Datasets:

QPromaQ
/

Docking_project

License:
Docking_project / scripts /prepare_pdb_ligand_dataset.py
Reset repository and upload final project (part 30)
504d922 verified 5 months ago
87.5 kB
from __future__ import annotations

import argparse
import csv
import json
import os
import shutil
import sys
import time
import urllib.parse
import urllib.request
from concurrent.futures import ProcessPoolExecutor, ThreadPoolExecutor, as_completed
from pathlib import Path

try:  # pragma: no cover
    from rdkit import Chem, DataStructs
    from rdkit.Chem import rdMolDescriptors
    from rdkit.Chem import rdFingerprintGenerator
    from rdkit.Chem.Scaffolds import MurckoScaffold

    RDKIT_AVAILABLE = True
except Exception:  # pragma: no cover
    Chem = None  # type: ignore[assignment]
    DataStructs = None  # type: ignore[assignment]
    rdMolDescriptors = None  # type: ignore[assignment]
    rdFingerprintGenerator = None  # type: ignore[assignment]
    MurckoScaffold = None  # type: ignore[assignment]
    RDKIT_AVAILABLE = False

try:  # pragma: no cover
    from tqdm import tqdm
except Exception:  # pragma: no cover
    tqdm = None  # type: ignore[assignment]

ROOT = Path(__file__).resolve().parents[1]
if str(ROOT) not in sys.path:
    sys.path.insert(0, str(ROOT))

from docking_pipeline.dataset import (
    auto_detect_reference_ligand,
    count_sdf_records,
    create_dataset_manifest,
    download_pdb_structure,
    extract_receptor_and_reference_ligand,
    list_hetero_ligands,
    list_known_good_complexes,
    prepare_dataset_target_with_rdock,
    resolve_known_good_defaults,
    validate_dataset_dir,
)
from docking_pipeline.provenance import CommandRunner, RDockPipelineError, fail_if_bad_command, probe_version, require_executable, require_file
from docking_pipeline.sdf import split_sdf_text
from docking_pipeline.validation import resolve_jobs


KNOWN_GOOD = ROOT / "configs" / "known_good_pdb_complexes.yaml"
DEFAULT_THRESHOLD_LADDER = [95, 90, 85, 80, 75, 70]
DEFAULT_BUNDLED_EXAMPLE = ROOT / "data" / "examples" / "example_smiles.smi"
DEFAULT_EXAMPLE_1000 = ROOT / "data" / "examples" / "example_smiles_1000.smi"


def _read_smiles(path: Path, n_ligands: int) -> list[tuple[str, str]]:
    rows: list[tuple[str, str]] = []
    for idx, line in enumerate(path.read_text(encoding="utf-8").splitlines()):
        text = line.strip()
        if not text or text.startswith("#"):
            continue
        parts = text.replace(",", " ").split()
        smiles = parts[0]
        ligand_id = parts[1] if len(parts) > 1 else f"lig_{idx:05d}"
        rows.append((smiles, ligand_id))
        if len(rows) >= n_ligands:
            break
    return rows


def _read_all_smiles(path: Path) -> list[tuple[str, str]]:
    rows: list[tuple[str, str]] = []
    for idx, line in enumerate(path.read_text(encoding="utf-8").splitlines()):
        text = line.strip()
        if not text or text.startswith("#"):
            continue
        parts = text.replace(",", " ").split()
        smiles = parts[0]
        ligand_id = parts[1] if len(parts) > 1 else f"lig_{idx:05d}"
        rows.append((smiles, ligand_id))
    return rows


def _default_ligand_jobs() -> int:
    cpu_total = os.cpu_count() or 8
    return max(1, cpu_total - 4)


def _mkdir_or_fail(path: Path) -> None:
    try:
        path.mkdir(parents=True, exist_ok=True)
    except PermissionError as exc:
        suggestion = Path.home() / "datasets" / path.name
        raise RDockPipelineError(
            f"Cannot create output directory {path}: permission denied. "
            f"Use a writable path such as {suggestion} or a relative path under your home/project directory."
        ) from exc


def _http_json(url: str, timeout: int, retries: int = 3, pause_seconds: float = 1.0) -> dict[str, object]:
    last_error: Exception | None = None
    for attempt in range(1, retries + 1):
        try:
            request = urllib.request.Request(url, headers={"User-Agent": "portable-rdock-pipeline/1.0"})
            with urllib.request.urlopen(request, timeout=timeout) as response:
                payload = response.read().decode("utf-8")
            return json.loads(payload)
        except Exception as exc:  # pragma: no cover - network-dependent
            last_error = exc
            if attempt < retries:
                time.sleep(pause_seconds * attempt)
    raise RDockPipelineError(f"HTTP JSON request failed after {retries} attempts for {url}: {last_error}")


def _http_post_json(
    url: str,
    data: bytes,
    content_type: str,
    timeout: int,
    retries: int = 3,
    pause_seconds: float = 1.0,
) -> dict[str, object]:
    last_error: Exception | None = None
    for attempt in range(1, retries + 1):
        try:
            request = urllib.request.Request(
                url,
                data=data,
                headers={"User-Agent": "portable-rdock-pipeline/1.0", "Content-Type": content_type},
                method="POST",
            )
            with urllib.request.urlopen(request, timeout=timeout) as response:
                payload = response.read().decode("utf-8")
            return json.loads(payload)
        except Exception as exc:  # pragma: no cover - network-dependent
            last_error = exc
            if attempt < retries:
                time.sleep(pause_seconds * attempt)
    raise RDockPipelineError(f"HTTP POST JSON request failed after {retries} attempts for {url}: {last_error}")


def _http_post_form_json(
    url: str,
    form: dict[str, str],
    timeout: int,
    retries: int = 3,
    pause_seconds: float = 1.0,
) -> dict[str, object]:
    body = urllib.parse.urlencode(form).encode("utf-8")
    return _http_post_json(url, body, "application/x-www-form-urlencoded", timeout, retries=retries, pause_seconds=pause_seconds)


def _write_smi(rows: list[tuple[str, str]], path: Path) -> Path:
    path.parent.mkdir(parents=True, exist_ok=True)
    path.write_text("\n".join(f"{smiles} {ligand_id}" for smiles, ligand_id in rows) + "\n", encoding="utf-8")
    return path


def _count_sdf_records_lenient(path: Path) -> int:
    if not path.exists() or path.stat().st_size <= 0:
        return 0
    text = path.read_text(encoding="utf-8", errors="ignore")
    return len(split_sdf_text(text))


