A Markov Chain Monte Carlo Method for Estimating the Statistical Significance of Proteoform Identifications by Top-Down Mass Spectrometry.
Qiang KouZhe WangRachele A LubeckyjSi WuLiangliang SunXiaowen LiuPublished in: Journal of proteome research (2019)
Top-down mass spectrometry is capable of identifying whole proteoform sequences with multiple post-translational modifications because it generates tandem mass spectra directly from intact proteoforms. Many software tools, such as ProSightPC, MSPathFinder, and TopMG, have been proposed for identifying proteoforms with modifications. In these tools, various methods are employed to estimate the statistical significance of identifications. However, most existing methods are designed for proteoform identifications without modifications, and the challenge remains for accurately estimating the statistical significance of proteoform identifications with modifications. Here we propose TopMCMC, a method that combines a Markov chain random walk algorithm and a greedy algorithm for assigning statistical significance to matches between spectra and protein sequences with variable modifications. Experimental results showed that TopMCMC achieved high accuracy in estimating E-values and false discovery rates of identifications in top-down mass spectrometry. Coupled with TopMG, TopMCMC identified more spectra than the generating function method from an MCF-7 top-down mass spectrometry data set.
Keyphrases
- mass spectrometry
- liquid chromatography
- capillary electrophoresis
- gas chromatography
- high performance liquid chromatography
- high resolution
- machine learning
- monte carlo
- density functional theory
- deep learning
- small molecule
- high throughput
- big data
- neural network
- artificial intelligence
- simultaneous determination
- binding protein
- genetic diversity