Tandem mass spectrometry data precursor ion detection model training method and precursor ion detection method

Abstract

The invention provides a method for training a tandem mass spectrometry parent ion detection model, comprising the following steps: 1) obtaining a known spectrum data set of the parent ion, and for each secondary spectrum, determining the candidate precursor ion of the secondary spectrum ; 2) Extract the eigenvector of each secondary spectrogram-candidate precursor ion combination, and carry out corresponding assignment according to whether the secondary spectrogram and candidate precursor ion are paired correctly; Wherein, the element of feature vector comprises: isotopic peak cluster is similar degree, the spectral peak intensity ratio in the fragmentation window, the chromatographic similarity and the virtual chromatographic similarity; 3) the eigenvectors of all secondary spectra-candidate precursor ion combinations are used as input, and the secondary spectra and candidate precursor ions are paired The correct or incorrect assignment is used as an output to train the MARS model and obtain a tandem mass spectrometry precursor ion detection model. The invention also provides a corresponding parent ion detection method. The invention can improve the recall rate of precursor ions and improve the detection speed of precursor ions.

Description

technical field [0001] The invention relates to the technical field of bioinformatics, in particular, the invention relates to a method for training a parent ion detection model of tandem mass spectrometry data in proteomics and a method for detecting precursor ions. Background technique [0002] The shotgun method is one of the important methods for identifying proteins in bioinformatics. It cuts proteins in biological samples into peptides, and then sends them to the tandem mass spectrometer to obtain the corresponding mass spectrometry data sets, and then searches through the tandem mass spectrometry database. Algorithms, such as SEQUEST, Mascot, pFind, etc., can identify peptides from tandem mass spectrometry data, and further infer peptides to proteins to obtain protein information in the sample. [0003] The generation of peptide biological samples to mass spectrometry data is divided into two stages: in the first stage, peptides enter the mass spectrometer in batches,...

AI Technical Summary

Problems solved by technology

However, the current algorithms such as Hardklor, pParse, and MaxQuant all need to manually select the relevant parameters of each feature, and compare repeatedly on multiple data sets, and gradually adjust the value of each parameter to achieve a certain recall rate, so it is very cumbersome to use.

In terms of detection speed, even with the fastest Hardklor algorithm, it takes more than 15 minutes to export the precursor ions of 9000 spectra, which is difficult to meet the requirements of real-time data analysis

Moreover, the recall rate of the existing precursor ion detection algorithm also needs to be further improved

Images