def _progress_log(out: Path, message: str, payload: dict[str, object] | None = None) -> None:
    line = message
    print(line, file=sys.stderr, flush=True)
    log_path = out / "logs" / "pubchem_progress.log"
    log_path.parent.mkdir(parents=True, exist_ok=True)
    with log_path.open("a", encoding="utf-8") as handle:
        handle.write(line + "\n")
    if payload is not None:
        status_path = out / "logs" / "pubchem_progress.json"
        status_path.write_text(json.dumps(payload, indent=2), encoding="utf-8")


def _write_partial_pubchem_hits(
    out: Path,
    rows: list[tuple[str, str]],
    metadata_rows: list[dict[str, object]],
) -> None:
    if not rows:
        return
    _write_smi(rows, out / "ligands" / "pubchem_partial_hits.smi")
    _write_csv(out / "ligands" / "pubchem_partial_metadata.csv", metadata_rows)


def _obabel_convert(runner: CommandRunner, stage: str, input_path: Path, output_path: Path, extra_args: list[str], cwd: Path) -> Path:
    obabel = require_executable("obabel")
    rec = runner.run(
        stage,
        [obabel, str(input_path.resolve()), *extra_args, "-O", str(output_path.resolve())],
        cwd,
        cwd / "logs" / f"{stage}.stdout.log",
        cwd / "logs" / f"{stage}.stderr.log",
    )
    fail_if_bad_command(rec, f"OpenBabel {stage}")
    return require_file(output_path, f"OpenBabel output {stage}")


def _merge_sdf_files(inputs: list[Path], output_path: Path) -> Path:
    output_path.parent.mkdir(parents=True, exist_ok=True)
    with output_path.open("w", encoding="utf-8") as handle:
        for path in inputs:
            if not path.exists() or path.stat().st_size <= 0:
                continue
            text = path.read_text(encoding="utf-8", errors="ignore")
            if text and not text.endswith("\n"):
                text += "\n"
            handle.write(text)
    return output_path


def _convert_single_ligand_with_obabel(
    runner: CommandRunner,
    row: tuple[str, str],
    work_dir: Path,
) -> tuple[Path | None, dict[str, object] | None]:
    smiles, ligand_id = row
    smi_path = work_dir / f"{ligand_id}.smi"
    sdf_path = work_dir / f"{ligand_id}.sdf"
    smi_path.write_text(f"{smiles} {ligand_id}\n", encoding="utf-8")
    try:
        _obabel_convert(runner, f"single_{ligand_id}_to_sdf", smi_path, sdf_path, ["--gen3d", "-h"], work_dir)
        if _count_sdf_records_lenient(sdf_path) != 1:
            raise RDockPipelineError(f"Expected 1 SDF record for ligand {ligand_id}, got {_count_sdf_records_lenient(sdf_path)}")
        return sdf_path, None
    except Exception as exc:
        return None, {"ligand_id": ligand_id, "smiles": smiles, "reason": str(exc), "stage": "single_ligand_fallback"}


def _convert_batch_with_obabel(
    runner: CommandRunner,
    batch_rows: list[tuple[str, str]],
    batch_dir: Path,
    batch_name: str,
    single_fallback_workers: int,
) -> tuple[Path, list[dict[str, object]]]:
    batch_dir.mkdir(parents=True, exist_ok=True)
    batch_smi = _write_smi(batch_rows, batch_dir / f"{batch_name}.smi")
    batch_sdf = batch_dir / f"{batch_name}.sdf"
    invalid_rows: list[dict[str, object]] = []
    try:
        _obabel_convert(runner, f"{batch_name}_to_sdf", batch_smi, batch_sdf, ["--gen3d", "-h"], batch_dir)
        if _count_sdf_records_lenient(batch_sdf) == len(batch_rows):
            return batch_sdf, invalid_rows
    except Exception as exc:
        invalid_rows.append({"ligand_id": batch_name, "smiles": "", "reason": str(exc), "stage": "batch_conversion"})

    single_dir = batch_dir / f"{batch_name}_single"
    single_dir.mkdir(parents=True, exist_ok=True)
    single_outputs: dict[str, Path] = {}
    with ThreadPoolExecutor(max_workers=min(max(1, single_fallback_workers), len(batch_rows))) as pool:
        futures = {
            pool.submit(_convert_single_ligand_with_obabel, runner, row, single_dir): row
            for row in batch_rows
        }
        for future in as_completed(futures):
            row = futures[future]
            output_path, invalid = future.result()
            if output_path is not None:
                single_outputs[row[1]] = output_path
            if invalid is not None:
                invalid_rows.append(invalid)
    ordered_outputs = [single_outputs[ligand_id] for _, ligand_id in batch_rows if ligand_id in single_outputs]
    _merge_sdf_files(ordered_outputs, batch_sdf)
    return batch_sdf, invalid_rows


def _convert_batch_with_obabel_worker(
    batch_rows: list[tuple[str, str]],
    batch_dir: str,
    batch_name: str,
    single_fallback_workers: int,
) -> tuple[str, list[dict[str, object]]]:
    batch_path = Path(batch_dir)
    runner = CommandRunner(batch_path / "logs" / "commands.log")
    sdf_path, invalid_rows = _convert_batch_with_obabel(
        runner,
        batch_rows,
        batch_path,
        batch_name,
        single_fallback_workers,
    )
    return str(sdf_path), invalid_rows


def _prepare_ligands_from_existing_smi(
    out: Path,
    runner: CommandRunner,
    batch_size: int,
    jobs: int | str,
    cpu_fraction: float,
    force_rebuild: bool = False,
    metadata_rows: list[dict[str, object]] | None = None,
) -> dict[str, object]:
    lig_root = out / "ligands"
    smi_path = require_file(lig_root / "all_ligands.smi", "all_ligands.smi for ligand preparation")
    rows = _read_all_smiles(smi_path)
    if not rows:
        raise RDockPipelineError(f"No usable SMILES rows found in {smi_path}")
    expected_total = len(rows)
    final_sdf = lig_root / "all_ligands.sdf"
    existing_final = _count_sdf_records_lenient(final_sdf)
    if existing_final > expected_total:
        raise RDockPipelineError(
            f"Existing {final_sdf} contains {existing_final} records but {smi_path} contains only {expected_total} ligands. "
            "Refuse to resume from an inconsistent dataset."
        )
    if not force_rebuild and existing_final == expected_total:
        metadata_path = lig_root / "ligand_metadata.csv"
        if not metadata_path.exists():
            _write_csv(metadata_path, [{"ligand_id": ligand_id, "smiles": smiles, "source": "smiles_file"} for smiles, ligand_id in rows])
        invalid_path = lig_root / "invalid_ligands.csv"
        if not invalid_path.exists():
            _write_csv(invalid_path, [])
        return {
            "mode": "prepare_ligands_only",
            "status": "already_complete",
            "expected_ligands": expected_total,
            "prepared_ligands": existing_final,
            "batch_size": batch_size,
            "jobs": _default_ligand_jobs() if str(jobs).strip().lower() == "auto" else resolve_jobs(jobs, cpu_fraction),
            "final_sdf": str(final_sdf),
        }

