This repository maintains the Python program of a pipeline for training, optimizing, and testing a random forest-based machine learning model in order to identify the essentiality of 6,638 Candida albicans genes.
To run this program, please make sure your environment meets the prerequisites and has the input data "Calbicans_13Features_6638genes_beforeImputation_210302.tsv" in the same directory as the code. The output files (i.e. the predictions and figures) will all be located in the same directory as well.
You can utilize the functions provided in this program to generate figures such as PR/ROC curves and distribution plots based on the source data attached to the paper. We provide the optimal random forest classifier trained from 5-fold cross-validation (rf_clf_optimal.joblib) so that you can load it directly, fit it with the whole GRACE gene set, and make predictions on all genes. Note that every time you fit this model, it is a new training process and the results may vary slightly in terms of prediction results. We chose one version of reasonable output and reported it in the paper.
-
Packages essential to the random forest pipeline and the versions that work:
- sklearn (0.23.2)
- numpy (1.16.2)
- pandas (1.0.2)
- joblib (0.14.1)
Packages essential to generating relevant figrues and the versions that work:
- matplotlib (3.0.3)
- sns (0.10.0)
-
Clone the repo
git clone https://github.com/csbio/C.albicans-ml-pipeline.git
-
Run the Python program (this program was built under Python 3.6.7 and should be able to run by a Python3 command)
Python3 rf_pipeline.py
Distributed under the MIT License. See LICENSE for more information.
Please email to Xiang Zhang (zhan6668@umn.edu) if you have any questions, comments, or suggestions regarding this program.
Project Link: https://github.com/csbio/C.albicans-ml-pipeline