setup_environments.sh

#!/bin/bash
# Setup script for creating tool-specific mamba environments for benchmarking.
#
# This script creates isolated environments for each benchmarking tool to avoid
# dependency conflicts and ensure reproducible results.

set -e  # Exit on any error

echo "Setting up tool-specific environments for benchmarking..."

# Auto-detect mamba/micromamba executable (portable across users and systems)
detect_mamba_executable() {
    # Try various common mamba executables in order of preference
    local mamba_candidates=(
        "mamba"           # Standard mamba
        "micromamba"      # Micromamba
        "conda"           # Fallback to conda
    )
    
    for candidate in "${mamba_candidates[@]}"; do
        if command -v "$candidate" &> /dev/null; then
            echo "$candidate"
            return 0
        fi
    done
    
    echo "Error: No mamba/conda executable found in PATH" >&2
    echo "Please install mamba, micromamba, or conda and ensure it's in your PATH" >&2
    return 1
}

# Auto-detect mamba root prefix (portable)
detect_mamba_root() {
    local mamba_cmd="$1"
    
    # Get the full path to the mamba executable
    local mamba_full_path
    mamba_full_path=$(command -v "$mamba_cmd")
    
    # Try to get root prefix from mamba itself (try multiple methods)
    local root_prefix
    
    # Method 1: mamba info --base
    root_prefix=$("$mamba_full_path" info --base 2>/dev/null | grep -o '/[^[:space:]]*' | head -1 || true)
    if [ -n "$root_prefix" ] && [ -d "$root_prefix" ]; then
        echo "$root_prefix"
        return 0
    fi
    
    # Method 2: mamba info (parse output)
    root_prefix=$("$mamba_full_path" info 2>/dev/null | grep "base environment" | sed 's/.*base environment[[:space:]]*:[[:space:]]*\([^[:space:]]*\).*/\1/' | head -1 || true)
    if [ -n "$root_prefix" ] && [ -d "$root_prefix" ]; then
        echo "$root_prefix"
        return 0
    fi
    
    # Method 3: Try environment variable
    if [ -n "$MAMBA_ROOT_PREFIX" ] && [ -d "$MAMBA_ROOT_PREFIX" ]; then
        echo "$MAMBA_ROOT_PREFIX"
        return 0
    fi
    
    # Fallback: use common locations
    local common_roots=(
        "$HOME/mamba"
        "$HOME/micromamba"  
        "$HOME/.conda"
        "$HOME/conda"
    )
    
    for root in "${common_roots[@]}"; do
        if [ -d "$root" ]; then
            echo "$root"
            return 0
        fi
    done
    
    echo "Error: Could not detect mamba root directory" >&2
    return 1
}

# Source the portable mamba detection script
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
source "$SCRIPT_DIR/detect_mamba_env.sh"

if [ -z "$MAMBA_EXE" ] || [ -z "$MAMBA_ROOT_PREFIX" ]; then
    echo "Error: Could not detect mamba configuration"
    exit 1
fi

echo "Using mamba executable: $MAMBA_EXE"
echo "Using mamba root: $MAMBA_ROOT_PREFIX"

# Set MAMBA_CMD for backward compatibility
MAMBA_CMD="$MAMBA_EXE"

# Function to create environment if it doesn't exist
create_env_if_missing() {
    local env_name=$1
    local packages=$2
    local channels=$3
    
    if "$MAMBA_CMD" env list | grep -q "^$env_name "; then
        echo "Environment $env_name already exists, skipping..."
    else
        echo "Creating environment: $env_name"
        echo "Packages: $packages"
        echo "Channels: $channels"
        echo "Running micromamba create command..."
        if [ -n "$channels" ]; then
            "$MAMBA_CMD" create -n "$env_name" $channels $packages -y -v
        else
            "$MAMBA_CMD" create -n "$env_name" $packages -y -v
        fi
        echo "✓ Created $env_name"
    fi
}

