The VBASE3 MCP Server (developer_scripts/vbase3_mcp_server.mjs) allows AI assistants and agentic coding environments (such as Google Antigravity IDE, Claude Desktop, Cursor, and custom LLM agents) to directly invoke the high-performance VBASE3 WebAssembly engine via standard JSON-RPC 2.0 over stdio.
Because the underlying engine runs as zero-dependency WebAssembly (web/static/pkg/dnaplot_wasm_bg.wasm), all sequence alignments, germline lookups, and synthesis boundary audits execute 100% locally within milliseconds. Proprietary therapeutic antibody sequences never leave your machine.
| Tool Name | Parameters | Description |
|---|---|---|
vbase3_analyze_sequence |
sequence, species, chain |
Performs full V(D)J alignment, IMGT unique numbering (positions 1–128), CDR1/2/3 loop detection, somatic hypermutation (SHM) counts, and in-frame productivity analysis. |
vbase3_verify_synthesis_boundary |
sequence, chain_type |
Audits variable heavy/light domain sequences for recombinant gene synthesis design. Verifies canonical C-terminal motifs (TVSS/TVSA for Heavy; VEIK/LEIK/TVL for Light) and flags any trailing constant region bleed-through. |
vbase3_trim_antibody_seq |
peptide_sequence, dna_sequence, chain_type |
Canonical sequence trimmer. Truncates trailing constant region overhang residues (e.g., GSA, RTV, GQPKAAPSD) from variable domain peptide and nucleotide sequences, terminating cleanly at the exact IMGT J-segment boundary. |
vbase3_lookup_gene |
gene_id, species |
Sub-millisecond lookup of any Class I functional V-gene or allele across 55 vertebrate species. Supports IMGT symbols (e.g. IGHV3-23*01) and VBASE gene IDs (e.g. humIGHV200187). |
vbase3_export_structure_input |
vh_sequence, vl_sequence, pipeline, identifier |
Formats paired synthesis-ready antibody VH/VL sequences for leading 3D structure prediction pipelines (colabfold, alphafold3, esmfold, boltz). Enforces synthesis boundaries before formatting. |
vbase3_calculate_biophysics |
cdr3_amino_acid |
Calculates CDR3 loop length, net charge at physiological pH 7.4, and Kyte-Doolittle hydropathy index to assess solubility and off-target liabilities. |
vbase3_get_server_status |
(none) | Returns VBASE3 database release version, vertebrate species coverage, gating policies, and regulatory Research Use Only (RUO) notice. |
For command-line AI agents and headless batch environments, the standalone WASM CLI provides sub-millisecond execution with zero external npm dependencies:
# 1. Sub-millisecond Class I gene lookup (returns JSON)
node developer_scripts/vbase3_wasm_cli.mjs --lookup IGHV3-23 --format json
# 2. Audit synthesis boundary
node developer_scripts/vbase3_wasm_cli.mjs --audit-boundary "EVQL...TVSSGSA" --format json
# 3. Canonical boundary trimming
node developer_scripts/vbase3_wasm_cli.mjs --trim-boundary "EVQL...TVSSGSA" --format json
# 4. CDR3 Biophysical profile
node developer_scripts/vbase3_wasm_cli.mjs --biophysics "CSRWGGDGFYAMDYW" --format json
# 5. Format paired FASTA for AlphaFold/ColabFold
node developer_scripts/vbase3_wasm_cli.mjs --structure "EVQL...TVSS" --vl "DIQM...VEIK" --pipeline colabfold --format jsonAdd the server definition to your claude_desktop_config.json:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"vbase3": {
"command": "node",
"args": [
"/absolute/path/to/vbase3_database/developer_scripts/vbase3_mcp_server.mjs"
]
}
}
}Add to your project or user MCP configuration (.gemini/config/mcp_config.json):
{
"mcpServers": {
"vbase3_immunogenomics": {
"command": "node",
"args": [
"developer_scripts/vbase3_mcp_server.mjs"
]
}
}
}"Analyze this variable heavy sequence for recombinant synthesis. Does it end cleanly at the canonical synthesis boundary?
EVQLVESGGGLVQPGGSLRLSCAASGFTFSDYAMSWVRQAPGKGLEWVAVISYDGSNKYYADSVKGRFTISRDNSKNTLYLQMNSLRAEDTAVYYCAKDRLSITIRPRYYGMDVWGQGTTVSSGSA"
{
"sequence_length": 126,
"detected_chain_type": "heavy",
"canonical_motif_found": "TVSS",
"is_synthesis_compliant": false,
"trailing_constant_residues": "GSA",
"recommendation": "Remove trailing constant region residues 'GSA' to terminate cleanly at 'TVSS'."
}"Your sequence contains a 3-amino acid constant-region overhang (
GSA) after the canonical IMGT position 122 boundary. For recombinant gene synthesis, you must truncate the sequence to end precisely atTVSS:...YYGMDVWGQGTTVSS."
Because all VBASE3 web applications execute client-side WebAssembly, AI assistants can also serve and embed the interactive web applications directly within IDE panels, sidecars, or webviews:
- Collier de Perles 2D Modal:
https://vbase3.github.io/vbase3_database/dnaplot.html - Single-Cell Repertoire Studio:
https://vbase3.github.io/vbase3_database/single_cell_repertoire_studio.html - Universal PaCMAP Manifold:
https://vbase3.github.io/vbase3_database/vgene_pacmap_universe.html
This architecture allows the LLM to perform deep sequence reasoning while embedding publication-ready interactive visualizations right in the developer's workspace.