arteMIS (Accelerated Ranking and Tuning using multi-metric Interpretability across Scores) is a framework that builds optimal molecular networks for downstream analysis. It can start from:
- global-mode — in-silico annotated metabolic features (MS2Query-based),
- seed-mode — a subset of well-annotated metabolic features, or
- target-mode — network construction optimised for a specific class of compounds.
git clone https://github.com/rtlortega/arteMIS.gitWith conda:
conda env create -f environment.yml
conda activate artemisCheck that everything is ok:
artemis-run --helpWithout conda:
pip install -e .Check config.yaml and update the input/output paths before running. Default mode is global:
artemis-run --config config.yaml --mode globalmgf_file: input spectra filechem_file: MS2Query results for global mode (a column such asnpc_pathway_resultsset viachem_levelinconfig.yaml)out_dir: name of your output directory — all results are saved as a folder here
mgf_file: input spectra filechem_file: table of the subset of annotated features (a column such aspathway_levelset viachem_level— this is what the chemical optimisation focuses on)out_dir: name of your output directory — all results are saved as a folder here
mgf_file: input spectra filechem_file: any table (MS2Query or otherwise) with the full features, with a column identifying the class to target and optimise for (seeconfig.yaml)out_dir: name of your output directory — all results are saved as a folder here
- Cleaned spectra
- LHS (Latin Hypercube Sampling) of the variable space exploration
- The top 3 graphs obtained with arteMIS
- The results from all networks, in a JSON file
top_configs.csv— the top 10 configurations, with their parameters and metrics
The code in this package is licensed under the MIT License.
Coming soon.
Please open a GitHub Issue for bugs/feature requests.
Maintainers: Rosina Torres-Ortega (rosina.torresortega@wur.nl) and Esteban Charria-Girón (esteban.charriagiron@wur.nl)