Back

ZAFONIX: A GUI Driven Platform for Personalized Therapeutics

Attique, Z.; Azhar, H. M. F.

2025-08-11 health informatics
10.1101/2025.08.07.25333242 medRxiv
Show abstract

The integration of genomic insights into clinical decision-making is central to personalized medicine, yet clinicians often face challenges navigating large pharmacologic datasets and identifying relevant drug-gene associations. We developed ZAFONIX, a desktop graphical tool tailored to streamline this process using a curated DrugBank-derived dataset comprising [~]17,000 entries. The application, built in Python with tkinter for the user interface, pandas for data management, and matplotlib for visualizations, introduces a secure, key-protected entry followed by a unified search panel. Users can query by disease, gene ID, or drug name, and apply filters for FDA-approved, experimental, or investigational drug statuses. The system returns a scrollable table of results and automatically generates visual summaries: drug status distribution, small-molecule prevalence, and top target actions. Users may export the search results as CSV or Excel files. In performance trials, ZAFONIX delivers response times under one second for standard queries on typical hardware. Visually, the interactive charts facilitate quick interpretation of pharmacologic landscapes surrounding specified genetic or disease contexts. The modular design supports future integration with live DrugBank APIs, incorporation of drug-drug interaction data, and extension toward clinical variant interpretation pipelines. ZAFONIX stands as a clinician-friendly, extensible application, bridging the gap between static pharmacogenomic data and real-time therapeutic decision support (https://github.com/ZarlishAttique/ZaFoniX).

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.