TidyMass: An Object-oriented Reproducible Analysis Framework for LC-MS Data

Abstract

Reproducibility and transparency have been longstanding but significant problems for the metabolomics field. Here, we present the tidyMass project (https://www.tidymass.org/), a comprehensive computational framework that can achieve the shareable and reproducible workflow needs of data processing and analysis for LC-MS-based untargeted metabolomics. TidyMass was designed based on the following strategies to address the limitations of current tools: 1) Cross-platform utility. TidyMass can be installed on all platforms; 2) Uniformity, shareability, traceability, and reproducibility. A uniform data format has been developed, specifically designed to store and manage processed metabolomics data and processing parameters, making it possible to trace the prior analysis steps and parameters; 3) Flexibility and extensibility. The modular architecture makes tidyMass a highly flexible and extensible tool, so other users can improve it and integrate it with their own pipeline easily.

Deep Learning-based Pseudo-Mass Spectrometry Imaging Analysis for Precision Medicine

“…The pseudo-MS image converter is designed and developed to convert the LC-MS-based untargeted metabolomics raw data to pseudo-MS images. Briefly, the LC-MS-based untargeted metabolomics raw data (from mass spectrometry instrument) is first converted to mzXML format data using msConvert 20 or massconverter 21 . And then, the mzXML format data is imported to the R environment using the readMSData function from the MSnbase package 22 .…”

“…Finally, the generated MS 1 metabolic feature table (peak table) includes the mass-to-charge ratio ( m/z ), retention time (RT, second), peak abundances for all the samples, and other information. This MS 1 metabolic feature table is used for the subsequent data cleaning using the masscleaner package from the tidyMass project 21 . Briefly, the features detected in less than 20% QC samples were removed as noisy from the metabolic feature table.…”

Deep learning-based pseudo-mass spectrometry imaging analysis for precision medicine

“…The pseudo-MS image converter is designed and developed to convert the LC-MS-based untargeted metabolomics raw data to pseudo-MS images. Briefly, the LC-MS-based untargeted metabolomics raw data (from mass spectrometry instrument) is first converted to mzXML format data using msConvert 20 or massconverter 21 . And then, the mzXML format data is imported to the R environment using the readMSData function from the MSnbase package 22 .…”

“…Alignment of two metabolic peak tables. Two metabolic feature tables were aligned according to m/z and RT using the masstools package (mz_rt_match function) from the tidyMass project 21 . Briefly, only the features in two metabolic feature tables within the setting cutoff for m/z matching (< 10 ppm) and RT matching (< 30 seconds) are considered the same features.…”

“…LC-MS-based untargeted metabolomics raw data processing. The mzXML format data (RPLC positive mode) were placed into different folders according to their class (for example ''Blank'', ''QC'' and ''Subject'') and then subjected to peak detection and alignment using the massprocesser package from the tidyMass project 21 based on XCMS 24 . Briefly, the peak detection and alignment were performed using the centWave algorithm 24 .…”