fragments of approximately 1000 nucleotides, the values of these parameters vary over time. For example, at the end of the analysis cycle, the peak widths increase, while the distances between them decrease, and the spectral peaks practically merge with each other. This paper examines algorithms for converting a nonstationary spectrum into a stationary one by transferring points of the original non-stationary spectrum to the scale of the output stationary spectrum using the inverse function of the time scale. The inverse function allows for correction of the non-linear time scale. The algorithms were tested on mathematical models and real signals. The test results are presented in the paper.