Base sequence determination apparatus, capillary array electrophoresis apparatus, and method

A base sequence determination apparatus comprising:

• a memory;
• an input device;
• a sequence output device;
• and
• a processor communicatively coupled to the memory, the input device and the sequence output device, wherein the processor is configured to:
• divide a plurality of signals of each channel of a color-converted signal into multiple blocks;
• output a mobility correction signal obtained by mobility correction of a time-series signal of a wavelength spectrum corresponding to each base using the multiple blocks;
• receive, via the input device, a tolerance that gives a weighting coefficient;
• changes the tolerance with an increment to a changed tolerance, the increment being within a range;
• execute a process of calculating a deconvoluted signal of the mobility correction signal respectively for a plurality of parameter candidates of point spread function, a process of calculating variance of peak intervals for the calculated deconvoluted signal, a process of specifying a parameter of the point spread function using the calculated variance and the changed tolerance, and a process of outputting the deconvoluted signal corresponding to the point spread function having the specified parameter as an updated deconvoluted signal;
• a peak waveform from the updated deconvoluted signal and outputs an updated peak-extracted signal;
• receive the updated peak-extracted signal and determine a base sequence;
• calculate a coincidence rate between the base sequence and a correct sequence;
• and
• calculate, using machine learning, an updated tolerance causing the coincidence rate to be set at a maximum value, such that the updated deconvolution signal is calculated using the updated tolerance;
• and
• tune, using machine learning, a plurality of parameters to determine a level of accuracy of the original peak, and
• wherein calculating the updated tolerance and tuning the plurality of parameters are alternately performed such that the specified parameter is optimized each time.