Improve integration of R and Python-based libraries for mass spectrometry data analysis

[jorainer]

Title

SpectriPy

Abstract

Several powerful software packages for mass spectrometry (MS) data analysis exist in both Python and R. These include among others the Python libraries matchms, pyOpenMS, Pyteomics and spectrum_utils, as well as Bioconductor's Spectra, and related packages from the RforMassSpectrometry ecosystem for metabolomics and proteomics data analysis in R. Each of these packages have their own specific use cases and functionality. To fully benefit from the rich and already validated functionality of these tools and to avoid re-implementation of functionality in another programming language, a better
integration is needed.

The SpectriPy R package provides a first proof-of-concept implementation for the seamless integration of functionality from Python libraries into R software. The package translates between R and Python MS data structures and allows to perform spectra similarity calculations using Python's matchms library, directly from within R.

In this hackathon we hope to benefit from contributions and input from participating Python and R developers as well as data analysts. We propose to extend the SpectriPy package to include more functionality also from other Python libraries. Ultimately this will enable the seamless integration of Python MS libraries in reproducible R-based analysis workflows.

Project Plan

• Implement and improve functionality to efficiently convert between R and python-based MS data representations and objects.
• Improve documentation and implement tutorials showcasing the functionality on dedicated use cases.
• Extend the functionality of the SpectriPy R package, adding support for additional Python libraries.

Contributions to this project can include actual code and functionality, or documentation, including definition of use cases for the developed software and contributions to tutorials, as well as participation in general discussions on approaches and concepts of better integration between R and Python software projects.

After the hackathon we will:

• Continue development of started projects. Contributions to SpectriPy are always highly welcome (independently of joining this hackathon).
• Consolidate tutorials and documentation and include and link them in the software package(s).
• Evaluate possibilities of a joint grant submission to get funding to continue integration of the various open-source MS software libraries.
• Submit the SpectriPy package to Bioconductor and eventually publish this R package as a research paper.

Technical Details

• Programming language: R and/or, Python.
• Existing MS software: matchms, spectrum_utils, pyOpenMS, Pyteomics, Spectra, SpectriPy
• Other relevant software that will be used: reticulate for integration of R and Python; basilisk for managing and bundling Conda environments for R packages.

Contact information

Johannes Rainer
Institute for Biomedicine, Eurac Research, Bolzano, Italy
[email protected]

[enryH]

Just a question of understanding: Is the focus rather on making Python libraries available in R?

[lgatto]

The idea is to be able to use/run methods available in Python directly on R/Bioconductor objects.

[jorainer]

exactly, use Python functionality/methods in R on R/Bioconductor MS objects to avoid re-implementation of established methods. we have already some Python libraries (i.e. matchms) that we can directly call, but it would be great to expand it to additional/other Python libraries.

[enryH]

thanks for the clarifications to both of you. The title reads a bit more general for me, so I just wanted to double check.

[jorainer]

What is needed for the hackathon/development:

• A recent version of R (>= 4.4) or one of the Bioconductor docker images.
• A set of R/Bioconductor packages that can be installed with:

install.packages("BiocManager")
BiocManager::install(c("Spectra", "msdata", "mzR"))
BiocManager::install(c("reticulate", "basilisk"))
BiocManager::install("RforMassSpectrometry/SpectriPy")

Create and synchronize a fork of the SpectriPy github repository.