Skip to content
mcp/harbor

DeepDNA changelog

ai.deepdna/knowledge

11 changes recorded. MCP Harbor re-reads this listing from the official registry every six hours and asks the live server what it exposes, and keeps what changed between one look and the next.

  • llms.txt +10 −11

    https://deepdna.ai/llms.txt 

    − > DeepDNA is a genomic knowledge and interpretation API for AI agents and developers, built in Europe. Live today: curated, cited records on genes, variants and drug–gene pairs, free with no key. Designed, and in private beta: parsing consumer DNA files (23andMe, AncestryDNA, MyHeritage; VCF later in the beta), annotating them against ClinVar, PharmGKB, gnomAD and GWAS, turning lab report PDFs into structured biomarkers, and crossing DNA with bloodwork. **The parsing endpoints are specified and − - Status: knowledge endpoints live (no key); parsing endpoints specified, not deployed, opened as a private beta partner by partner; MCP server live at https://deepdna.ai/mcp for the knowledge layer (no key, takes no genotype); local package deepdna-mcp (PyPI, MIT; registry ai.deepdna/local) reads a raw DNA file on the user's own computer at the curated positions and sends DeepDNA nothing from it− - Private beta: `POST /dna/parse`, `POST /dna/annotate`, `POST /bloodwork/parse`, `POST /crossover`− - Pricing: knowledge endpoints stay free; parsing free during the beta, then priced by usage with a free tier− - Data: nothing personal is received by the knowledge endpoints; parsing endpoints are designed for zero retention; for the beta DeepDNA acts as processor (GDPR Art. 28) for special-category data (Art. 9); based in Spain, under Spanish and EU law− - [Homepage](https://deepdna.ai/): What the API does, what is live, what is in beta, and the beta request form− - [API discovery](https://deepdna.ai/api/v1): Live JSON listing live and beta endpoints− - [Pricing](https://deepdna.ai/pricing/): Knowledge endpoints stay free; parsing free in beta, then priced by usage with a free tier− - [Methodology](https://deepdna.ai/methodology/): Variant interpretation, evidence grading, validation standards− - [A Pharmacogenomics MCP Server for Claude, Cursor and VS Code](https://deepdna.ai/blog/pharmacogenomics-mcp-server/): Connect https://deepdna.ai/mcp in one line (no key): medication screen, gene, rsID and drug–gene lookups with sources and a basis field (40 of 51 pairs cite CPIC/DPWG), why it refuses genotypes, and what the server does not do− - [Best 23andMe Alternatives in Europe (2026)](https://deepdna.ai/blog/23andme-alternatives-europe-2026/): Comparison of Promethease, SelfDecode, Genomelink, and other raw data analysis services available in Europe+ > DeepDNA serves curated, cited records on genes, variants and drug–gene pairs to AI agents and developers, built in Europe. Three things exist today, all free with no key and no account: a remote MCP server at https://deepdna.ai/mcp, the same records as a REST API at https://deepdna.ai/api/v1, and an open-source local package, deepdna-mcp, that reads a raw DNA file on the user's own computer and sends nothing from the file to DeepDNA. **DeepDNA does not parse DNA files or lab reports on its ser+ - Status: knowledge endpoints live (no key); MCP server live at https://deepdna.ai/mcp for the knowledge layer (no key, takes no genotype); local package deepdna-mcp (PyPI, MIT; registry ai.deepdna/local) reads a raw DNA file on the user's own computer at the curated positions and sends DeepDNA nothing from it+ - Pricing: free. The knowledge endpoints, the MCP server and the local package need no key and no account+ - Data: nothing personal is received by the knowledge endpoints or the MCP server; DeepDNA does not parse DNA files or lab reports on its servers, and the local package sends nothing from the file to DeepDNA; based in Spain, under Spanish and EU law+ - [Homepage](https://deepdna.ai/): What DeepDNA is: the MCP server, the knowledge API and the local package+ - [Connect the MCP server](https://deepdna.ai/connect/): Setup for Claude, ChatGPT, Claude Code, Cursor, VS Code and Gemini CLI; how to check it works+ - [API discovery](https://deepdna.ai/api/v1): Live JSON listing the live endpoints and the MCP server+ - [Methodology](https://deepdna.ai/methodology/): How the curated records are written from guidelines and the literature, what each field means, how they are checked, and what the API does not do+ - [Changelog](https://deepdna.ai/changelog/): Reviewed changes to the records and when each source (CPIC, ClinVar, FDA labels, AlphaMissense, Europe PMC) was last checked; JSON at https://deepdna.ai/api/v1/freshness+ - [A Pharmacogenomics MCP Server for Claude, Cursor and VS Code](https://deepdna.ai/blog/pharmacogenomics-mcp-server/): Connect https://deepdna.ai/mcp in one line (no key): medication screen, gene, rsID and drug–gene lookups with sources and a basis field (72 of 83 pairs cite CPIC/DPWG), why it refuses genotypes, and what the server does not do

