Tutorial Request: Custom Model Integration with gReLU

[yuanlizhanshi]

gReLU is a powerful framework that unifies preprocessing, training, interpretation, variant effect prediction and sequence design for diverse DNA sequence models.

However, current documentation does not include a tutorial that shows how a user-defined PyTorch model can:

1. Be registered into gReLU’s training pipeline
2. Expose outputs compatible with prediction transform layers
3. Work with interpretation methods (saliency / ISM / TF-MoDISco)
4. Support downstream tasks such as variant effect prediction and design