MCnebula: Chemical Clustering Visualization Analysis Strategy Based On Non-targeted LC-MS/MS Technology To Quickly Analyze The Complex Chemical Components Of Traditional Chinese Medicin

Objective:Establishing a convenient LC-MS/MS analysis technique adapted to the current development of artificial intelligence technology in the field of mass spectrometry.Methods:Combine cutting-edge predictive technologies of software of SIRIUS,visualization of molecular network,statistical screening,and object-oriented programming in R language to develop workflows for LC-MS/MS analysis.Results:We established a framework called MCnebula(Multiple-Chemical nebula)to facilitate mass spectrometry data analysis process by focusing on critical chemical classes and visualization in multiple dimensions.It consisted of three vital steps:(1)abundance-based classes(ABC)selection algorithm,(2)critical chemical classes to classify ’features’(compounds),(3)visualization as multiple Child-Nebulae(network graph)with annotation,chemical classification and structure.Notably,MCnebula can be applied to explore classification and structural characteristic of unknown compounds that beyond the limit of spectral library.What’s more,it is intuitive and convenient for pathway analysis and biomarker discovery due to its function of ABC selection and visualization.MCnebula was implemented in the R language.We provided a series of tools in the R packages to facilitate downstream analysis in a MCnebula-featured way,including feature selection(statistical analysis of binary comparisons),homology tracing of top features,pathway enrichment analysis,heat map clustering analysis,spectral visualization analysis,chemical information query and output analysis reports,etc.We evaluated the performance of MCnebula for chemical classifying and identification.The classified performance of MCnebula was evaluated with three datasets:Stability after adding noise(89.5%,81.2%);Precision(69.8%,67.1%,67.4%);Recall(82.2%,81.6%,81.6%).MCnebula chemical classifying outperformed than the counterpart GNPS method.the identification accuracy of MCnebula fluctuated around 70%.In order to illustrate the broad utility of MCnebula,we investigated a human-derived serum dataset for metabolomics analysis.The results indicated that ’Acyl carnitines’ were screened out by tracing structural classes of biomarkers which was consistent with the reference;We also identified other chemical classes associated with disease development,’Lysophosphatidylcholines’ and ’Bile acids,alcohols and derivatives’;we identified 1086 metabolites.We investigated a plant-derived dataset of herbal E.ulmoides to achieve a rapid unknown compound annotation and discovery,and identified 565 chemical compounds.The usage of MCnebula is introduced at:http://www.mcnebula.org/.Conclusion:MCnebula workflows are broadly powerful and adaptable to complex metabolomics data analysis and phytopharmaceutical data analysis.