The main outputs of the code are written into plots directories. Due to data use restrictions within Genomics England, certains files in this directory will not run outside of the system. For more details on the contents of this repository:
| Folder/File | Description |
|---|---|
| code/01_move_gihawi_preds_to_csv.R | R script that loads the Gihawi et al. data and their trained models, and write the trained models' predictions on the data to .csv files which can be loaded into python. The inputs for this file are available within Genomics England. |
| code/02_generate_figures.py | Python script which loads the Gihawi et al. model predictions, and runs variations of analysis methods to evaluate the strength of signal in these predictions, and writes all resulting plots into the plots folder, which encompasses the plots from all figures in our associated manuscript |
| code/helper_functions.py | python file containing functions used in the code/02_generate_figures.py file. |
| data/ | Is empty within the repository, but the 01_move_gihawi_preds_to_csv.R script runs when the contents of this folder are populated with data used by Gihawi et al.. The contents of this folder are available within Genomics England |
| models_out/ | Is empty within the repository, but the 01_move_gihawi_preds_to_csv.R script runs when the contents of this folder are populated with the trained models used by Gihawi et al.. The contents of this folder are available within Genomics England |
| gihawi_model_predictions/ | contains .csv files of model prediction scores applied to the samples from Genomics England; contents of this folder are written by the 01_move_gihawi_preds_to_csv.R script. |
| Rpackages/ | folder containg R packages used by Gihawi et al.'s code. This folder is empty within github, but is populated within the Genomics England system. |
| plots/ | Contains .pdf files of this repository's analysis results. All files within this fodler are produced by the 02_generate_figures.py script. |
For any questions, please reach out to gia2105@columbia.edu, or feel free to use this repository's issues page.