Week 24, 2025 (June 10 - June 16)
Status: Actively working on deploying the StickForStats platform to production. Made significant progress with frontend on Vercel and backend on HPC infrastructure, currently resolving integration issues.
| Component | Progress | Details |
|---|---|---|
| Frontend Deployment |
Deployed React frontend to Vercel after resolving critical memory issues during build process. Implemented memory optimization strategies including increased Node.js heap size allocation. Fixed critical theme configuration issues causing runtime errors with Material-UI components. Working on: Resolving API connectivity issues between frontend and backend.
React 18
Material-UI
Vercel
Node.js
|
|
| Backend Deployment |
Deployed Django backend on IGIB HPC infrastructure. Configured CORS settings for cross-origin requests from Vercel frontend. Set up server running with session management. Challenges: Public tunnel stability issues, working on permanent solution.
Django 4.2
PostgreSQL
HPC Linux
Gunicorn
|
|
| Integration & Testing |
Established connection between frontend and backend using SSH tunneling. Testing public tunnel configuration for HPC backend accessibility. In Progress: Verifying module functionality in production environment. Created deployment documentation and automation scripts.
Deployment URLs (Testing):
|
Current Challenges & Progress:
Collaborator: Mahima (DMD splicing analysis project)
Duration: 2 days (Friday-Saturday, June 14-15)
| Component | Progress | Details |
|---|---|---|
| Pipeline Development |
Developed comprehensive branchpoint prediction pipeline for Duchenne Muscular Dystrophy (DMD) gene analysis. Integrated LaBranchoR deep learning model for accurate branchpoint prediction. Created automated scripts for processing multiple splice variants. Implemented downstream analysis tools for functional interpretation.
Python
LaBranchoR
BioPython
Pandas
|
|
| Analysis Tools |
Created visualization tools for branchpoint location analysis relative to exon boundaries. Developed verification scripts to validate predicted branchpoints against known databases. Implemented interactive reporting system with network visualization capabilities. Added batch processing support for analyzing multiple DMD mutations simultaneously. |
|
| Documentation |
Created comprehensive documentation for pipeline usage and interpretation. Developed improvement plan for future enhancements including additional ML models. Prepared detailed analysis reports for DMD splice variants. |
Status: All modules successfully deployed and verified in production
| Module | Status | Features Verified |
|---|---|---|
| Confidence Intervals | ✓ Operational | Mean, proportion, variance calculations with multiple confidence levels |
| Design of Experiments (DOE) | ✓ Operational | Factorial designs, response surface methodology, optimization |
| Principal Component Analysis | ✓ Operational | Dimensionality reduction, scree plots, biplot visualization |
| Statistical Quality Control | ✓ Operational | Control charts, capability analysis, measurement system analysis |
| Probability Distributions | ✓ Operational | Parameter estimation, distribution fitting, visualization |
| Advanced Statistics | ✓ Operational | Time series, survival analysis, Bayesian methods |
| Machine Learning | ✓ Operational | AutoML, model registry, deep learning integration |
| RAG System with LLM | ✓ Operational | Context-aware assistance, knowledge retrieval, intelligent suggestions |
This week focused on deploying the StickForStats platform to production. The deployment process involved significant technical challenges including memory constraints during frontend builds, module path issues on HPC infrastructure, and theme configuration problems. While substantial progress was made with frontend deployment on Vercel and backend setup on HPC, integration between the two components remains in progress.
The deployment architecture leverages modern cloud infrastructure with the frontend hosted on Vercel's global CDN and the backend running on IGIB's HPC infrastructure. This hybrid approach provides optimal performance for users while maintaining computational resources for intensive statistical analyses. The implementation of secure tunneling ensures reliable connectivity between components.
In parallel, I completed a focused 2-day project developing a branchpoint prediction pipeline for DMD gene analysis. This bioinformatics tool integrates state-of-the-art deep learning models for splice site prediction, addressing a critical need in understanding DMD mutations. The pipeline includes comprehensive analysis tools, visualization capabilities, and batch processing support.
Key technical achievements this week:
The StickForStats platform deployment is actively progressing with the frontend successfully deployed to Vercel and backend running on HPC infrastructure. Current focus is on resolving integration challenges to ensure stable connectivity between components. Once complete, the platform will serve as a comprehensive statistical analysis suite with advanced AI capabilities for researchers and data scientists. Immediate next steps include establishing stable frontend-backend communication, implementing a permanent public access solution for the HPC backend, and conducting comprehensive integration testing of all modules.