Data processing and pattern recognition in high-throughput capillary electrophoresis

A specific method for massive Capillary Electrophoresis data analysis based on pattern recognition techniques in the wavelet domain is presented. Low-resolution, denoised electropherograms are obtained by applying several pre-processing algorithms including discrete wavelet transform, denoising, detection of region of interest and baseline correction. The resultant signal is mapped into multi-character sequences exploiting the first derivative information and multi-level peak height quantization. Next, local alignment algorithms are applied on the coded sequence for peak pattern recognition. Finally, Gaussian approximation is performed to assure precise peak-height measurements.