Objective To perform the identification at the subspecies-level of Mycobacterium abscessus(M.abscessus)and analyze its characteristics based on matrix-assisted laser desorption ionization time-of-flight mass spectrometry(MALDI-TOF MS).Methods Bi-otyper software was used to construct the predicted peak spectrum of M.abscessus subsp.abscessus and M.abscessus subsp.massiliense.The predicted peak spectrum was constructed with expected maximum peak value number of 70 and peak frequency of 100％in the ex-perimental group and control group,respectively.A blind test was performed on 31 strains of M.abscessus that were not used to con-struct predictive peak spectra to evaluate the identification efficiency of predictive peak spectra.FlexAnalysis software was used to sum-marize and analyze the list of mass spectral peak value of M.abscessus,and screen the specific peaks in mass spectra of different sub-species of M.abscessus.The principal component analysis(PCA)algorithm was used to perform the cluster analysis for the data from mass spectrometry of M.abscessus,and explore the feasibility of PCA clustering in distinguishing the subspecies of M.abscessus.Results In the experimental group,96.8％(30/31)of the strains were correctly identified,and one strain of M.abscessus subsp.massiliense with rough colony form was mistakenly identified as M.abscessus subsp.abscessus.In control group,77.4％(24/31)of the strains were correctly identified,but 7 strains of M.abscessus subsp.massiliense were incorrectly identified or unable to be identified.The identification efficiency in the experimental group was significantly better than that in the control group with statistical difference(X2=5.167,P=0.026).M.abscessus subsp.abscessus exhibited three specific peaks(m/z 4 001.67,4 386.81 and 4 963.17),and M.abscessus subsp.massiliense also exhibited three specific peaks(m/z 4 950.48,4 381.78 and 5 214.90).In the PCA 3D scatter plot,the data points of M.abscessus subsp.abscessus and M.abscessus subsp.massiliense were relatively dispersed without obvious clus-tering.The PC A dendrograph could be divided into six branches in which only four branches were composed of a single subspecies.The minimum level value of distance between M.abscessus subsp.abscessus and M.abscessus subsp.massiliense was about 0.1.Conclusion The predicted peak spectrum based on MALDI-TOF MS with the expected maximum peak number of 70 could accurately identify M.abscessus at the subspecies level.The specific peak of mass spectrometry method in this study should be feasible to distinguish the subspecies of M.abscessus subsp.abscessus and the subspecies of M.abscessus subsp.Massiliense,but PCA cluster analysis cannot be used as a means to distinguish M.abscessus subsp.abscessus from M.abscessus subsp.massiliense.

Objective To perform the identification at the subspecies-level of Mycobacterium abscessus(M.abscessus)and analyze its characteristics based on matrix-assisted laser desorption ionization time-of-flight mass spectrometry(MALDI-TOF MS).Methods Bi-otyper software was used to construct the predicted peak spectrum of M.abscessus subsp.abscessus and M.abscessus subsp.massiliense.The predicted peak spectrum was constructed with expected maximum peak value number of 70 and peak frequency of 100％in the ex-perimental group and control group,respectively.A blind test was performed on 31 strains of M.abscessus that were not used to con-struct predictive peak spectra to evaluate the identification efficiency of predictive peak spectra.FlexAnalysis software was used to sum-marize and analyze the list of mass spectral peak value of M.abscessus,and screen the specific peaks in mass spectra of different sub-species of M.abscessus.The principal component analysis(PCA)algorithm was used to perform the cluster analysis for the data from mass spectrometry of M.abscessus,and explore the feasibility of PCA clustering in distinguishing the subspecies of M.abscessus.Results In the experimental group,96.8％(30/31)of the strains were correctly identified,and one strain of M.abscessus subsp.massiliense with rough colony form was mistakenly identified as M.abscessus subsp.abscessus.In control group,77.4％(24/31)of the strains were correctly identified,but 7 strains of M.abscessus subsp.massiliense were incorrectly identified or unable to be identified.The identification efficiency in the experimental group was significantly better than that in the control group with statistical difference(X2=5.167,P=0.026).M.abscessus subsp.abscessus exhibited three specific peaks(m/z 4 001.67,4 386.81 and 4 963.17),and M.abscessus subsp.massiliense also exhibited three specific peaks(m/z 4 950.48,4 381.78 and 5 214.90).In the PCA 3D scatter plot,the data points of M.abscessus subsp.abscessus and M.abscessus subsp.massiliense were relatively dispersed without obvious clus-tering.The PC A dendrograph could be divided into six branches in which only four branches were composed of a single subspecies.The minimum level value of distance between M.abscessus subsp.abscessus and M.abscessus subsp.massiliense was about 0.1.Conclusion The predicted peak spectrum based on MALDI-TOF MS with the expected maximum peak number of 70 could accurately identify M.abscessus at the subspecies level.The specific peak of mass spectrometry method in this study should be feasible to distinguish the subspecies of M.abscessus subsp.abscessus and the subspecies of M.abscessus subsp.Massiliense,but PCA cluster analysis cannot be used as a means to distinguish M.abscessus subsp.abscessus from M.abscessus subsp.massiliense.