Move minimap2 download to runtime (avoids build failures on HF cpu-basic)

Phase 3: Sequencing pipeline (FASTQ -> QC -> alignment -> variant calling -> report)

fix: load .env.deploy explicitly in config.py

feat: molecular docking (AutoDock Vina) + docs site + Sentry + cache stats + Phase 2 docs update

feat: complete Phase 2 — API keys, share links, export, pipeline steps, guest upgrade

fix: use valid EBI email (example.com), remove PhyML build from Docker

fix: remove PhyML source build from Docker (fails on HF), add graceful missing-binary error in phylo.py

feat: dedicated phylo router (NJ/UPGMA/ML), local PhyML binary, standalone phylo page, PhyloTreeViewer redesign

feat: in-memory pipeline v2 with MSA, phylo tree, domains; PipelineResults component; standalone phylo page; report page; wizard step checkboxes

fix: add logging to init_redis() so Redis failures are visible in server logs

fix: Foldseek alignments are [dict] lists, parse target format

fix: debug alignment structure types

fix: debug Foldseek result structure

fix: Foldseek alignments are dicts, parse with .get() correctly

fix: debug Foldseek alignment format (list vs dict)

fix: add full traceback to Foldseek error

fix: use aiohttp for Foldseek multipart upload

fix: debug Foldseek submit response type

fix: use sync httpx in thread pool for Foldseek submit

fix: restore nested alignments loop, remove debug raw

fix: return raw Foldseek data for debugging

fix: Foldseek returns list directly, no alignments sub-key

fix: move fetch inside httpx async with block

fix: manual multipart for Foldseek, add secrets import

fix: use sync httpx.Client for Foldseek multipart upload

fix: add error detail to Foldseek, use bytes, longer timeout

fix: simplify Foldseek integration, use full PDB, encode as bytes

fix: primer error handling, interactions scores, structure comparison API

fix: handle null name in InterPro domain response

feat: domains, interactions, primers, structure analysis, wizard, phylo tree, conservation

Fix KEGG search: use find/hsa with symbol matching + parse get/PATHWAY section

Add Reactome diagrams, KEGG search, pathway enrichment (pipeline + standalone)

Security, crash fixes, demo prep, dead code removal

Fix Reactome pathway search: correct param name and response parsing

Add Render Blueprint, GitHub Actions auto-deploy, revert to CORS fix, local startup scripts

Fix structure viewer: correct PDB API URL (+/entry/), fix RCSB UniProt search attribute + payload format

Fix MSA: use correct EBI result type 'fa' instead of 'aln-fasta' — confirmed via live API test

Fix MSA: dynamically discover EBI result types instead of hardcoding 'aln-fasta'

Fix MSA alignment: add Accept headers, result retry with delay, better error detail

Fix Phase 1 endpoints: resolve UniProt→PDB in structure viewer, use proper Reactome search, surface real backend errors in frontend

Sprint 3-7: Alignment, Structure Viewer, Pathway Analysis, Utility Tools, analyze page fully live

Sprint 1-2: JWT auth middleware + UniProt dedicated page

full dark migration: CSS overrides for all legacy classes, framer-motion animations across every page/component, rebrand to Bio Nexus

design.md + rules.md compliance: dark theme, teal palette, pulsing dot, Pydantic models, api.ts routing, config fixes, db/max_hits threading, Supabase migration 004

Fix PostgREST in() filter: quote status values in startup hook

Add RefSeq->UniProt mapping step in pipeline + demo uniprot_accession

Add startup resume hook for stuck jobs + persist NCBI RID

fix: extract user UUID from JWT for rate limiter, skip anon

fix: logger.exception instead of stderr, fail job on thread crash

fix: move first _set_step into try block + stderr fallback in run_pipeline_sync