AIRI: automatic interesting rules identification

This is the main branch for general AIRI development. For AIRItaxa (currently under peer-review), please see the AIRItaxa branch: https://github.com/pascoalf/AIRIexperiment/tree/AIRItaxa

Source data files description (data/)

• mosj_ASV_df.rds - ASV table of MOSJ dataset

• mosj_env_data.rds - metadata for ASV table of MOSJ dataset

• mosj_environmental_data_category.rds - discretized metadata values for MOSJ dataset

• emose_df - ASV table with metadata from EMOSE dataset

• emose_curated_taxonomy - taxonomic data on ASVs from EMOSE dataset

Rule sets obtained (rule-sets/)

Case study 1:

• mosj_full_rules_set_as_data_frame.rds - Association rule mining of MOSJ results as data.frame object
• mosj_full_rules_set.rds - Association rule mining os MOSJ results as transactions object
• mosj_airi_by_complexity.csv - AIRI results by complexity for MOSJ data
• mosj_airi_by_improvement.csv - AIRI results by improvement for MOSJ data
• mosj_airi_by_mutual_information.csv - AIRI results by mutual information for MOSJ data

Case study 2:

• emose_rules_df.rds - Association rule mining of EMOSE results as data.frame object
• emose_rules.rds - Association rule mining of EMOSE results as transactions object
• emose_airi_by_complexity.csv - AIRI results by complexity for EMOSE data
• emose_airi_by_improvement.csv - AIRI results by improvement for EMOSE data
• emose_airi_by_mutual_information.csv - AIRI results by mutual information for EMOSE data

Instructions to reproduce results (R/)

Assuming a dedicated environment with all the files in R/ and data/ directories inside:

1. prepare_data.R - loades R packages needed.

2. Case study 1 - MOSJ dataset

2.1. cs1_prepare_data.R - loads MOSJ dataset and pre-processing

2.2. cs1_prepare_metadata.R - loads contextual environmental data from MOSJ and preprocessing

2.3. cs1_arm.R - asociation rule mining on MOSJ dataset

2.4. cs1_airi_steps.R - AIRI implementation on MOSJ dataset

3. Case study 2 - EMOSE dataset

3.1. cs2_prepare_emose_data.R - load EMOSE dataset and metadata, preprocessing

3.2. cs2_arm_emose.R - association rule mining for EMOSE dataset

3.3. cs2_airi_emose.R - AIRI for EMOSE dataset

4. Comparison: AIRI vs alternatives

4.1. cs1_compare.R - AIRI vs alternatives for MOSJ dataset

4.2. cs2_compare.R - AIRI vs alternatives for EMOSE dataset

4.3. cs1_syst_compare.R - systematic comparison for all metric values - MOSJ dataset

4.4. cs2_syst_compare.R - systematic comparison for all metric values - EMOSE dataset

Citation

Source code citation

See AIRItaxa branch: https://github.com/pascoalf/AIRIexperiment/tree/AIRItaxa

Additional citation for source data

If you use the source data available at this repository, please cite the original papers describing and presenting the datasets:

• MOSJ: Pascoal, F., Branco, P., Torgo, L. et al. Definition of the microbial rare biosphere through unsupervised machine learning. Commun Biol 8, 544 (2025). https://doi.org/10.1038/s42003-025-07912-4
• EMOSE: Pascoal, F., Tomasino, M. P., Piredda, R., Quero, G. M., Torgo, L., Poulain, J., Galand, P. E., Fuhrman, J. A., Mitchell, A., Tinta, T., Turk Dermastia, T., Fernandez-Guerra, A., Vezzi, A., Logares, R., Malfatti, F., Endo, H., Dąbrowska, A. M., De Pascale, F., Sánchez, P., Henry, N., Fosso, B., Wilson, B., Toshchakov, S., Ferrant, G. K., Grigorov, I., Vieira, F. R. J., Costa, R., Pesant, S., Magalhães, C. (2023). Inter-comparison of marine microbiome sampling protocols. ISME Communications, 3(1), 84. https://doi.org/10.1038/s43705-023-00278-w