This folder contains the notebooks and results presented in the paper
- results: Results for different BO runs for each protein dataset
- Different_featurizations.ipynb: Notebook for the other two encoding apporach to obtain results in Table 4
- Example_study_GB1_2016.ipynb: notebook to generate results for GB1 (4) dataset from Wu et al (2016) in Table3, Figure 4, Figure 5B and Figure 6
- ODBO_for_different_datasets.ipynb: notebook to generate results for other three datasets
- Plot_results.ipynb: Notebook to systematiclly plot results and compute statstics
- Update_XGBOD_GB1_2016.ipynb: Notebook that performs XBGOD every BO iteration instead of before BO procedure in Figure S1
- XGBOD_accuracy_comparison_GB1_2016.ipynb: Notebook to analyze the XGBOD accuracy with different thresholds (Table 2)
The following .npy files store the encoding information for
- Physicochemical encoding: sequences in example_protein_georgiev_IndexToCombo.npy and feature vectors in example_protein_georgiev_Normalized.npy
- One-hot encoding: sequences in example_protein_onehot_IndexToCombo.npy and feature vectors in example_protein_onehot_Normalized.npy
The following .npy files store the initial selected experiment information
- GB1 (4)
- 40 initial samples selected by Algorithm S2, sequence in sele_experiment_GB1_2016.npy, fitness in sele_fitness_GB1_2016.npy
- 40 randomly selected samples, sequence locations in sele_indices_GB1_2016_random1.npy, sele_indices_GB1_2016_random2.npy, sele_indices_GB1_2016_random3.npy, sele_indices_GB1_2016_random4.npy, sele_indices_GB1_2016_random5.npy; shuffle orders in shuffle_order_GB1_2016_random1.npy, shuffle_order_GB1_2016_random2.npy, shuffle_order_GB1_2016_random3.npy, shuffle_order_GB1_2016_random4.npy, shuffle_order_GB1_2016_random5.npy
- GB1 (55)
- 40 initial samples selected by Algorithm S3, sequence locations in sele_indices_GB1_2014.npy; shuffle orders in shuffle_order_GB1_2014.npy
- Ube4b
- 40 initial samples selected by Algorithm S3, sequence locations in sele_indices_Ube4b_2013.npy; shuffle orders in shuffle_order_Ube4b_2013.npy
- avGFP
- 40 initial samples selected by Algorithm S3, sequence locations in sele_indices_avGFP_2016.npy; shuffle orders in shuffle_order_avGFP_2016.npy