基因命名解析服务

# Install the latest version from GitHub
remotes::install_github("armish/hgnc.mcp")

# First, generate documentation (required before installation)
source("generate-docs.R")

# Then install the package
devtools::install(".")

roxygen2::roxygenise()
devtools::install(".")

# Load HGNC data (downloads and caches on first use)
hgnc <- load_hgnc_data()

# Check cache status
get_hgnc_cache_info()

# Force refresh the cache
hgnc <- load_hgnc_data(force = TRUE)

# Or explicitly download
download_hgnc_data(force = TRUE)

# Clear cache
clear_hgnc_cache()

# Check if cache is fresh (default: 30 days)
is_hgnc_cache_fresh()
is_hgnc_cache_fresh(max_age_days = 7)

# Load the package
library(hgnc.mcp)

# Check dependencies
check_mcp_dependencies()

# Start with stdio transport (recommended for desktop clients like Claude Desktop)
start_hgnc_mcp_server(transport = "stdio")

# Or start HTTP server on default port 8080 (for web-based or remote clients)
start_hgnc_mcp_server(transport = "http")

# Customize HTTP server configuration
start_hgnc_mcp_server(
  transport = "http",
  port = 9090,
  host = "0.0.0.0",
  swagger = TRUE
)

# Start with stdio transport (for desktop clients)
Rscript inst/scripts/run_server.R --stdio

# Start HTTP server (default behavior)
Rscript inst/scripts/run_server.R

# Custom port for HTTP mode
Rscript inst/scripts/run_server.R --port 9090

# Update cache before starting
Rscript inst/scripts/run_server.R --update-cache

# Disable Swagger UI (HTTP mode only)
Rscript inst/scripts/run_server.R --no-swagger

# Get help
Rscript inst/scripts/run_server.R --help

{
  "mcpServers": {
    "hgnc": {
      "command": "Rscript",
      "args": ["-e", "hgnc.mcp::start_hgnc_mcp_server(transport='stdio')"],
      "env": {
        "HGNC_CACHE_DIR": "${HOME}/.cache/hgnc"
      }
    }
  }
}

{
  "mcpServers": {
    "hgnc": {
      "command": "docker",
      "args": ["run", "--rm", "-i", "-v", "hgnc-cache:/home/hgnc/.cache/hgnc", "ghcr.io/armish/hgnc.mcp:latest", "--stdio"]
    }
  }
}

http://localhost:8080/__docs__/

# Pull the pre-built image (supports both amd64 and arm64)
docker pull ghcr.io/armish/hgnc.mcp:latest

# For Claude Desktop (stdio mode)
docker run --rm -i \
  -v hgnc-cache:/home/hgnc/.cache/hgnc \
  ghcr.io/armish/hgnc.mcp:latest --stdio

# For HTTP server mode
docker run -d \
  --name hgnc-mcp-server \
  -p 8080:8080 \
  -v hgnc-cache:/home/hgnc/.cache/hgnc \
  ghcr.io/armish/hgnc.mcp:latest

# Access the server (HTTP mode only)
open http://localhost:8080/__docs__/

# Clone the repository
git clone https://github.com/armish/hgnc.mcp.git
cd hgnc.mcp

# Build the Docker image
docker build -t hgnc-mcp:latest .

# Run in stdio mode (for Claude Desktop)
docker run --rm -i \
  -v hgnc-cache:/home/hgnc/.cache/hgnc \
  hgnc-mcp:latest --stdio

# Or run in HTTP mode
docker run -d \
  --name hgnc-mcp-server \
  -p 8080:8080 \
  -v hgnc-cache:/home/hgnc/.cache/hgnc \
  hgnc-mcp:latest

# Start the server and supporting services
docker compose up -d

# View logs
docker compose logs -f

# Test the server
docker compose --profile test up hgnc-test-client

# Stop the server
docker compose down

services:
  hgnc-mcp-server:
    image: ghcr.io/armish/hgnc.mcp:latest
    ports:
      - "127.0.0.1:8080:8080"  # Only expose to localhost
    volumes:
      - hgnc-cache:/home/hgnc/.cache/hgnc
    restart: unless-stopped
    deploy:
      resources:
        limits:
          cpus: '2'
          memory: 4G
        reservations:
          cpus: '0.5'
          memory: 1G
    healthcheck:
      test: ["CMD", "Rscript", "-e", "tryCatch(httr::GET('http://localhost:8080/__docs__/'), error = function(e) quit(status=1))"]
      interval: 30s
      timeout: 10s
      retries: 3

