俞卓琦, 王伟民, 王华, 徐炳寅, 吴焕铭, 唐科奇, 丁力. 一种平面静电离子阱非正弦镜像电荷/电流信号的定量转换新算法[J]. 质谱学报, 2024, 45(3): 332-342. DOI: 10.7538/zpxb.2024.0017
引用本文: 俞卓琦, 王伟民, 王华, 徐炳寅, 吴焕铭, 唐科奇, 丁力. 一种平面静电离子阱非正弦镜像电荷/电流信号的定量转换新算法[J]. 质谱学报, 2024, 45(3): 332-342. DOI: 10.7538/zpxb.2024.0017
YU Zhuo-qi, WANG Wei-min, WANG Hua, XU Bing-yin, WU Huan-ming, TANG Ke-qi, DING Li. A New Algorithm for Quantitative Conversion of Non-sinusoidal Image Charge or Current Signal from Planar Electrostatic Ion Trap[J]. Journal of Chinese Mass Spectrometry Society, 2024, 45(3): 332-342. DOI: 10.7538/zpxb.2024.0017
Citation: YU Zhuo-qi, WANG Wei-min, WANG Hua, XU Bing-yin, WU Huan-ming, TANG Ke-qi, DING Li. A New Algorithm for Quantitative Conversion of Non-sinusoidal Image Charge or Current Signal from Planar Electrostatic Ion Trap[J]. Journal of Chinese Mass Spectrometry Society, 2024, 45(3): 332-342. DOI: 10.7538/zpxb.2024.0017

一种平面静电离子阱非正弦镜像电荷/电流信号的定量转换新算法

A New Algorithm for Quantitative Conversion of Non-sinusoidal Image Charge or Current Signal from Planar Electrostatic Ion Trap

  • 摘要: 傅里叶变换质谱(FTMS)的质量分辨率取决于镜像电荷/电流的采集时间。为了在较短的时间内获得较高的分辨率,除增加分析器场强外,还可以利用镜像电荷信号中的高次谐波。平面静电离子阱获得的镜像电荷信号具有非正弦波形,其中包含许多高次谐波成分,增加了频谱分析的复杂程度,例如,难以辨认1个谱峰的谐波次数,以及当不同次的谐波发生重叠时出现定量困难。为了将多质荷比离子的镜像电荷信号转换成质谱,本文开发了一种新的定量算法,包含谱峰评分分类(SC)和最小二乘法拟合(LSF)。通过评分可以确定每个峰属于哪一次谐波,从而确定其对应的基频。打分算法列出所有候选离子的基频,并使用确定的基频构建所有基信号,然后用其对原频谱进行LSF,以确定每种质荷比离子的数量。使用48种不同质荷比离子的仿真镜像电荷信号对SC-LSF 算法进行测试。结果表明,该方法允许较宽的质量范围,即使在高噪声条件下,通过 SC-LSF算法也能准确得到各种不同质荷比离子的数量。因为频谱中通常存在大量的空白频点,可以选择1个频点子集进行LSF,这相对于时域中的LSF,能够大大减少计算量,提高效率和准确度。此外,只有使用复数频谱数据进行LSF时才能获得良好的定量效果,而使用幅值数据则会对质荷比接近的离子定量造成较大的误差,这是因为质荷比接近的离子信号的低次谐波峰未能彻底分解,而复数的幅值不具有可加性,从而造成拟合的错误。

     

    Abstract: Mass resolution of Foureir transform mass spectrometer (FTMS) highly depends on the acquisition time of image charge/current signal. In order to achieve higher resolution with a shorter transient time, apart from increasing the field strength of the analyzer, the high order harmonics in the image charge signal may also be exploited. The image charge or current signal obtained from the planar electrostatic ion trap has a non-sinusoidal waveform which contains many high-order harmonic components. However, presence of high harmonics increases the complexity of spectral analysis, such as identification of a peak for its correct harmonic order, and avoiding quantitation error due to the peak overlapping from different harmonic groups. A new quantitative algorithm consists of a scoring-based peak classification and the least square fitting (SC-LSF), which has been developed to convert image charge or current signal to mass spectrum. The scoring process will go through all the peaks identified above the noise background, for assumptions that the peak belongs to a certain harmonic order. The score will go up when a relevant lower harmonic peak is confirmed. The harmonic order which achieves the highest score, is assigned to the peak so its fundamental frequency can be determined. Through the SC tests, the candidates of all fundamental frequencies are found for all possible m/z of ions. The basis signals for all possible m/z are constructed using the identified fundamental frequencies and are brought to the LSF to determine the intensities of each species. The SC-LSF algorithm was tested using simulated signal from a mixture of 48 different m/z ions. Different levels of artificial noise were added to the signal to challenge the algorithm. The results show a wide range of mass and ion numbers in the sample mixture can be accurately returned through the SC-LSF algorithm even if the transient signal is under high noise condition. LSF in frequency domain is more efficient than in time domain because a subset of frequency points may be selected, so the amount of calculation is massively reduced. In addition, the test proves that good quantitation can be achieved only when LSF is carried out using data in complex number, while fitting with magnitude data results in large errors for ions in the closed mass group. This is because those closed mass peaks are overlapping at certain low-order harmonics and the amplitudes of a FFT spectrum are not addable, thus leading to the fitting errors.

     

/

返回文章
返回