Omics-based Predictive Platform (OPP)

The Omics-based Predictive Platform (OPP) is a solution for deploying ML/DL models trained on high-throughput omics data, with a primary focus on DNA methylation microarrays (450K, EPIC, EPICv2).

 

🖥️OPP ensures:

  • Standardization: All data undergo harmonized preprocessing (masking, normalization, imputation) using Nextflow-based pipelines.

  • Security: Data submission and interaction are secured via TLS protocol and containerized with Docker.

  • Scalability: Celery queue and Redis broker manage tasks efficiently, allowing multiple concurrent users and scalable inference.

  • Transparency: All processing steps, models, and results are version-controlled and documented.

 

📤Inference

To use the models implemented within OPP, a registered user must submit a sample through the submit subpage. Once the inference pipeline completes, the generated report is, by default, accessible only to its owner (private status).

However, if needed, the user can make the report publicly available (public status) by checking the Public checkbox on the update inference task subpage.

 

🔐Data security 

  • All communications with the platform are encrypted using TLS (Transport Layer Security).
  • User passwords are securely encrypted and stored in compliance with industry best practices.
  • The database is secured against unauthorized access to ensure the integrity and confidentiality of stored information.

 

👷‍♂️Task processing overview 

After a request is submitted, a new task is added to the queue with the status PENDING. When resources become available, a worker retrieves the task from the FIFO (First-In-First-Out) queue and begins processing. At this point, the task status changes to STARTED.

From here, there are two possible outcomes:

  • If the worker completes the task successfully, the status will change to SUCCESS, and the inference report will become available.
  • If an error occurs during processing, the task status will change to FAILED.

If your task has failed, please contact us and provide the TASK ID, which is visible in the top-left corner of the task record in the history subpage.

 

💾Exemplary data

  1. GSM2940967 - Glioblastoma, RTK I - Grn IDAT, Red IDAT
  2. GSM8531341 - Prostate tumor - Grn IDAT, Red IDAT
  3. GSM3519721 - Pulmonary metastasis of head and neck squamous cell carcinoma - Grn IDAT, Red IDAT

 

🤝Collaboration

We welcome contributions and collaboration from the scientific community for model integration and deployment within OPP.

 

✉️Contact

To report bug or issue, please contact: jan.binkowski@pum.edu.pl