v1.0.1: Hardened Knowledge Graph, Concurrency & Layer Visibility

What's Changed in v1.0.1

Critical Fixes & Production Hardening

• CLI & Docker serve sub-command: Fixed command mismatch between Dockerfile (CMD ["serve"]) and CLI parser (added alias server).
• GO Layer Visibility: Fixed silent failure in KnowledgeGraphBuilder where annotate_gene was called incorrectly; added layer_status tracking and warnings propagation.
• REST API Concurrency: Resolved event loop blocking by switching to ThreadingHTTPServer with a dedicated asyncio worker thread and 120s timeout.
• Scientific Integrity & Scoring: Removed artificial 1.0 confidence defaults; curated and literature edges are now unscored (weight=None), preventing fabricated probabilities.
• Traversal Accuracy: Added whole-word target phrase matching, isolated PUBLICATION nodes from non-literature queries, and added --target-type filter.
• Test Suite: Added 35 offline robustness tests (tests/test_offline_robustness.py) and CI rerun logic for resilient automated builds.

Full Changelog: v1.0.0...v1.0.1

v1.0.0: General Availability - Python SDK & REST API v1

What's Changed in v1.0.0

Unified Python SDK & REST API v1 (PRD-18)

• Shipped unified high-level Python client BioContext in src/biocontext/__init__.py.
• Built standalone REST API v1 server (biocontext api --port 8000) supporting /health, /v1/gene, /v1/protein, /v1/disease, /v1/graph, and /v1/traverse.
• Added developer documentation in docs/API.md and complete integration test suite in tests/test_sdk_and_api.py.
• Promoted BioContext to v1.0.0 General Availability.

Full Changelog: v0.9.0...v1.0.0

v0.9.0: Ontological Multi-hop Traversal & Reasoning Engine

What's Changed in v0.9.0

Ontological Multi-Hop Traversal Engine (PRD-22)

• Implemented deterministic graph traversal (TraversalEngine) linking biological queries across genes, proteins, pathways, and diseases.
• Explores connected paths up to configurable max_hops with cumulative edge scoring.
• Added natural language template explanation generating mechanistic rationale for every discovered path.
• Exposed new MCP tool traverse_biological_graph and CLI command biocontext traverse.
• Integrated full end-to-end traversal test cases in tests/test_graph.py.

Full Changelog: v0.8.0...v0.9.0

v0.8.0 - Biological Knowledge Graph Architecture

What's Changed

• Knowledge Graph Core (PRD-21): Implemented ontological biological Knowledge Graph models, in-memory store, and subgraph builder pipeline.
• Multi-layer Entity Connections: Seamlessly traverses and builds subgraphs across Genes (HGNC), Proteins (UniProt), Transcripts (Ensembl), Pathways (Reactome), GO terms (QuickGO), Diseases (Open Targets), and Publications (Europe PMC).
• FastMCP Tool: Added build_entity_subgraph tool to server.py for LLM agents.
• CLI Command: Added biocontext graph <query> to CLI with selective layer inclusion flags.
• Testing: Added unit and integration tests with 100% passing rate.

v0.7.0: Literature Intelligence (Europe PMC & PubMed)

BioContext v0.7.0: Literature Intelligence

What's New

• Europe PMC & PubMed Literature Adapter:
  • Full querying capabilities for biomedical publications via Europe PMC REST API.
  • Resolves queries by gene symbol, disease entity, PubMed ID (PMID), PubMed Central ID (PMCID), or Digital Object Identifier (DOI).
  • Rich citation metadata extraction including titles, authors, journals, publication years, full abstract text, citation count, and direct open-access URLs.
• FastMCP Literature Tools:
  • get_supporting_publications(query: str, limit: int = 5): Search biomedical literature for gene/disease evidence.
  • get_publication_details(identifier: str): Fetch granular paper metadata and abstract by PMID, PMCID, or DOI.
• CLI Literature Commands:
  • biocontext literature <query>: Interactive multi-paper literature search with formatted terminal tables.
  • biocontext paper <identifier>: Granular abstract and citation viewer directly in the terminal.
• Resilience & Testing:
  • Full test coverage for literature adapter queries and MCP tool interfaces (tests/test_literature_adapter.py).

