7 Subtype prediction
To expand the power of signatures to clinical application, based on signature activity, we can go further build neutral network model prediction model with keras. This feature is implemented in sigminer’s child package sigminer.prediction.
If you are studying copy number signatures of prostate cancer, you can directly use the trained model, the usage has been described in the README file.
For other situations, you can reuse the functions modeling_and_fitting()
and batch_modeling_and_fitting()
(for batch processing) for building models.