The code that supports the findings in our paper titled “Cumulative Learning Enables Convolutional Neural Network Representations for Small Mass Spectrometry Data Classification”
Data availability
All MS data can be found in : Canine sarcoma raw library is accessible on the ProteomeXchange consortium: PXD010990. Human ovarian datasets can be accessed through FDA-NCI Clinical Proteomics at https://home.ccr.cancer.gov/ncifdaproteomics/ppatterns.asp. Microorganisms, beef liver, and rat brain raw libraries are accessible on https://data.mendeley.com/datasets/33cbb37cs2/1.
Data description
Raw SpiderMass spectra are converted into mzXML format using the 64-bit MSConvert tool (version 3.0), part of the ProteoWizard suite. Spectra with a total ion count (TIC) exceeding 1e4 count for irradiation detection are selected using the MSnbase package (version 1.20.7, R version 3.4.4) and converted into a csv file. Raw ovarian datasets are imported into a csv file format.
Citation request
[1] Please consider citing the following paper: Seddiki K., Saudemont K., Precioso F., Ogrinc N., Wisztorski M., Salzet M., Fournier I., Arnaud Droit A. submitted. Cumulative Learning with Convolutional Neural Networks Enables Small Mass Spectrometry Data Classification.
Contributing
For any questions, feel free to open an issue or contact at arnaud.droit@crchudequebec.ulaval.ca