Download Full Text (1.0 MB)
Background: Metabolomic and lipidomic studies generate vast quantities of data that are often analysed in a closed software environment with little to no access to the underlying algorithms. As a result, data processed via different software pipelines yield different results thus leading to a widespread problem of low reproducibility within these fields. To address this problem, we are developing LipidAnalyst; an R based lipidomics software pipeline. As a part of this project, we are creating a simple statistical analysis and graphing module in R to generate accurate, reproducible, high-resolution figures.
Methods: R scripts were developed under version 3.5.3 with the capability to undertake statistical analyses (e.g. ANOVA) and post-hoc tests (e.g. Tukey). Additional code plotted resultant information as high resolution violin and box plots that depicted statistical significance. Thereafter, lipidomic and metabolomic data were analysed by this code and compared against commercial software and Metaboanalyst, a primary software used in metabolomic and lipidomic research.
Results: Code generated in house demonstrated the same results as those generated using commercial software (e.g. JMP 14.0 Pro) but were different from results obtained by using the MetaboAnalyst pipeline.
Conclusions: This study demonstrated the prevalent danger of using closed-source software pipelines for the analysis of lipidomic and metabolomic data without validating the analysis outcomes via open-source software. Open source software such as LipidAnalyst, that has also been independently validated using multiple data sets, can then be published with the results to enable transparency of data analysis and improve the replicability of results across different labs.
Pharmacy and Pharmaceutical Sciences
Is Part Of
VCU Graduate Research Posters