    batch_root = lig_root / "obabel_batches"
    batch_root.mkdir(parents=True, exist_ok=True)
    progress_path = lig_root / "ligand_preparation_progress.json"
    resume_prefix_count = 0
    recovered_prefix_sdf: Path | None = None
    if not force_rebuild and existing_final > 0:
        existing_batch_files = list(batch_root.glob("batch_*/*.sdf"))
        if not existing_batch_files:
            resume_prefix_count = existing_final
            recovered_prefix_sdf = batch_root / "recovered_prefix.sdf"
            shutil.copy2(final_sdf, recovered_prefix_sdf)
    batches: list[tuple[str, list[tuple[str, str]]]] = []
    for batch_index, start in enumerate(range(0, expected_total, batch_size), start=1):
        batches.append((f"batch_{batch_index:05d}", rows[start : start + batch_size]))

    resolved_jobs = _default_ligand_jobs() if str(jobs).strip().lower() == "auto" else resolve_jobs(jobs, cpu_fraction)
    invalid_rows: list[dict[str, object]] = []
    completed_batch_outputs: dict[str, Path] = {}
    if recovered_prefix_sdf is not None:
        completed_batch_outputs["__recovered_prefix__"] = recovered_prefix_sdf
    pending_batches: list[tuple[str, list[tuple[str, str]]]] = []
    ligands_skipped_by_prefix = 0
    for batch_number, (batch_name, batch_rows) in enumerate(batches, start=1):
        batch_start = (batch_number - 1) * batch_size
        batch_end = batch_start + len(batch_rows)
        if resume_prefix_count and batch_end <= resume_prefix_count:
            ligands_skipped_by_prefix += len(batch_rows)
            continue
        if resume_prefix_count and batch_start < resume_prefix_count < batch_end:
            prefix_skip = resume_prefix_count - batch_start
            ligands_skipped_by_prefix += prefix_skip
            batch_rows = batch_rows[prefix_skip:]
            if not batch_rows:
                continue
        batch_dir = batch_root / batch_name
        batch_sdf = batch_dir / f"{batch_name}.sdf"
        if (not force_rebuild) and _count_sdf_records_lenient(batch_sdf) == len(batch_rows):
            completed_batch_outputs[batch_name] = batch_sdf
            continue
        pending_batches.append((batch_name, batch_rows))

    progress_path.write_text(
        json.dumps(
            {
                "expected_ligands": expected_total,
                "prepared_batches": len(completed_batch_outputs),
                "total_batches": len(batches),
                "pending_batches": [name for name, _ in pending_batches[:1000]],
                "resumed_from_existing_final_sdf_records": resume_prefix_count,
                "ligands_skipped_by_prefix_resume": ligands_skipped_by_prefix,
                "jobs": resolved_jobs,
                "batch_size": batch_size,
            },
            indent=2,
        ),
        encoding="utf-8",
    )

    if pending_batches:
        progress = tqdm(total=len(pending_batches), desc="Ligand 3D batches", unit="batch") if tqdm is not None else None
        ligand_progress = tqdm(total=expected_total, desc="Ligands prepared", unit="lig") if tqdm is not None else None
        if ligand_progress is not None:
            ligand_progress.update(sum(_count_sdf_records_lenient(path) for path in completed_batch_outputs.values()))
        max_workers = min(resolved_jobs, len(pending_batches))
        executor_cls = ProcessPoolExecutor
        pool = None
        try:
            pool = executor_cls(max_workers=max_workers)
        except Exception:
            executor_cls = ThreadPoolExecutor
            pool = executor_cls(max_workers=max_workers)
        with pool:
            futures = {
                pool.submit(
                    _convert_batch_with_obabel_worker,
                    batch_rows,
                    str(batch_root / batch_name),
                    batch_name,
                    max(1, min(4, resolved_jobs)),
                ): batch_name
                for batch_name, batch_rows in pending_batches
            }
            for future in as_completed(futures):
                batch_name = futures[future]
                batch_sdf, batch_invalid = future.result()
                completed_batch_outputs[batch_name] = Path(batch_sdf)
                invalid_rows.extend(batch_invalid)
                batch_prepared = _count_sdf_records_lenient(Path(batch_sdf))
                if progress is not None:
                    progress.update(1)
                    progress.set_postfix(last_batch=batch_name, invalid=len(invalid_rows), worker=executor_cls.__name__)
                if ligand_progress is not None:
                    ligand_progress.update(batch_prepared)
                    ligand_progress.set_postfix(last_batch=batch_name, invalid=len(invalid_rows), worker=executor_cls.__name__)
                prepared_count = sum(_count_sdf_records_lenient(path) for path in completed_batch_outputs.values())
                progress_path.write_text(
                    json.dumps(
                        {
                            "expected_ligands": expected_total,
                            "prepared_batches": len(completed_batch_outputs),
                            "total_batches": len(batches),
                            "prepared_ligands_estimate": prepared_count,
                            "resumed_from_existing_final_sdf_records": resume_prefix_count,
                            "ligands_skipped_by_prefix_resume": ligands_skipped_by_prefix,
                            "jobs": resolved_jobs,
                            "batch_size": batch_size,
                            "worker_backend": executor_cls.__name__,
                            "last_completed_batch": batch_name,
                            "invalid_count": len(invalid_rows),
                        },
                        indent=2,
                    ),
                    encoding="utf-8",
                )
        if progress is not None:
            progress.close()
        if ligand_progress is not None:
            ligand_progress.close()

