BioDeep 0.1.0

BioDeepagents CLI and skill bundles.

BioDeepagents CLI

Skills

Skills are sourced from BioClaw. The following skills are sourced from claude-scientific-skills:

• Bioinformatics & Genomics
• Sequence Analysis: Process DNA/RNA/protein sequences with BioPython and pysam
• Single-Cell Analysis: Analyze 10X Genomics data with Scanpy, identify cell types, infer GRNs with Arboreto
• Variant Annotation: Annotate VCF files with Ensembl VEP, query ClinVar for pathogenicity
• Variant Database Management: Build scalable VCF databases with TileDB-VCF for incremental sample addition, efficient population-scale queries, and compressed storage of genomic variant data
• Gene Discovery: Query NCBI Gene, UniProt, and Ensembl for comprehensive gene information
• Network Analysis: Identify protein-protein interactions via STRING, map to pathways (KEGG, Reactome)

Skill	Description
bio-orchestrator	Meta-agent that routes bioinformatics requests to specialised sub-skills. Handles file type detection, analysis planning, report generation, and reproducibility export.
claw-ancestry-pca	Ancestry decomposition PCA against the Simons Genome Diversity Project.
claw-metagenomics	Shotgun metagenomics profiling — taxonomy, resistome, and functional pathways.
claw-semantic-sim	Semantic Similarity Index for disease research literature using PubMedBERT embeddings.
drug-photo	Photo of a medication to get pharmacogenomic dosage guidance via a CPIC single-drug lookup.
equity-scorer	Compute HEIM diversity and equity metrics from VCF or ancestry data. Generates heterozygosity, FST, PCA plots, and a composite HEIM Equity Score with markdown reports.
genome-compare	Compare your genome to George Church (PGP-1) and estimate ancestry composition.
labstep	Interact with the Labstep electronic lab notebook API using labstepPy. Query experiments, protocols, resources, inventory, and other lab entities.
lit-synthesizer	Search PubMed and bioRxiv, summarise papers with LLM, build citation graphs, and generate literature review sections.
nutrigx-advisor	Personalised nutrition report from genetic data (23andMe, AncestryDNA, or VCF).
pharmgx-reporter	Pharmacogenomic report from DTC genetic data (23andMe/AncestryDNA).
repro-enforcer	Export any bioinformatics analysis as a reproducible bundle with Conda environment, Singularity container definition, and Nextflow pipeline.
scrna-orchestrator	Automate single-cell RNA-seq analysis with Scanpy or Seurat. QC, normalisation, clustering, DE analysis, and visualisation.
seq-wrangler	Sequence QC, alignment, and BAM processing. Wraps FastQC, BWA/Bowtie2, SAMtools for automated read-to-BAM pipelines.
struct-predictor	Local protein structure prediction with AlphaFold, Boltz, or Chai. Compare predicted structures, compute RMSD, visualise 3D models.
vcf-annotator	Annotate VCF variants with VEP, ClinVar, gnomAD frequencies, and ancestry-aware context. Generates prioritised variant reports.

This project is basing on deepagents CLI, an open source coding assistant that runs in your terminal, similar to Claude Code.

Key Features:*

• Built-in Tools: File operations (read, write, edit, glob, grep), shell commands, web search, and subagent delegation
• Customizable Skills: Add domain-specific capabilities through a progressive disclosure skill system
• Persistent Memory: Agent remembers your preferences, coding style, and project context across sessions
• Project-Aware: Automatically detects project roots and loads project-specific configurations

🐳 Docker

Build the CLI image from the repository root so both the CLI sources and the shared deepagents package are available to Docker:

This project is a docker-compose hosted a deepagents-cli, and inside container some other agents(e.g. dsl) were also hosted.

docker compose run --rm --service-ports --build flask-app

Copy the sample configuration into place and edit the credentials or tracing toggles you need:

docker run --rm -it -P \
  -v "$(pwd)/workspace/:/workspace/project" \
  --env-file ./.env \
  deepagents-cli

Type naturally as you would in a chat interface. The agent will use its built-in tools, skills, and memory to help you with tasks.

[!WARNING] Human-in-the-Loop (HITL) Approval Required

Potentially destructive operations require user approval before execution:

• File operations: write_file, edit_file
• Command execution: shell, execute
• External requests: web_search, fetch_url
• Delegation: task (subagents)

Each operation will prompt for approval showing the action details. Use --auto-approve to skip prompts:

deepagents --auto-approve