v0.6.0: MONDO Disease Ontology & Open Targets Grounding

BioContext v0.6.0 Release Notes

🌟 New Features & Grounding Capabilities

• MONDO Disease Ontology Integration (MondoAdapter):
  • Added authoritative clinical disease normalization via EBI OLS4 API.
  • Resolves canonical MONDO:xxxxxxx identifiers, preferred names, descriptions, synonyms, and cross-references (OMIM, Orphanet, DOID, UMLS).
  • Flexible free-text disease search.
• Open Targets Platform Integration (OpenTargetsAdapter):
  • Direct target-disease evidence mapping via Open Targets GraphQL API (api.platform.opentargets.org/api/v4/graphql).
  • Yields aggregate association confidence scores (0.0 - 1.0) with granular evidence breakdowns across genetic associations, somatic mutations, and literature.
• Expanded MCP Server (10 Production Tools):
  • Exposes resolve_disease(query, limit) and get_target_diseases(gene, limit).
• New CLI Commands:
  • biocontext disease <query>
  • biocontext targets <gene>

v0.5.1: Ensembl REST Resilience & Multi-Tier Fallback

BioContext v0.5.1 Release Notes

🛠️ Bug Fixes & Resilience Improvements

• Ensembl Adapter Degradation Resolution (#4):
  • Implemented resilient 4-stage exponential backoff and token-bucket rate limiting against Ensembl REST HTTP 429 and HTTP 500/503 service degradations.
  • Added intraservice endpoint fallback: automatic degradation from /homology/id to /homology/symbol and from expand=1 to expand=0 with symbol model enrichment.
  • Re-enabled and verified all 5 previously skipped Ensembl unit and MCP tests.
• Documentation & Asset Modernization:
  • Streamlined README.md with official high-resolution logo hero banner.
  • Added verified AI client badges for Cursor, Claude Desktop/Code, OpenAI Codex, Antigravity IDE, and Goose.
  • Moved system architecture details to native docs/ARCHITECTURE.md.

v0.5.0 — Phase 1 Milestone: Multi-Authority Resolution & 100 Benchmark Cases

BioContext v0.5.0 — Phase 1 Milestone Release

Authoritative Biological Entity Resolution & Contextual Intelligence Framework.

Key Highlights

• Multi-Authority Biological Resolution: 7 operational adapters (HGNC, NCBI Entrez, UniProtKB, Ensembl, MGI, QuickGO, Reactome).
• 100 Benchmark Test Cases: Validated with 100% resolution accuracy across human cancer/signaling genes, clinical aliases (HER2, p53, Bcl-xL, p27Kip1), Entrez Gene IDs, UniProt accessions, and mouse models (Mus musculus).
• End-to-End Multilayer Context: Deep biological context pipeline verified in tests/test_e2e_pipeline.py (Entity Resolution $\rightarrow$ QuickGO Functional Annotations $\rightarrow$ Reactome Pathways).
• FastMCP Protocol Integration: 8 production-ready MCP tools compatible with Cursor, Antigravity IDE, Claude Desktop, Claude Code, and Goose.
• Embedded Persistence: Zero-configuration SQLite cache with alias lookup optimizations.
• Comprehensive Documentation: Complete guides in README.md, docs/QUICKSTART.md, and docs/ARCHITECTURE.md.
• Roadmap Evolution: Formally established Phase 1.5 (Disease, Drug & Clinical Context) bridging Phase 1 to Phase 2 (Knowledge Graph & AI Reasoning).

v0.1.0 - Phase 0 MVP: Biological Entity Resolution & MCP Server

BioContext v0.1.0 (Phase 0 MVP)

First official release of BioContext, establishing authoritative biological entity resolution and native Model Context Protocol (MCP) tool-calling support.

Highlights:

• Authoritative Adapters: HGNC (human nomenclature), NCBI Entrez (cross-species & gene IDs), and UniProtKB (protein accessions).
• Explainable Disambiguation: Deterministic rule-based matching with ResolutionContext (chromosome and biotype filtering).
• Unified Standalone CLI: Native biocontext command for single queries, batch processing, cache management, and benchmarks.
• Embedded Persistence: SQLite key-value cache with TTL and dynamic client-side token-bucket rate limiter.
• Model Context Protocol (MCP): Native stdio server compliant with MCP 2.x for Claude Desktop, Cursor, and Antigravity IDE.
• Empirical Validation: 100% pass rate across 25-case benchmark evaluation suite (exceeding >= 95% target KPI).