    ordered_outputs: list[Path] = []
    if "__recovered_prefix__" in completed_batch_outputs:
        ordered_outputs.append(completed_batch_outputs["__recovered_prefix__"])
    ordered_outputs.extend(completed_batch_outputs[batch_name] for batch_name, _ in batches if batch_name in completed_batch_outputs)
    _merge_sdf_files(ordered_outputs, final_sdf)
    prepared_count = _count_sdf_records_lenient(final_sdf)
    metadata_payload = metadata_rows or _existing_ligand_metadata_payload(lig_root, expected_total)
    if metadata_payload is None:
        metadata_payload = [{"ligand_id": ligand_id, "smiles": smiles, "source": "smiles_file"} for smiles, ligand_id in rows]
    _write_csv(lig_root / "ligand_metadata.csv", metadata_payload)
    _write_csv(lig_root / "invalid_ligands.csv", invalid_rows)
    progress_path.write_text(
        json.dumps(
            {
                "expected_ligands": expected_total,
                "prepared_ligands": prepared_count,
                "invalid_ligands": len(invalid_rows),
                "prepared_batches": len(ordered_outputs),
                "total_batches": len(batches),
                "resumed_from_existing_final_sdf_records": resume_prefix_count,
                "ligands_skipped_by_prefix_resume": ligands_skipped_by_prefix,
                "jobs": resolved_jobs,
                "batch_size": batch_size,
                "final_sdf": str(final_sdf),
                "status": "complete" if prepared_count > 0 else "empty",
            },
            indent=2,
        ),
        encoding="utf-8",
    )
    return {
        "mode": "prepare_ligands_only",
        "status": "complete" if prepared_count > 0 else "empty",
        "expected_ligands": expected_total,
        "prepared_ligands": prepared_count,
        "invalid_ligands": len(invalid_rows),
        "batch_size": batch_size,
        "jobs": resolved_jobs,
        "resumed_from_existing_final_sdf_records": resume_prefix_count,
        "ligands_skipped_by_prefix_resume": ligands_skipped_by_prefix,
        "final_sdf": str(final_sdf),
        "progress_json": str(progress_path),
    }


def _reference_ligand_to_smiles(runner: CommandRunner, reference_sdf: Path, cwd: Path) -> str:
    obabel = require_executable("obabel")
    stdout_log = cwd / "logs" / "reference_ligand_to_smiles.stdout.log"
    stderr_log = cwd / "logs" / "reference_ligand_to_smiles.stderr.log"
    rec = runner.run(
        "reference_ligand_to_smiles",
        [obabel, str(reference_sdf.resolve()), "-osmi"],
        cwd,
        stdout_log,
        stderr_log,
    )
    fail_if_bad_command(rec, "OpenBabel reference ligand to SMILES")
    text = stdout_log.read_text(encoding="utf-8", errors="ignore").strip()
    if not text:
        raise RDockPipelineError(f"OpenBabel did not produce SMILES for reference ligand {reference_sdf}")
    first = text.splitlines()[0].strip()
    parts = first.split()
    if not parts:
        raise RDockPipelineError(f"Could not parse OpenBabel SMILES output for {reference_sdf}: {first!r}")
    return parts[0]


def _reference_ligand_sdf_block(reference_sdf: Path) -> str:
    text = reference_sdf.read_text(encoding="utf-8", errors="ignore")
    first = text.split("$$$$", 1)[0].strip()
    if not first:
        raise RDockPipelineError(f"Reference ligand SDF is empty or unreadable: {reference_sdf}")
    return first + "\n$$$$\n"


def _canonicalize_smiles_for_filter(smiles: str) -> str:
    return smiles.strip()


def _looks_organic_smiles(smiles: str) -> bool:
    text = _canonicalize_smiles_for_filter(smiles)
    if not text or "." in text:
        return False
    return "C" in text or "c" in text


def _parse_threshold_ladder(text: str) -> list[int]:
    try:
        values = [int(part.strip()) for part in text.split(",") if part.strip()]
    except Exception as exc:
        raise RDockPipelineError(f"Invalid --pubchem-threshold-ladder value {text!r}: {exc}") from exc
    if not values:
        raise RDockPipelineError("Empty --pubchem-threshold-ladder")
    return values


def _resolve_threshold_ladder(args: argparse.Namespace) -> list[int]:
    ladder_override = str(getattr(args, "similarity_thresholds", "") or "").strip()
    if ladder_override:
        return _parse_threshold_ladder(ladder_override)
    start = getattr(args, "pubchem_threshold_start", None)
    stop = getattr(args, "pubchem_threshold_stop", None)
    step = int(getattr(args, "pubchem_threshold_step", 1) or 1)
    if start is not None or stop is not None:
        start_value = int(99 if start is None else start)
        stop_value = int(70 if stop is None else stop)
        if step <= 0:
            raise RDockPipelineError("--pubchem-threshold-step must be a positive integer")
        if start_value < stop_value:
            raise RDockPipelineError(
                f"Invalid PubChem threshold range: start {start_value} is lower than stop {stop_value}. "
                "Use a descending range such as --pubchem-threshold-start 99 --pubchem-threshold-stop 80."
            )
        return list(range(start_value, stop_value - 1, -step))
    return _parse_threshold_ladder(args.pubchem_threshold_ladder)


def _pubchem_similarity_cids_from_sdf(reference_sdf_block: str, threshold: int, max_records: int, timeout: int) -> list[int]:
    url = (
        "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/fastsimilarity_2d/"
        f"sdf/cids/JSON?Threshold={int(threshold)}&MaxRecords={int(max_records)}&MaxSeconds={int(timeout)}"
    )
    payload = _http_post_form_json(url, {"sdf": reference_sdf_block}, timeout)
    info = payload.get("IdentifierList", {}) if isinstance(payload, dict) else {}
    cids = info.get("CID", []) if isinstance(info, dict) else []
    return [int(cid) for cid in cids]


def _pubchem_identity_cids_from_sdf(reference_sdf_block: str, timeout: int) -> list[int]:
    url = (
        "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/fastidentity/"
        f"sdf/cids/JSON?identity_type=same_connectivity&MaxRecords=10&MaxSeconds={int(timeout)}"
    )
    payload = _http_post_form_json(url, {"sdf": reference_sdf_block}, timeout)
    info = payload.get("IdentifierList", {}) if isinstance(payload, dict) else {}
    cids = info.get("CID", []) if isinstance(info, dict) else []
    return [int(cid) for cid in cids]


def _pubchem_identity_cids_from_smiles(reference_smiles: str, timeout: int) -> list[int]:
    encoded = urllib.parse.quote(reference_smiles, safe="")
    identity_url = (
        "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/fastidentity/"
        f"smiles/{encoded}/cids/JSON?identity_type=same_connectivity&MaxRecords=10&MaxSeconds={int(timeout)}"
    )
    identity_payload = _http_json(identity_url, timeout)
    identity_info = identity_payload.get("IdentifierList", {}) if isinstance(identity_payload, dict) else {}
    identity_cids = identity_info.get("CID", []) if isinstance(identity_info, dict) else []
    return [int(cid) for cid in identity_cids]


def _pubchem_reference_cid(
    reference_smiles: str,
    timeout: int,
    reference_sdf_block: str | None = None,
) -> int | None:
    try:
        identity_cids = _pubchem_identity_cids_from_smiles(reference_smiles, timeout)
        if identity_cids:
            return int(identity_cids[0])
    except Exception:  # pragma: no cover - network-dependent
        pass
    if reference_sdf_block:
        try:
            identity_cids = _pubchem_identity_cids_from_sdf(reference_sdf_block, timeout)
            if identity_cids:
                return int(identity_cids[0])
        except Exception:  # pragma: no cover - network-dependent
            pass
    return None


