Hi HN,

I've been working on QUBE Predict, a cloud platform for drug response prediction built around graph-based biological descriptor learning.

Rather than relying solely on end-to-end models, the framework focuses on reusable biological descriptor representations that can be paired with conventional machine learning algorithms.

The platform includes:

• REST API

• Web interface

• Batch prediction

• Authentication and API keys

• Cloud deployment

The methodology has been described in a publicly available technical paper.

Paper:

https://doi.org/10.5281/zenodo.21455765

Live Demo:

https://qube-predict.streamlit.app/

GitHub:

https://github.com/gulfamh1999-afk/qube-predict-ui

I'd really appreciate technical feedback, especially on the representation-learning methodology and platform architecture.