A multimodal approach to antimicrobial resistance prediction for drug recommendation and large-scale classification.
This repository contains the code used to conduct experiments for the paper A multimodal approach to antimicrobial resistance prediction for drug recommendation and large-scale classification. from Giovanni Visonà, Yiran Li, Diane Duroux, Lucas Miranda, Emese Sukei, Karsten Borgwardt, and Carlos Oliver.
The experiments are organized based on the authors that performed them, while the general utilities and processed data are contained in the data_split and processed_data forlders.
lm_experiments contains scripts to evaluate the performance of some baseline machine learning models on individual drug-species subsets of the DRIAMS dataset.
dd_experiments includes the training of the Siamese Network model and the evaluation of all the recommendation models except the ResMLP.
gv_experiments contains scripts to train the baseline PCA+LR experiments and the scripts related to the ResMLP experiments, including classification, recommendation, and ablation experiments.
To train a ResMLP classifier model with the same configuration as the one presented in the paper, the command used is:
cd gv_experiments
python3 training_scripts/train_ResAMR_classifier.py --experiment_name "myExperiment" --experiment_group "ResMLP" --driams_dataset "B" --seed 0 --split_type "random" --driams_long_table "../processed_data/DRIAMS_combined_long_table.csv" --drugs_df "../processed_data/drug_fingerprints.csv" --spectra_matrix "../data/DRIAMS-B/spectra_binned_6000_2018.npy" --n_epochs 500 --learning_rate 0.0003 --drug_emb_type "fingerprint" --fingerprint_class "morgan_1024" --fingerprint_size 1024 --patience 50 --batch_size 128
The necessary files to run the script (the matrices of the processed spectra) are available on request.