Chemical and Random Additive Noise Elimination (CRANE)

CRANE is a wavelet transform based method for de-noising electrospray ionisation - liquid chromatography mass spectrometry data [1]

Set up environment

Inorder to set up the environment, conda is a prerequisite (https://docs.conda.io/projects/conda/en/latest/user-guide/install/). In the command line run setup-env.sh

bash setup-env.sh
conda activate crane

Set up path

path to the directory where crane is stored should be setup in the python code before being able to import the CRANE classes.

sys.path.append(crane_path)

where crane_path is the absolute path to the directory where crane is stored. See the crane_denoise_multicenter_*.py scripts in the case study section for an example.

Usage

Current implementation of CRANE accepts mass spectrometry data files in the toffee file format (Docker Image - cmriprocan/toffee:0.14.2) [2] as input. Currently the toffee file format only support Time-Of-Flight (TOF), Data-Independent Acquisition (DIA) data. However the general approach of CRANE de-noising would support all forms of electrospray ionisation - liquid chromatography mass spectrometry data.

• App class of crane.app manages the denoising of the MS file and creating a new denoised data file.
• App accepts openSWATH compatible SRLs. Alternatively one could convert PeakView compatible SRLs using the OpenSwathLibraryFromPeakview class of OpenMSToffee (Docker Image - cmriprocan/openms-toffee:0.14.3)
• Since vast amount of computer memory is required to denoise an entire MS window at once, strips of data are extracted from the MS window and processed independently. Therefore a suitable instance of the MassRangeCalculatorBase class of crane.mass_ranges should be passed when initializing crane.app
  • If the entire window is to be denoised use MassRangeCalculatorFullMapSplit class
  • If only the areas looked at by OpenSWATH is to be denoised use an instance of either MassRangeCalculatorNonOverlappingPPMFullWidth or MassRangeCalculatorNonOverlappingPixelHalfWidth
    • The n_isotopes parameter of the App class can be used to select how many isotope m/z's should be used when calculating the mass ranges based on the SRL
    • When fill_non_denoised is TRUE the mass ranges outside the mass ranges determined by the mass range calculators are copied over to the de-noised file
• De-noising parameters are managed by an instance of DenoiserBase class of crane.denoiser
• The MS1 window can be de-noised with a different set of parameters than that of MS2 by passing a different instance of DenoiserBase class as the ms1_denoiser when initializing App class
• window_name and include_ms1 parameters of the App class can be used to select which windows are to be denoised

Please refer the crane_denoise_multicenter_*.py scripts in the case study section for an example.

Running tests

pip install pytest
pytest crane

References

1. Akila J Seneviratne, Sean Peters, David Clarke, Michael Dausmann, Michael Hecker, Brett Tully, Peter G Hains and Qing Zhong, "Improved identification and quantification of peptides in mass spectrometry data via chemical and random additive noise elimination (CRANE)"

2. Brett Tully, "Toffee – a highly efficient, lossless file format for DIA-MS". Scientific Reports 2020;10(1):8939