  • server.json 2 fields changed
    repository_url
    — https://github.com/alfonsoguv/deepdna-mcp
    version
    1.3.0 1.3.1

  • Tools −1
    • − request_beta_access
  • server.json 1 field changed
    version
    1.2.0 1.3.0

  • llms.txt +3 −2

    https://deepdna.ai/llms.txt 

    − - Status: knowledge endpoints live (no key); parsing endpoints specified, not deployed, opened as a private beta partner by partner; MCP server live at https://deepdna.ai/mcp for the knowledge layer (no key), parsing tools join it with the beta− - [A Pharmacogenomics MCP Server for Claude, Cursor and VS Code](https://deepdna.ai/blog/pharmacogenomics-mcp-server/): Connect https://deepdna.ai/mcp in one line (no key): medication screen, gene, rsID and drug–gene lookups with sources and a basis field (40 of 51 pairs cite CPIC/DPWG), an optional genotype that only counts allele copies, and what the server does not do+ - Status: knowledge endpoints live (no key); parsing endpoints specified, not deployed, opened as a private beta partner by partner; MCP server live at https://deepdna.ai/mcp for the knowledge layer (no key, takes no genotype); local package deepdna-mcp (PyPI, MIT; registry ai.deepdna/local) reads a raw DNA file on the user's own computer at the curated positions and sends DeepDNA nothing from it+ - [A Pharmacogenomics MCP Server for Claude, Cursor and VS Code](https://deepdna.ai/blog/pharmacogenomics-mcp-server/): Connect https://deepdna.ai/mcp in one line (no key): medication screen, gene, rsID and drug–gene lookups with sources and a basis field (40 of 51 pairs cite CPIC/DPWG), why it refuses genotypes, and what the server does not do+ - [Read Your 23andMe File With Claude, Without Uploading It](https://deepdna.ai/blog/read-23andme-file-with-claude/): A free local MCP package, deepdna-mcp, reads a 23andMe, AncestryDNA, MyHeritage or FamilyTreeDNA file on the user's own computer and reports each curated position with its sources (APOE, BRCA1 and HFE only when named). DeepDNA receives nothing from the file, the results go to the user's AI model, and no metabolizer status is called.

  • server.json 1 field changed
    version
    1.1.1 1.2.0
  • Prompts +2
    • + medication_review
    • + explain_variant
  • Tools +2
    • + check_medications
    • + request_beta_access

  • llms.txt +3

    https://deepdna.ai/llms.txt 

    + - [A Pharmacogenomics MCP Server for Claude, Cursor and VS Code](https://deepdna.ai/blog/pharmacogenomics-mcp-server/): Connect https://deepdna.ai/mcp in one line (no key): medication screen, gene, rsID and drug–gene lookups with sources and a basis field (40 of 51 pairs cite CPIC/DPWG), an optional genotype that only counts allele copies, and what the server does not do+ - [The 23andMe API Is Gone: Reading 23andMe Data in Python](https://deepdna.ai/blog/23andme-api-alternative-python/): What happened to api.23andme.com (switched off in 2018), the raw data file that replaced it, four format details that break parsers (GRCh37, i-IDs, `--`, haploid calls), a dependency-free reader and a GRCh38 remap with `snps`+ - [Your Whole Genome Is Sequenced. Why Has It Done Nothing?](https://deepdna.ai/blog/whole-genome-sequenced-now-what/): Why a sequenced genome sits idle (read before ClinVar, ACMG SF and gnomAD existed), what to ask it once (ACMG v3.3 actionable genes, pharmacogenomics, carrier status), genotypes that change the reading of HbA1c, Lp(a), ferritin, bilirubin, neutrophils and PSA, somatic signals in a blood genome (CHIP, loss of Y), what long reads add, and why it can't predict lifespan

  • server.json 2 fields changed
    description
    Cited gene, variant (rsID) and CPIC drug–gene lookups for AI agents. Read-only, no key. Pharmacogenomics for AI agents: cited drug–gene, gene and rsID lookups plus a medication screen.
    version
    1.0.0 1.1.1

  • llms.txt +2

    https://deepdna.ai/llms.txt 

    + - [CYP2C19 and SLCO1B1 From a Raw DNA File in Python](https://deepdna.ai/blog/pharmacogenomics-raw-dna-python/): Parse 23andMe and AncestryDNA files, call CYP2C19 star alleles and SLCO1B1 rs4149056, map them to CPIC phenotypes and fetch cited guidance from /pgx, with working code+ - [PharmGKB API Not Working? Migrate to ClinPGx in Python](https://deepdna.ai/blog/pharmgkb-api-clinpgx-migration/): api.pharmgkb.org was retired on 20 July 2026; moving to api.clinpgx.org with a 2 req/s throttle, loud errors and a self-test, plus the 2026 CPIC API renames and what the licences say
DeepDNA MCP server