def _pubchem_similarity_cids_from_cid(reference_cid: int, threshold: int, max_records: int, timeout: int) -> list[int]:
    cid_url = (
        "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/fastsimilarity_2d/"
        f"cid/{int(reference_cid)}/cids/JSON?Threshold={int(threshold)}&MaxRecords={int(max_records)}&MaxSeconds={int(timeout)}"
    )
    cid_payload = _http_json(cid_url, timeout)
    cid_info = cid_payload.get("IdentifierList", {}) if isinstance(cid_payload, dict) else {}
    cid_hits = cid_info.get("CID", []) if isinstance(cid_info, dict) else []
    return [int(cid) for cid in cid_hits]


def _pubchem_similarity_cids(
    reference_smiles: str,
    threshold: int,
    max_records: int,
    timeout: int,
    reference_sdf_block: str | None = None,
    reference_cid: int | None = None,
) -> list[int]:
    errors: list[str] = []
    if reference_cid is not None:
        try:
            parsed = _pubchem_similarity_cids_from_cid(reference_cid, threshold, max_records, timeout)
            if parsed:
                return parsed
        except Exception as exc:  # pragma: no cover - network-dependent
            errors.append(f"cid_fastsim:{exc}")
    if reference_sdf_block:
        try:
            parsed = _pubchem_similarity_cids_from_sdf(reference_sdf_block, threshold, max_records, timeout)
            if parsed:
                return parsed
        except Exception as exc:  # pragma: no cover - network-dependent
            errors.append(f"sdf_fastsim:{exc}")
        try:
            identity_cids = _pubchem_identity_cids_from_sdf(reference_sdf_block, timeout)
        except Exception as exc:  # pragma: no cover - network-dependent
            identity_cids = []
            errors.append(f"sdf_identity:{exc}")
        if identity_cids:
            ref_cid = int(identity_cids[0])
            cid_url = (
                "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/fastsimilarity_2d/"
                f"cid/{ref_cid}/cids/JSON?Threshold={int(threshold)}&MaxRecords={int(max_records)}&MaxSeconds={int(timeout)}"
            )
            try:
                cid_payload = _http_json(cid_url, timeout)
                cid_info = cid_payload.get("IdentifierList", {}) if isinstance(cid_payload, dict) else {}
                cid_hits = cid_info.get("CID", []) if isinstance(cid_info, dict) else []
                parsed = [int(cid) for cid in cid_hits]
                if parsed:
                    return parsed
            except Exception as exc:  # pragma: no cover - network-dependent
                errors.append(f"cid_fastsim_from_sdf:{exc}")
    encoded = urllib.parse.quote(reference_smiles, safe="")
    url = (
        "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/fastsimilarity_2d/"
        f"smiles/{encoded}/cids/JSON?Threshold={int(threshold)}&MaxRecords={int(max_records)}&MaxSeconds={int(timeout)}"
    )
    try:
        payload = _http_json(url, timeout)
        info = payload.get("IdentifierList", {}) if isinstance(payload, dict) else {}
        cids = info.get("CID", []) if isinstance(info, dict) else []
        parsed = [int(cid) for cid in cids]
        if parsed:
            return parsed
    except Exception as exc:  # pragma: no cover - network-dependent
        errors.append(f"smiles_fastsim:{exc}")
    identity_url = (
        "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/fastidentity/"
        f"smiles/{encoded}/cids/JSON?identity_type=same_connectivity&MaxRecords=10&MaxSeconds={int(timeout)}"
    )
    try:
        identity_payload = _http_json(identity_url, timeout)
        identity_info = identity_payload.get("IdentifierList", {}) if isinstance(identity_payload, dict) else {}
        identity_cids = identity_info.get("CID", []) if isinstance(identity_info, dict) else []
    except Exception as exc:  # pragma: no cover - network-dependent
        identity_cids = []
        errors.append(f"smiles_identity:{exc}")
    if not identity_cids:
        return []
    try:
        return _pubchem_similarity_cids_from_cid(int(identity_cids[0]), threshold, max_records, timeout)
    except Exception:  # pragma: no cover - network-dependent
        return []


def _pubchem_name_search_cids(query: str, max_records: int, timeout: int) -> list[int]:
    encoded = urllib.parse.quote(query.strip(), safe="")
    url = f"https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/name/{encoded}/cids/JSON?MaxRecords={int(max_records)}"
    payload = _http_json(url, timeout)
    info = payload.get("IdentifierList", {}) if isinstance(payload, dict) else {}
    cids = info.get("CID", []) if isinstance(info, dict) else []
    return [int(cid) for cid in cids]


def _default_pubchem_workers() -> int:
    cpu_count = os.cpu_count() or 4
    return max(2, min(8, cpu_count // 2 or 1))


def _pubchem_fetch_properties_chunk(chunk: list[int], timeout: int) -> list[dict[str, object]]:
    cid_text = ",".join(str(cid) for cid in chunk)
    url = (
        "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/cid/"
        f"{cid_text}/property/SMILES,ConnectivitySMILES,IUPACName,Title,MolecularFormula,MolecularWeight,"
        "XLogP,TPSA,HBondDonorCount,HBondAcceptorCount,RotatableBondCount,HeavyAtomCount/JSON"
    )
    payload = _http_json(url, timeout)
    table = payload.get("PropertyTable", {}) if isinstance(payload, dict) else {}
    props = table.get("Properties", []) if isinstance(table, dict) else []
    return [item for item in props if isinstance(item, dict)]


def _pubchem_fetch_properties_chunk_resilient(
    chunk: list[int],
    timeout: int,
    out: Path | None = None,
    progress_prefix: str = "",
    chunk_label: str = "",
) -> list[dict[str, object]]:
    if not chunk:
        return []
    try:
        return _pubchem_fetch_properties_chunk(chunk, timeout)
    except Exception as exc:  # pragma: no cover - network-dependent
        if len(chunk) <= 1:
            if out is not None:
                _progress_log(
                    out,
                    f"{progress_prefix} property fetch failed for single CID chunk {chunk_label}: {exc}",
                )
            return []
        mid = max(1, len(chunk) // 2)
        left = chunk[:mid]
        right = chunk[mid:]
        if out is not None:
            _progress_log(
                out,
                f"{progress_prefix} property chunk {chunk_label} failed ({exc}); splitting {len(chunk)} CIDs into {len(left)} + {len(right)}",
            )
        left_rows = _pubchem_fetch_properties_chunk_resilient(
            left,
            timeout,
            out=out,
            progress_prefix=progress_prefix,
            chunk_label=f"{chunk_label}.L",
        )
        right_rows = _pubchem_fetch_properties_chunk_resilient(
            right,
            timeout,
            out=out,
            progress_prefix=progress_prefix,
            chunk_label=f"{chunk_label}.R",
        )
        return left_rows + right_rows


