nujhatmaliha99/medirepolab

1. High-level shape

MedReproLab is a two-tier MERN application: a React/Vite frontend talks to an Express.js REST backend, which persists research workspace data in MongoDB via Mongoose. The repository also includes a separate lightweight admin-panel UI under the backend tree.

2. Frontend shell and navigation

The main UI lives in src/ and is organized around a dashboard-style shell. App.jsx, DashboardLayout.jsx, and the page components under src/pages/ define the user-facing workflows for projects, datasets, experiments, reviews, reports, and published research. Styling is split across App.css and index.css.

3. Client-side state and access control

Authentication is handled on the client by AuthContext.jsx, with protected navigation enforced by PrivateRoute.jsx. The login/register pages are the public entry points; once authenticated, users move into the dashboard and workspace pages. src/api/axios.js centralizes API calls, which helps keep request/auth wiring consistent across pages.

4. Core product boundaries in the UI

The frontend is organized around research-workspace concepts rather than generic CRUD pages:

- project management: Projects, ProjectDetails

- datasets: Datasets

- experiments and comparisons: Experiments, ExperimentDetails, Compare, Leaderboard

- research outputs: Manuscript, ModelCards, Reports, ReproPackage, Reproducibility

- review and quality workflows: QualityReview, ReviewRequests, XAIEvidence, ClinicalTriage, ClinicalCases

- public discovery: PublishedResearch

- user/account views: Profile, Dashboard

5. Frontend data flows

The typical flow is: page component -> axios API request -> authenticated backend route -> controller -> Mongoose model -> MongoDB -> response back to the page. The utility modules in src/utils/ suggest client-side shaping of domain-specific outputs such as activity summaries, artifact lists, dataset risk scoring, and reproducibility checks before rendering.

6. Backend API organization

The Express backend under server/server/ is split by domain:

- routes/ defines REST endpoints

- controllers/ contains request handlers

- models/ defines Mongoose schemas

- middleware/authMiddleware.js protects authenticated routes

- config/db.js handles MongoDB connection

This is a clean controller-route-model separation, with auth middleware sitting between the client and protected resources.

7. Backend domain model

The persistence layer centers on User, Project, Dataset, Experiment, ClinicalCase, ReviewRequest, and Activity. That tells you the system stores not just research objects, but also workflow state: review submissions, audit history, and clinical triage/case metadata.

8. Review, approval, and publication workflow

A major architectural boundary is between private research workspaces and approved public research. reviewRequestController.js and the admin-panel UI indicate a submission/approval queue. Once approved, research appears in PublishedResearch, so there is an explicit transition from internal workspace data to public-facing visibility.

9. AI-assisted review layer

aiController.js and aiRoutes.js indicate a dedicated API for automated reviewer suggestions. In the product description this is used for identifying missing validation, artifacts, and documentation gaps, so the AI subsystem functions as advisory tooling layered on top of the core research records rather than as a primary data store.

10. Audit/activity tracking

activityLogger.js, activityController.js, and the Activity model point to an event stream or activity log used across the platform. This is important for a reproducibility product because it captures what changed, when, and by whom across project, experiment, and review actions.

11. Admin surface

server/server/admin-panel/ is a separate static interface with its own app.js, index.html, and styles.css. Based on the README, it serves the admin review queue on a different local port from the main app, so the repository contains two user interfaces: the researcher app and the admin approval console.

12. Infrastructure and runtime

The stack is Node.js/Express on the server, React on the client, and MongoDB as the backing store. JWT authentication is the main security boundary. The README also exposes health/status endpoints for both app and database, which suggests operational visibility is part of the backend’s responsibility.

13. Demo and data bootstrap

The seed scripts (seedSample.js, seedBreastSample.js, seedPublishedResearch.js) create representative datasets and approved public research for demos. These scripts are part of the architecture because they establish the initial state needed for the platform’s main workflows: workspace exploration, review, and public browsing.

14. How to draw the diagram

The most accurate diagram should show:

- a browser/client tier with the React app

- a separate admin-panel UI

- a REST API tier with auth, AI, activity, and domain controllers

- a MongoDB persistence layer

- workflow arrows for authentication, workspace CRUD, experiment/review activity, AI suggestions, and approval/publication

The key architectural idea is that MedReproLab is a research-workflow system with strong separation between private workspace data, review/approval processes, and publicly published research.