medical-informatics-platform/platform-backend

Purpose and Runtime

This repository is the Java backend for the MIP platform. It is a Spring-based HTTP service packaged as a Docker image, configured for PostgreSQL persistence, Keycloak authentication, and an external Exaflow server. Application bootstrap is src/main/java/hbp/mip/MIPApplication.java; runtime defaults and environment-variable binding are in src/main/resources/application.yml and config/application.tmpl.

API Layer

REST-facing components are organized by domain:

• Algorithms: src/main/java/hbp/mip/algorithm/AlgorithmsAPI.java

• Data models: src/main/java/hbp/mip/datamodel/DataModelAPI.java

• Experiments: src/main/java/hbp/mip/experiment/ExperimentAPI.java

• Active user context: src/main/java/hbp/mip/user/ActiveUserAPI.java

These APIs receive requests, use DTOs as boundary contracts, and delegate business work to services rather than directly managing persistence or remote calls.

Experiment Orchestration

Experiments are the central persisted execution domain. src/main/java/hbp/mip/experiment/ExperimentService.java coordinates experiment creation/retrieval and execution-oriented behavior, while src/main/java/hbp/mip/experiment/ExperimentDAO.java and src/main/java/hbp/mip/experiment/ExperimentRepository.java separate data access from service logic. Execution payloads are represented by src/main/java/hbp/mip/experiment/ExperimentExecutionDTO.java.

The principal flow is API request → experiment service → repository/DAO and, where execution is required, the configured Exaflow endpoint.

Algorithm and Data-Model Catalogs

Algorithm requests/specifications are handled through src/main/java/hbp/mip/algorithm/AlgorithmService.java, src/main/java/hbp/mip/algorithm/AlgorithmsSpecs.java, and DTOs in the same package. Data-model access is similarly isolated in src/main/java/hbp/mip/datamodel/DataModelService.java and src/main/java/hbp/mip/datamodel/DataModelDTO.java.

These components form the input-definition boundary for experiments: algorithm specifications and data-model metadata are resolved before or alongside experiment execution.

Identity and Authorization Boundary

Security configuration and authentication integration reside in src/main/java/hbp/mip/configurations/SecurityConfiguration.java, src/main/java/hbp/mip/configurations/AuthenticationEventListener.java, and src/main/java/hbp/mip/configurations/SpaRedirectAuthenticationSuccessHandler.java. Keycloak is the external identity provider; its URL, realm, client ID, and secret are supplied at deployment time.

User identity is converted into platform-level user state through src/main/java/hbp/mip/user/ActiveUserService.java, src/main/java/hbp/mip/user/UserRepository.java, and claim helpers in src/main/java/hbp/mip/utils/ClaimUtils.java. Authentication can be disabled for development.

Persistence and Schema

PostgreSQL is required in both development and container deployments. Persistence wiring is centralized in src/main/java/hbp/mip/configurations/PersistenceConfiguration.java; initial schema creation is versioned in src/main/resources/db/migration/V1__InitialSchema.sql. Repository/DAO pairs in the experiment and user domains provide the persistence boundary.

Operational Boundaries

The service is containerized by Dockerfile. Logging is configured through src/main/resources/log4j2.yml; log levels are environment-controlled. API documentation configuration is located at src/main/java/hbp/mip/configurations/OpenApiConfig.java. Cross-cutting request failures are normalized by src/main/java/hbp/mip/utils/ControllerExceptionHandler.java.

CI includes image publishing and security scanning workflows under .github/workflows/publish_images.yml, .github/workflows/sast.yml, and .github/workflows/container-scan.yml.