def _pubchem_fetch_properties(
    cids: list[int],
    timeout: int,
    out: Path | None = None,
    progress_prefix: str = "",
    workers: int = 1,
    chunk_size: int = 200,
) -> list[dict[str, object]]:
    if not cids:
        return []
    chunk_size = max(1, int(chunk_size))
    workers = max(1, int(workers))
    chunks = [cids[start : start + chunk_size] for start in range(0, len(cids), chunk_size)]
    rows: list[dict[str, object]] = []
    if workers == 1 or len(chunks) == 1:
        for index, chunk in enumerate(chunks, start=1):
            if out is not None:
                _progress_log(
                    out,
                    f"{progress_prefix} fetching properties chunk {index}/{len(chunks)} ({len(chunk)} CIDs)",
                )
            rows.extend(
                _pubchem_fetch_properties_chunk_resilient(
                    chunk,
                    timeout,
                    out=out,
                    progress_prefix=progress_prefix,
                    chunk_label=f"{index}/{len(chunks)}",
                )
            )
        return rows

    with ThreadPoolExecutor(max_workers=min(workers, len(chunks))) as pool:
        futures = {
            pool.submit(_pubchem_fetch_properties_chunk, chunk, timeout): (index, chunk)
            for index, chunk in enumerate(chunks, start=1)
        }
        for future in as_completed(futures):
            index, chunk = futures[future]
            if out is not None:
                _progress_log(
                    out,
                    f"{progress_prefix} fetching properties chunk {index}/{len(chunks)} ({len(chunk)} CIDs)",
                )
            try:
                rows.extend(future.result())
            except Exception as exc:  # pragma: no cover - network-dependent
                if out is not None:
                    _progress_log(
                        out,
                        f"{progress_prefix} property chunk {index}/{len(chunks)} failed asynchronously ({exc}); retrying with recursive split",
                    )
                rows.extend(
                    _pubchem_fetch_properties_chunk_resilient(
                        chunk,
                        timeout,
                        out=out,
                        progress_prefix=progress_prefix,
                        chunk_label=f"{index}/{len(chunks)}",
                    )
                )
    return rows


def _resolve_pubchem_max_records(
    needed: int,
    record_multiplier: int,
    min_records_per_threshold: int,
    max_records_per_threshold: int,
) -> int:
    candidate = max(int(needed) * int(record_multiplier), int(min_records_per_threshold))
    return max(1, min(candidate, int(max_records_per_threshold)))


def _pubchem_similarity_rows(
    reference_smiles: str,
    reference_sdf_block: str,
    n_ligands: int,
    threshold_ladder: list[int],
    timeout: int,
    allow_partial: bool = False,
    out: Path | None = None,
    property_workers: int = 1,
    property_chunk_size: int = 200,
    record_multiplier: int = 8,
    min_records_per_threshold: int = 1000,
    max_records_per_threshold: int = 5000,
) -> tuple[list[tuple[str, str]], list[dict[str, object]], list[dict[str, object]], str | None]:
    collected_smiles: dict[str, tuple[str, str]] = {}
    metadata: list[dict[str, object]] = []
    diagnostics: list[dict[str, object]] = []
    reference_cid = _pubchem_reference_cid(reference_smiles, timeout, reference_sdf_block=reference_sdf_block)
    if out is not None:
        _progress_log(
            out,
            f"PubChem reference CID resolution: {'resolved to CID ' + str(reference_cid) if reference_cid is not None else 'not resolved; using direct similarity fallbacks'}",
            {
                "mode": "pubchem_similarity",
                "reference_cid": reference_cid,
                "reference_cid_resolved": reference_cid is not None,
            },
        )
    for threshold in threshold_ladder:
        needed = max(0, n_ligands - len(collected_smiles))
        if needed <= 0:
            break
        max_records = _resolve_pubchem_max_records(
            needed,
            record_multiplier=record_multiplier,
            min_records_per_threshold=min_records_per_threshold,
            max_records_per_threshold=max_records_per_threshold,
        )
        if out is not None:
            _progress_log(
                out,
                f"PubChem similarity threshold {threshold}: requesting up to {max_records} records; collected so far {len(collected_smiles)}/{n_ligands}",
                {
                    "mode": "pubchem_similarity",
                    "threshold": threshold,
                    "requested_ligands": n_ligands,
                    "collected_ligands": len(collected_smiles),
                    "needed_ligands": needed,
                    "max_records": max_records,
                },
            )
        similarity_kwargs: dict[str, object] = {
            "max_records": max_records,
            "timeout": timeout,
            "reference_sdf_block": reference_sdf_block,
        }
        if reference_cid is not None:
            similarity_kwargs["reference_cid"] = reference_cid
        try:
            cids = _pubchem_similarity_cids(
                reference_smiles,
                threshold,
                **similarity_kwargs,
            )
        except Exception as exc:  # pragma: no cover - network-dependent
            diagnostics.append(
                {
                    "threshold": threshold,
                    "returned_cids": 0,
                    "accepted_new_smiles": 0,
                    "collected_total": len(collected_smiles),
                    "error": str(exc),
                }
            )
            if out is not None:
                _progress_log(
                    out,
                    f"PubChem similarity threshold {threshold}: failed ({exc}); continuing to next threshold",
                )
            continue
        if out is not None:
            _progress_log(out, f"PubChem similarity threshold {threshold}: received {len(cids)} CIDs")
        try:
            props = _pubchem_fetch_properties(
                cids,
                timeout,
                out=out,
                progress_prefix=f"PubChem threshold {threshold}:",
                workers=property_workers,
                chunk_size=property_chunk_size,
            )
        except Exception as exc:  # pragma: no cover - network-dependent
            diagnostics.append(
                {
                    "threshold": threshold,
                    "returned_cids": len(cids),
                    "accepted_new_smiles": 0,
                    "collected_total": len(collected_smiles),
                    "error": f"property_fetch_failed: {exc}",
                }
            )
            if out is not None:
                _progress_log(
                    out,
                    f"PubChem similarity threshold {threshold}: property fetch failed ({exc}); continuing to next threshold",
                )
            continue
        accepted = 0
        for item in props:
            smiles = str(
                item.get("SMILES")
                or item.get("ConnectivitySMILES")
                or item.get("CanonicalSMILES", "")
            ).strip()
            cid = int(item.get("CID", 0) or 0)
            if not smiles or cid <= 0 or not _looks_organic_smiles(smiles):
                continue
            ligand_id = f"pubchem_cid_{cid}"
            if ligand_id in collected_smiles:
                continue
            collected_smiles[ligand_id] = (smiles, ligand_id)
            metadata.append(
                {
                    "ligand_id": ligand_id,
                    "smiles": smiles,
                    "source": "pubchem_similarity",
                    "pubchem_cid": cid,
                    "pubchem_threshold": threshold,
                    "pubchem_title": item.get("Title", ""),
                    "iupac_name": item.get("IUPACName", ""),
                    "molecular_formula": item.get("MolecularFormula", ""),
                    "molecular_weight": item.get("MolecularWeight", ""),
                    "xlogp": item.get("XLogP", ""),
                    "tpsa": item.get("TPSA", ""),
                    "hbd": item.get("HBondDonorCount", ""),
                    "hba": item.get("HBondAcceptorCount", ""),
                    "rotatable_bonds": item.get("RotatableBondCount", ""),
                    "heavy_atom_count": item.get("HeavyAtomCount", ""),
                }
            )
            accepted += 1
            if len(collected_smiles) >= n_ligands:
                break
        diagnostics.append(
            {
                "threshold": threshold,
                "returned_cids": len(cids),
                "accepted_new_smiles": accepted,
                "collected_total": len(collected_smiles),
            }
        )
        if out is not None:
            _write_partial_pubchem_hits(out, list(collected_smiles.values()), metadata)
            _progress_log(
                out,
                f"PubChem similarity threshold {threshold}: accepted {accepted} new ligands; total {len(collected_smiles)}/{n_ligands}",
                {
                    "mode": "pubchem_similarity",
                    "threshold": threshold,
                    "requested_ligands": n_ligands,
                    "collected_ligands": len(collected_smiles),
                    "accepted_new_smiles": accepted,
                    "returned_cids": len(cids),
                },
            )
            if len(collected_smiles) >= n_ligands:
                _progress_log(
                    out,
                    f"PubChem similarity target reached: collected {len(collected_smiles)}/{n_ligands} ligands",
                    {
                        "mode": "pubchem_similarity",
                        "threshold": threshold,
                        "requested_ligands": n_ligands,
                        "collected_ligands": len(collected_smiles),
                        "target_reached": True,
                    },
                )
    warning: str | None = None
    if len(collected_smiles) < n_ligands:
        warning = (
            f"PubChem similarity search collected {len(collected_smiles)} ligands after thresholds {threshold_ladder}. "
            f"Requested {n_ligands}."
        )
    if len(collected_smiles) < n_ligands and not allow_partial:
        raise RDockPipelineError(
            f"PubChem similarity search collected only {len(collected_smiles)} ligands for reference SMILES after thresholds "
            f"{threshold_ladder}. Requested {n_ligands}. Lower the threshold ladder, request fewer ligands, or provide --smiles-file."
        )
    rows = list(collected_smiles.values())[:n_ligands]
    if out is not None:
        _write_partial_pubchem_hits(out, rows, metadata[:n_ligands])
    return rows, metadata[:n_ligands], diagnostics, warning


