[A novel analyzing method for the signal denoising of DNA sequencing]

Fluorescence signals in DNA sequencing are often contaminated by noise, which has negative influence on the accuracy and detection limit of analysis. Wavelet analysis has excellent time and frequency domain resolution for signal denoising compared to other conventional filtering methods. Before the signal denoising process, a key problem is how to choose a suitable wavelet base, decomposition level and denoising threshold, which have great influence on the quality of signal denoising. In order to construct the same noise model as that in experiment and evaluate the denoising algorithm precisely, a novel method is presented: the real noise signal acquired from the experimental system was added to an ideal signal to simulate a noisy DNA sequencing signal, thus the denoising efficiency could be evaluated accurately. The denoising results indicate that using sym 7 wavelet base, decomposition level at 5 and using fixed form soft threshold can effectively reduce the noise. After being processed, the SNR was improved more than 5 times. When the same algorithm was applied to the experimental DNA sequencing data, the results were more credible than those obtained through other algorithms based on the random noise model.