# Tag for AWS ECR
docker tag hgnc-mcp:latest <account-id>.dkr.ecr.<region>.amazonaws.com/hgnc-mcp:latest

# Push to ECR
docker push <account-id>.dkr.ecr.<region>.amazonaws.com/hgnc-mcp:latest

# Deploy using ECS task definition

# Tag for Google Container Registry
docker tag hgnc-mcp:latest gcr.io/<project-id>/hgnc-mcp:latest

# Push to GCR
docker push gcr.io/<project-id>/hgnc-mcp:latest

# Deploy to Cloud Run
gcloud run deploy hgnc-mcp \
  --image gcr.io/<project-id>/hgnc-mcp:latest \
  --port 8080 \
  --memory 4Gi \
  --cpu 2

# Create resource group
az group create --name hgnc-mcp-rg --location eastus

# Deploy container
az container create \
  --resource-group hgnc-mcp-rg \
  --name hgnc-mcp-server \
  --image ghcr.io/armish/hgnc.mcp:latest \
  --ports 8080 \
  --cpu 2 \
  --memory 4

# Clone the repository
git clone https://github.com/armish/hgnc.mcp.git
cd hgnc.mcp

# Install dependencies
R -e "install.packages('remotes'); remotes::install_deps()"

# Install the package
R CMD INSTALL .

# Start the server
Rscript inst/scripts/run_server.R --port 8080

export HGNC_CACHE_DIR=/data/hgnc
export MCP_SERVER_PORT=9090
Rscript inst/scripts/run_server.R

#!/bin/bash
# health_check.sh
if curl -f -s http://localhost:8080/__docs__/ > /dev/null; then
  echo "Server is healthy"
  exit 0
else
  echo "Server is unhealthy"
  exit 1
fi

# Find what's using the port
lsof -i :8080
# Use a different port
docker run -p 9090:8080 hgnc-mcp:latest

# Increase Docker memory limit
docker update --memory="4g" hgnc-mcp-server

# Verify volume exists
docker volume inspect hgnc-cache
# Check mount point
docker inspect hgnc-mcp-server | grep Mounts -A 10

library(hgnc.mcp)

# Search for genes
results <- hgnc_find("BRCA")

# Fetch a specific gene
gene <- hgnc_fetch("symbol", "BRCA1")

# Resolve a symbol (handles aliases and previous symbols)
resolution <- hgnc_resolve_symbol("BRCA1", mode = "lenient")

# Normalize a list of gene symbols
symbols <- c("BRCA1", "tp53", "EGFR", "OLD_SYMBOL", "invalid")
result <- hgnc_normalize_list(symbols)

# View results
print(result$results)
print(result$summary)
print(result$warnings)

# Validate a gene panel
panel <- c("BRCA1", "BRCA2", "TP53", "ATM", "CHEK2")
validation <- hgnc_validate_panel(panel)

# Check for issues
print(validation$summary)
print(validation$report)

# Find genes modified in the last 30 days
recent_changes <- hgnc_changes(since = Sys.Date() - 30)
print(recent_changes$summary)

# Track symbol changes since a specific date
symbol_changes <- hgnc_changes(
  since = "2024-01-01",
  change_type = "symbol"
)

# Search for gene groups
kinases <- hgnc_search_groups("kinase")

# Get members of a specific group
members <- hgnc_group_members("Protein kinases")

# Load a clinical panel from a CSV file
clinical_panel <- read.csv("hereditary_cancer_panel.csv")$gene_symbol

# Validate against HGNC standards
validation <- hgnc_validate_panel(clinical_panel, policy = "HGNC")

# Review any issues
if (validation$summary$status != "PASS") {
  cat("Issues found:\n")
  print(validation$report)

  # Get replacement suggestions
  if (!is.null(validation$suggestions)) {
    cat("\nSuggested updates:\n")
    print(validation$suggestions[, c("input_symbol", "suggested_symbol", "reason")])
  }
}

