论文部分内容阅读
Accurate peak detection is essential for analyzing high-throughput dataset generated by chromatograph,mass spectrometry and Raman spectrometer.Local maximum with signal to noise ratio(SNR)over a certain threshold is frequently used for peak detection,but its sensitivity decreases severely for high noise and baseline dataset.Continuous wavelet transform(CWT)-based method is more practical and popular in this situation,which can increase accuracy and reliability by identifying peaks across scales in wavelet space and implicit removal of noise and baseline.However,its computational load is relatively high and the estimated peak position may not be accurate in the case of special peaks such as overlapping,asymmetric and weak peaks.In this study,we present an approach for peak detection by taking full advantage of additional information in wavelet space including ridge lines,valley lines and zero-crossing,which have all the advantages of CWT-based method and can estimate peak position accurately even for these special peaks.Furthermore,our approach has been designed and implemented efficiently in Python and Cython programming language.Therefore,it is about an order of magnitude faster than Mass Spec Wavelet,which is particularly suitable for detecting peaks in high-throughput and hyphenated datasets.Our approach was comprehensively evaluated with both simulation dataset and typical analytical instruments datasets.Results show that our approach has significantly higher identification accuracy and speed by comparing it with the popular Mass Spec Wavelet package using these datasets.