M2ara

M²ara is a software tool to facilitate the exploration of metabolomic responses in complex matrix-assisted laser desorption/ionization mass spectrometry (MALDI MS) bioassays. The app is intended for the evaluation of metabolomic drug actions by using the mass-to-charge ratios of hundreds of metabolites and it is particularly useful in defining novel pharmacodynamic biomarkers for high-throughput applications.

M²ara is based on the R package MALDIcellassay (published in Unger et. al. 2021, Nature Protocols) and extends its capabilities with a GUI and adds helpful features like clustering of curves, PCA analysis as well as the Curve Response Score (CRS) which enables fast screening for molecules regulated by drug treatment.

For more information please check out the publication.

How to use

This application simplifies the analysis of Molecular High Content Screening (MHCS) MALDI-TOF MS assay data and the evaluation of complex drug actions. After your data has been loaded, you can adjust settings as needed and start the processing. From here, you can analyze your data by selecting entries in the data table, visually inspect and rank mass features using the Curve Response Score (CRS) fingerprints, and save the curve fit and peak profile of your chosen m/z value.

This app is specifically designed for use with Bruker flex series raw data but also features support for mzML.

For more detailed information please take a look at the Manual that is also available inside the app.

Quality metrics

In M²ara the following quality metrics might be used (see Figure for an graphical explanation):

• $log_2FC$: Unadjusted magnitude of the curve. Given by the ratio between the upper ($a_u$) and lower ($a_l$) asymptotes.
• $F_Z$: Incorporates the inherent variability of the measurements given by the standard deviation ($σ_u$ and $σ_l$) and the upper ($µ_u$) and lower ($µ_l$) mean values of the two highest and lowest concentrations.
• $F_V$: Takes into account the residuals and relating them to the difference between the upper ($a_u$) and lower ($a_l$) asymptotes.

Finally, the Curve response score (CRS) combines the tree quality factors above to give a ranking of the features and a decision boundary from which on a curve can be surely classified as non-artificial.

How to install

R

Clone the GitHub repository to your local machine (please make sure to have R installed, tested with R v4.3.2) and start the app by sourcing the app.R file.

git clone https://github.com/CeMOS-Mannheim/M2ara.git

# install all packages needed
source("install_packages.R")

# run app
source("app.R")

Docker

Install the docker container, run it and access localhost:3838 to interact with the app. Don't forget to change the path c:/path/to/massSpecData to your data so that it can be mounted when running the container.

Stable version

docker pull thomasenzlein/m2ara:release

Development version

docker pull thomasenzlein/m2ara:main

Starting the container

docker run -p 3838:3838 -v  c:/path/to/massSpecData:/mnt thomasenzlein/m2ara

Stand-alone installer for Windows

Use the stand-alone installer (Windows only, no R installation needed). The installer can be downloaded here.

Example data

To test the app please use the example data on FigShare. You can find 5 datasets on FigShare, each with the corresponding processing parameter file.

For completeness find the processing parameters and expected results of two datasets (one as Bruker Flex format and one in mzML) here:

Unger2020_OATP2B1_inhibition_mzML.zip

The file contains mzML data (converted from Bruker Flex using MSConvert) originally published in Unger, et. al., 2020.

To replicate the results shown use the following parameters:

• under Settings set File Format to mzML
• set Concentration unit to uM
• set Normalization/re-calibration m/z to 354.1418 (D4-E3S, [M-H]^-)
• set recalibration tolerance to 0.1 Da
• set normalization to m/z
• deactivate smoothing and activate baseline removal
• set Aggregation method to mean
• set SNR to 3
• set alignment to 0 mDa (no alignment)
• set binning tolerance to 100 ppm
• select the folder mzML (parent folder of the mzML files) from the .zip file, please make sure that no other files are in this folder.

Alternatively, copy the this file as settings.csv into the main folder of the app.

The target m/z is 349.11 (E3S, [M-H]^-) the pIC50 value should be 6.1.

Weigt2018_BCR-ABL_inhibition_Dasatinib_BrukerFlex.zip

The file contains data in the Bruker Flex format originally published in Weigt, et. al., 2018.

To replicate the results shown use the following parameters:

• under Settings set File Format to Bruker Flex
• set Concentration unit to uM
• set Normalization/re-calibration m/z to 760.5851 (PC(34:1) [M+H]⁺)
• set recalibration tolerance to 0.1 Da
• set normalization to TIC
• activate smoothing and baseline removal
• set Aggregation method to mean
• set SNR to 3
• set alignment to 0 mDa (no alignment)
• set binning tolerance to 100 ppm
• select the the folder curve from the .zip file, make sure no other files/folders are present.

Alternatively, copy the this file as settings.csv into the main folder of the app.

The target is m/z 826.5722 (PC(36:1) [M+K]⁺) and m/z 616.1767 (Heme B [M+H]⁺) the pIC50 values should be 9.5 and 9.7.