Liquid chromatography noise characteristics based on wavelet smoothing on orbitrap LC-MS data

Liquid Chromatography (LC) peak shape is assumed to be Gaussian shape by certain researchers, but from many observations and calculations it is clearly seen that LC peak shape is not Gaussian and in most cases not symmetric. Many LC peaks appear to be skewed and have a long trailing tail, other LC peaks are bimodal. To find a more accurate model for the representation of LC peak shape one must determine the characteristics and noise characteristics of LC peaks. LC peak shape is determine by many complex factors such as solvent or gradient used in the column separation, the flow rate of the elution process, physicochemical properties, hydrophobicity, and chemical structure. Elution temperature and other factors have not been characterized but could also determine the shape of LC peaks. By determining these characteristics one can developed a more accurate peak detection algorithm, but current filtering methods do not address these noise characteristics and are therefore deficient.

In this work we give a brief introduction into this area, and a thorough overview of the background. We then discuss what has been done in the literature. In particular we study noise analysis based on estimation and goodness of fit parameters, noise analysis based on variance of intensity, and LC noise filtering based on LC peak shape. After this we go over filtering methods. The methods we used are Savitzky-Golay and wavelet denoising filters. We also discuss how we determine the intensity levels and how we calculated the histograms. We then discuss the results, conclusions, and deficiencies of current methods.