pocketminer/main.nf

#!/usr/bin/env nextflow

nextflow.enable.dsl=2

// Pipeline parameters
params.pdb = 's3://omic/eureka/Pocketminer/1HSG.pdb'
params.outdir = 's3://omic/eureka/Pocketminer/output'
params.model_path = '/workspace/gvp/models/pocketminer'
params.debug = false

process POCKETMINER {
    container 'harbor.cluster.omic.ai/omic/pocketminer:latest'
    publishDir params.outdir, mode: 'copy'

    input:
        val pdb_path

    output:
        path "*-preds.npy", emit: predictions_npy
        path "*-predictions.txt", emit: predictions_txt
        path "*-summary.json", emit: summary
        path "*_X.npy", optional: true, emit: features_debug
        path "*_S.npy", optional: true, emit: sequence_debug
        path "*_mask.npy", optional: true, emit: mask_debug
        path "run.log", emit: log

    script:
        def debug_flag = params.debug ? '--debug' : ''
        """
        set +e
        touch run.log

        echo "=== Environment Debug ===" >> run.log
        echo "Input path: ${pdb_path}" >> run.log
        echo "AWS_ENDPOINT_URL: \${AWS_ENDPOINT_URL:-not set}" >> run.log
        echo "AWS_ACCESS_KEY_ID set: \${AWS_ACCESS_KEY_ID:+yes}" >> run.log
        ls /omic/eureka/ >> run.log 2>&1 || echo "/omic/eureka not mounted" >> run.log
        echo "=== End Debug ===" >> run.log

        PDB_FILE=""

        # Method 1: Direct path (if PVC mounted or local)
        if [ -f "${pdb_path}" ]; then
            echo "Found at direct path" >> run.log
            cp "${pdb_path}" input.pdb && PDB_FILE="input.pdb"
        fi

        # Method 2: Convert s3://omic/... to /omic/... (PVC mount path)
        if [ -z "\$PDB_FILE" ]; then
            PVC_PATH="\$(echo '${pdb_path}' | sed 's|^s3://|/|')"
            if [ -f "\$PVC_PATH" ]; then
                echo "Found at PVC path: \$PVC_PATH" >> run.log
                cp "\$PVC_PATH" input.pdb && PDB_FILE="input.pdb"
            else
                echo "Not found at PVC path: \$PVC_PATH" >> run.log
            fi
        fi

        # Method 3: Download from MinIO using python boto3/S3
        if [ -z "\$PDB_FILE" ]; then
            S3_PATH="${pdb_path}"
            ENDPOINT="\${AWS_ENDPOINT_URL:-http://datalake-hl.datalake.svc.cluster.local:9000}"
            echo "Downloading via python S3 client from: \$ENDPOINT" >> run.log
            echo "S3 path: \$S3_PATH" >> run.log
            echo "AWS_ACCESS_KEY_ID: \${AWS_ACCESS_KEY_ID:+set}" >> run.log
            echo "AWS_SECRET_ACCESS_KEY: \${AWS_SECRET_ACCESS_KEY:+set}" >> run.log
            echo "AWS_ENDPOINT_URL: \${AWS_ENDPOINT_URL:-not set}" >> run.log
            python -c "
import os, sys
s3_path = '\$S3_PATH'
endpoint = '\$ENDPOINT'

# Parse s3://bucket/key
path = s3_path.replace('s3://', '')
parts = path.split('/', 1)
bucket = parts[0]
key = parts[1] if len(parts) > 1 else ''

print(f'Bucket: {bucket}, Key: {key}')
print(f'Endpoint: {endpoint}')

try:
    import boto3
    from botocore.client import Config
    s3 = boto3.client('s3',
        endpoint_url=endpoint,
        aws_access_key_id=os.environ.get('AWS_ACCESS_KEY_ID', ''),
        aws_secret_access_key=os.environ.get('AWS_SECRET_ACCESS_KEY', ''),
        config=Config(signature_version='s3v4'),
        region_name='us-east-1'
    )
    s3.download_file(bucket, key, 'input.pdb')
    print(f'Downloaded via boto3 ({os.path.getsize(\"input.pdb\")} bytes)')
except ImportError:
    print('boto3 not available, trying urllib with signing...')
    # Fallback: use subprocess to call python with hmac signing
    import urllib.request, hmac, hashlib, datetime
    access_key = os.environ.get('AWS_ACCESS_KEY_ID', '')
    secret_key = os.environ.get('AWS_SECRET_ACCESS_KEY', '')
    if access_key and secret_key:
        # Simple S3 GET with AWS Signature V2
        date_str = datetime.datetime.utcnow().strftime('%a, %d %b %Y %H:%M:%S GMT')
        string_to_sign = f'GET\n\n\n{date_str}\n/{bucket}/{key}'
        signature = hmac.new(secret_key.encode(), string_to_sign.encode(), hashlib.sha1)
        import base64
        sig_b64 = base64.b64encode(signature.digest()).decode()
        url = f'{endpoint}/{bucket}/{key}'
        req = urllib.request.Request(url)
        req.add_header('Date', date_str)
        req.add_header('Authorization', f'AWS {access_key}:{sig_b64}')
        with urllib.request.urlopen(req) as resp:
            with open('input.pdb', 'wb') as f:
                f.write(resp.read())
        print(f'Downloaded via signed URL ({os.path.getsize(\"input.pdb\")} bytes)')
    else:
        print('No AWS credentials available')
        sys.exit(1)
except Exception as e:
    print(f'Download failed: {e}')
    sys.exit(1)
" >> run.log 2>&1
            if [ -f input.pdb ] && [ -s input.pdb ]; then
                echo "S3 download successful (\$(wc -c < input.pdb) bytes)" >> run.log
                PDB_FILE="input.pdb"
            else
                echo "S3 download failed" >> run.log
                rm -f input.pdb
            fi
        fi

        set -e

        if [ -z "\$PDB_FILE" ] || [ ! -f input.pdb ]; then
            echo "ERROR: Could not retrieve PDB file from any source" >> run.log
            cat run.log
            exit 1
        fi

        echo "PDB file ready (\$(wc -c < input.pdb) bytes)" >> run.log
        PDB_BASENAME="\$(basename '${pdb_path}' .pdb)"

        python /workspace/entrypoint.py \\
            --pdb input.pdb \\
            --output-folder . \\
            --output-name "\$PDB_BASENAME" \\
            --model-path ${params.model_path} \\
            ${debug_flag} 2>&1 | tee -a run.log

        echo "Pipeline completed successfully" >> run.log
        """
}

workflow {
    POCKETMINER(params.pdb)
}