def _pubchem_name_rows(
    queries: list[str],
    n_ligands: int,
    timeout: int,
) -> tuple[list[tuple[str, str]], list[dict[str, object]], list[dict[str, object]]]:
    collected_smiles: dict[str, tuple[str, str]] = {}
    metadata: list[dict[str, object]] = []
    diagnostics: list[dict[str, object]] = []
    for query in [item.strip() for item in queries if item.strip()]:
        needed = max(0, n_ligands - len(collected_smiles))
        if needed <= 0:
            break
        cids = _pubchem_name_search_cids(query, max_records=max(needed * 4, needed), timeout=timeout)
        props = _pubchem_fetch_properties(cids, timeout)
        accepted = 0
        for item in props:
            smiles = str(
                item.get("SMILES")
                or item.get("ConnectivitySMILES")
                or item.get("CanonicalSMILES", "")
            ).strip()
            cid = int(item.get("CID", 0) or 0)
            if not smiles or cid <= 0 or not _looks_organic_smiles(smiles):
                continue
            ligand_id = f"pubchem_cid_{cid}"
            if ligand_id in collected_smiles:
                continue
            collected_smiles[ligand_id] = (smiles, ligand_id)
            metadata.append(
                {
                    "ligand_id": ligand_id,
                    "smiles": smiles,
                    "source": "pubchem_compound_search",
                    "pubchem_cid": cid,
                    "pubchem_query": query,
                    "pubchem_title": item.get("Title", ""),
                    "iupac_name": item.get("IUPACName", ""),
                    "molecular_formula": item.get("MolecularFormula", ""),
                    "molecular_weight": item.get("MolecularWeight", ""),
                    "xlogp": item.get("XLogP", ""),
                    "tpsa": item.get("TPSA", ""),
                    "hbd": item.get("HBondDonorCount", ""),
                    "hba": item.get("HBondAcceptorCount", ""),
                    "rotatable_bonds": item.get("RotatableBondCount", ""),
                    "heavy_atom_count": item.get("HeavyAtomCount", ""),
                }
            )
            accepted += 1
            if len(collected_smiles) >= n_ligands:
                break
        diagnostics.append(
            {
                "query": query,
                "returned_cids": len(cids),
                "accepted_new_smiles": accepted,
                "collected_total": len(collected_smiles),
            }
        )
    return list(collected_smiles.values())[:n_ligands], metadata[:n_ligands], diagnostics


def _known_good_entry(pdb_id: str) -> dict[str, object] | None:
    for item in list_known_good_complexes(KNOWN_GOOD):
        if str(item.get("pdb_id", "")).upper() == pdb_id.upper():
            return item
    return None


def _pubchem_diagnostics_payload(
    source_name: str,
    reference_smiles: str,
    threshold_ladder: list[int],
    diagnostics: list[dict[str, object]],
    errors: list[str],
) -> dict[str, object]:
    return {
        "source": source_name,
        "reference_smiles": reference_smiles,
        "thresholds_tried": threshold_ladder,
        "requested_ligands": None,
        "collected_ligands": None,
        "endpoints_tried": [
            "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/fastsimilarity_2d/sdf/cids/JSON (POST form field: sdf=...)",
            "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/fastidentity/sdf/cids/JSON (POST form field: sdf=...)",
            "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/fastsimilarity_2d/smiles/{smiles}/cids/JSON",
            "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/name/{query}/cids/JSON",
            "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/cid/{cids}/property/.../JSON",
        ],
        "diagnostics": diagnostics,
        "errors": errors,
    }


