nclgbd/geneGPT

geneGPT

Data required:

• AML_datasets.RData - initial dataset as provided by paper
• gene.RData - created when running the genesignature.ipynb notebook
• data.RData - created when running the splitdata.ipynb notebook

How to run

1. Ensure AML_datasets.RData is in project directory.

The datasets created are too big to be committed to the repository, and thus the feature selection and data split is left to be created by the user.

2. Run genesignature.ipynb notebook to generate gene.RData. This creates the list of gene signatures to train the models using the feature selection method proposed in the paper
3. Run splitdata.ipynb notebook to generate data.RData. This creates the train/test splits for the models to use for training.
4. Run models.ipynb. This trains the models first using the gene signatures selected via the proposed method in the paper, then by using the alternative gene signatures as proposed by group Genome Seeker, and finally cross validation is performed using the first set of gene signatures