# 1. Scanpy environment
echo "=== Setting up scanpy environment ==="
create_env_if_missing "scanpy-env" \
    "python=3.10 scanpy pandas numpy scipy matplotlib seaborn anndata pyyaml" \
    "-c conda-forge -c bioconda"

# 2. MELD environment  
echo "=== Setting up MELD environment ==="
create_env_if_missing "meld-env" \
    "python=3.10 pandas numpy scipy matplotlib seaborn anndata scanpy pyyaml" \
    "-c conda-forge -c bioconda"

# Install MELD via pip (not available in conda)
echo "Installing MELD via pip..."
echo "This may take a few minutes..."
"$MAMBA_CMD" run -n meld-env pip install meld -v
echo "✓ MELD installed"

# 3. scvi-tools environment (for mrVI)
echo "=== Setting up scvi-tools environment ==="
create_env_if_missing "scvi-env" \
    "python=3.10 pandas numpy scipy matplotlib seaborn anndata scanpy pytorch pyyaml" \
    "-c conda-forge -c bioconda -c pytorch"

# Install scvi-tools via pip
echo "Installing scvi-tools via pip..."
echo "This may take several minutes (large package)..."
"$MAMBA_CMD" run -n scvi-env pip install scvi-tools -v
echo "✓ scvi-tools installed"

# Install mrvi package for multi-resolution variational inference
echo "Installing mrvi package..."
"$MAMBA_CMD" run -n scvi-env pip install mrvi -v
echo "✓ mrvi installed"

# 4. R environments
echo "=== Setting up R environments ==="

# miloR environment
create_env_if_missing "milo-env" \
    "r-base=4.3 r-devtools r-biocmanager r-reticulate r-matrix r-igraph r-dplyr r-yaml python=3.10 anndata pyyaml pkg-config freetype fontconfig bioconductor-milor r-systemfonts" \
    "-c conda-forge -c bioconda"

# miloR packages will be installed via renv

# Seurat environment
create_env_if_missing "seurat-env" \
    "r-base=4.3 r-seurat r-reticulate r-matrix r-mass python=3.10 anndata numpy pandas scipy pyyaml zlib xz liblzma-devel pkg-config freetype-devel-cos7-x86_64 fontconfig-devel-cos7-x86_64" \
    "bioconductor-hdf5array bioconductor-rhdf5" \
    "-c conda-forge -c bioconda"

# DESeq2 environment
create_env_if_missing "deseq2-env" \
    "r-base=4.3 r-devtools r-biocmanager r-reticulate r-mass python=3.10 anndata pyyaml bioconductor-deseq2 zlib xz liblzma-devel pkg-config freetype-devel-cos7-x86_64 fontconfig-devel-cos7-x86_64" \
    "bioconductor-hdf5array bioconductor-rhdf5" \
    "-c conda-forge -c bioconda"

echo "✓ DESeq2 installed via conda"

# LEMUR environment
create_env_if_missing "lemur-env" \
    "r-base=4.3 r-devtools r-biocmanager r-reticulate r-matrix r-mass python=3.10 anndata pyyaml zlib xz liblzma-devel pkg-config freetype-devel-cos7-x86_64 fontconfig-devel-cos7-x86_64" \
    "bioconductor-hdf5array bioconductor-rhdf5" \
    "-c conda-forge -c bioconda"

# LEMUR packages will be installed via renv

echo ""
echo "=== Environment Setup Complete ==="
echo ""
echo "Available environments:"
"$MAMBA_CMD" env list | grep -E "(scanpy-env|meld-env|scvi-env|milo-env|seurat-env|deseq2-env|lemur-env)"

echo ""
echo "To test an environment:"
echo "  micromamba activate scanpy-env"
echo "  python -c 'import scanpy; print(scanpy.__version__)'"
echo ""
echo "To test R environments:"
echo "  micromamba activate milo-env"
echo "  R -e 'library(miloR); packageVersion(\"miloR\")'"
echo ""
echo "Environment setup complete! You can now run tool benchmarking."