# Generate normalized panel for clinical use
normalized <- hgnc_normalize_list(
  clinical_panel,
  return_fields = c("symbol", "name", "hgnc_id", "omim_id", "location")
)

# Export for lab reporting system
write.csv(normalized$results, "validated_panel.csv", row.names = FALSE)

# Combine gene lists from different studies
study1_genes <- read.csv("rnaseq_study1.csv")$gene
study2_genes <- read.csv("microarray_study2.csv")$gene
study3_genes <- read.csv("proteomics_study3.csv")$gene

all_genes <- unique(c(study1_genes, study2_genes, study3_genes))

# Normalize to current HGNC symbols
normalized <- hgnc_normalize_list(
  all_genes,
  return_fields = c("symbol", "name", "hgnc_id", "entrez_id", "ensembl_gene_id"),
  dedupe = TRUE
)

# Map back to original datasets with unified nomenclature
# This eliminates false negatives from symbol inconsistencies

# Build a kinase inhibitor target panel
kinase_groups <- hgnc_search_groups("kinase")
all_kinases <- hgnc_group_members("Protein kinases")

# Filter for specific kinase families of interest
tyrosine_kinases <- hgnc_search_groups("tyrosine kinase")
target_kinases <- hgnc_group_members("Receptor tyrosine kinases")

# Get comprehensive cross-references for target validation
targets_with_xrefs <- hgnc_normalize_list(
  target_kinases$symbol,
  return_fields = c("symbol", "name", "hgnc_id", "entrez_id",
                   "ensembl_gene_id", "uniprot_id", "omim_id")
)

# Track any nomenclature changes quarterly
quarterly_changes <- hgnc_changes(since = Sys.Date() - 90, change_type = "all")
target_updates <- quarterly_changes$changes[
  quarterly_changes$changes$symbol %in% target_kinases$symbol,
]

# Document panel version with HGNC provenance
create_compliance_report <- function(panel_genes, panel_name) {
  # Normalize genes
  normalized <- hgnc_normalize_list(
    panel_genes,
    return_fields = c("symbol", "name", "hgnc_id", "status", "location")
  )

  # Validate
  validation <- hgnc_validate_panel(panel_genes)

  # Get cache info for provenance
  cache_info <- get_hgnc_cache_info()

  # Create report
  report <- list(
    panel_name = panel_name,
    report_date = Sys.Date(),
    hgnc_version = cache_info$download_date,
    hgnc_source = cache_info$source_url,
    total_genes = length(panel_genes),
    valid_genes = sum(normalized$results$status == "Approved"),
    validation_status = validation$summary$status,
    normalized_genes = normalized$results,
    validation_report = validation$report,
    warnings = normalized$warnings
  )

  # Save for audit trail
  saveRDS(report, paste0(panel_name, "_", Sys.Date(), "_compliance.rds"))

  return(report)
}

# Use for regulatory submission
panel <- c("BRCA1", "BRCA2", "TP53", "PTEN", "ATM")
compliance <- create_compliance_report(panel, "BRCA_Panel_v2")

# Parse gene symbols from abstract/full text (hypothetical)
extracted_genes <- c("p53", "BRCA-1", "EGF receptor", "HER2", "ERBB2")

# Resolve to standard HGNC symbols
resolved <- lapply(extracted_genes, function(g) {
  result <- hgnc_resolve_symbol(g, mode = "lenient")
  if (!is.null(result$approved_symbol)) {
    data.frame(
      original = g,
      approved = result$approved_symbol,
      confidence = result$confidence
    )
  }
})

# Combine results
gene_mapping <- do.call(rbind, resolved)
print(gene_mapping)

# Result:
#     original approved confidence
# 1        p53     TP53      alias
# 2     BRCA-1    BRCA1   approved
# 3  ERBB2     ERBB2   approved

# Start the MCP server
Rscript -e "library(hgnc.mcp); start_hgnc_mcp_server()"

# List all vignettes
vignette(package = "hgnc.mcp")

# View a specific vignette
vignette("getting-started", package = "hgnc.mcp")