def _write_pubchem_diagnostics(out: Path, payload: dict[str, object]) -> Path:
    target = out / "logs" / "pubchem_diagnostics.json"
    target.parent.mkdir(parents=True, exist_ok=True)
    target.write_text(json.dumps(payload, indent=2), encoding="utf-8")
    qc_target = out / "qc" / "similarity_collection_diagnostics.json"
    qc_target.parent.mkdir(parents=True, exist_ok=True)
    qc_target.write_text(json.dumps(payload, indent=2), encoding="utf-8")
    return target


def _resolve_smiles_source(args: argparse.Namespace) -> Path | None:
    if args.smiles_file:
        return Path(args.smiles_file)
    if args.ligand_source == "bundled_example":
        return DEFAULT_BUNDLED_EXAMPLE
    return None


def _write_csv(path: Path, rows: list[dict[str, object]]) -> Path:
    path.parent.mkdir(parents=True, exist_ok=True)
    fields: list[str] = []
    for row in rows:
        for key in row:
            if key not in fields:
                fields.append(key)
    with path.open("w", encoding="utf-8", newline="") as handle:
        writer = csv.DictWriter(handle, fieldnames=fields)
        writer.writeheader()
        writer.writerows(rows)
    return path


def _mol_from_smiles(smiles: str):
    if not RDKIT_AVAILABLE:
        return None
    try:
        return Chem.MolFromSmiles(str(smiles))
    except Exception:
        return None


def _canonical_smiles(smiles: str) -> str:
    mol = _mol_from_smiles(smiles)
    if mol is None:
        return str(smiles).strip()
    try:
        return str(Chem.MolToSmiles(mol, canonical=True))
    except Exception:
        return str(smiles).strip()


def _scaffold_smiles(mol) -> str:
    if not RDKIT_AVAILABLE or mol is None:
        return ""
    try:
        return str(MurckoScaffold.MurckoScaffoldSmiles(mol=mol) or "")
    except Exception:
        return ""


def _morgan_fp(mol):
    if not RDKIT_AVAILABLE or mol is None:
        return None
    try:
        generator = rdFingerprintGenerator.GetMorganGenerator(radius=2, fpSize=1024)
        return generator.GetFingerprint(mol)
    except Exception:
        try:
            return rdMolDescriptors.GetMorganFingerprintAsBitVect(mol, 2, nBits=1024)
        except Exception:
            return None


def _existing_ligand_metadata_payload(lig_root: Path, expected_total: int) -> list[dict[str, object]] | None:
    metadata_path = lig_root / "ligand_metadata.csv"
    if not metadata_path.exists():
        return None
    try:
        with metadata_path.open(newline="", encoding="utf-8") as handle:
            rows = list(csv.DictReader(handle))
    except Exception:
        return None
    return rows if len(rows) == expected_total else None


def _annotate_near_duplicate_analogs(
    metadata_rows: list[dict[str, object]],
    *,
    tanimoto_threshold: float = 0.55,
    max_heavy_atom_delta: int = 3,
) -> tuple[list[dict[str, object]], dict[str, object]]:
    if not metadata_rows:
        return metadata_rows, {"analog_group_count": 0, "analog_grouped_ligands": 0, "analog_grouping_available": bool(RDKIT_AVAILABLE)}
    enriched: list[dict[str, object]] = []
    representatives_by_scaffold: dict[str, list[int]] = {}
    groups: dict[str, list[int]] = {}
    group_parent: dict[str, str] = {}
    for idx, row in enumerate(metadata_rows):
        item = dict(row)
        ligand_id = str(item.get("ligand_id", f"lig_{idx:05d}"))
        smiles = str(item.get("smiles", "")).strip()
        mol = _mol_from_smiles(smiles)
        canonical = _canonical_smiles(smiles)
        scaffold = _scaffold_smiles(mol) or canonical
        heavy = int(mol.GetNumHeavyAtoms()) if mol is not None else 0
        fp = _morgan_fp(mol)
        group_id = ""
        rule = "singleton"
        for rep_idx in representatives_by_scaffold.get(scaffold, []):
            rep = enriched[rep_idx]
            if canonical and canonical == str(rep.get("canonical_smiles", "")):
                group_id = str(rep["analog_group_id"])
                rule = "canonical_duplicate"
                break
            rep_heavy = int(rep.get("heavy_atom_count_for_grouping", 0) or 0)
            rep_fp = rep.get("_analog_fp")
            if fp is None or rep_fp is None:
                continue
            if abs(heavy - rep_heavy) > max_heavy_atom_delta:
                continue
            try:
                sim = float(DataStructs.TanimotoSimilarity(fp, rep_fp))
            except Exception:
                sim = 0.0
            if sim >= tanimoto_threshold:
                group_id = str(rep["analog_group_id"])
                rule = f"same_scaffold_tanimoto_{sim:.3f}"
                break
        if not group_id:
            group_id = f"analog_group_{len(groups) + 1:06d}"
            representatives_by_scaffold.setdefault(scaffold, []).append(idx)
            group_parent[group_id] = ligand_id
        item["canonical_smiles"] = canonical
        item["scaffold_core"] = scaffold
        item["heavy_atom_count_for_grouping"] = heavy
        item["analog_group_id"] = group_id
        item["analog_parent_id"] = group_parent.get(group_id, ligand_id)
        item["analog_group_rule"] = rule
        item["_analog_fp"] = fp
        groups.setdefault(group_id, []).append(idx)
        enriched.append(item)
    for group_id, indices in groups.items():
        size = len(indices)
        parent_id = str(enriched[indices[0]].get("analog_parent_id", enriched[indices[0]].get("ligand_id", "")))
        for variant_idx, row_idx in enumerate(indices, start=1):
            enriched[row_idx]["analog_group_size"] = size
            enriched[row_idx]["analog_variant_index"] = variant_idx
            enriched[row_idx]["analog_group_weight"] = 1.0 / max(1, size)
            enriched[row_idx]["analog_parent_id"] = parent_id
            enriched[row_idx].pop("_analog_fp", None)
    summary = {
        "analog_grouping_available": bool(RDKIT_AVAILABLE),
        "analog_group_count": len(groups),
        "analog_grouped_ligands": sum(len(indices) for indices in groups.values() if len(indices) > 1),
        "analog_largest_group_size": max((len(indices) for indices in groups.values()), default=0),
        "analog_tanimoto_threshold": float(tanimoto_threshold),
        "analog_max_heavy_atom_delta": int(max_heavy_atom_delta),
    }
    return enriched, summary


def _dataset_uniqueness_summary(metadata_rows: list[dict[str, object]]) -> dict[str, object]:
    seen_smiles: set[str] = set()
    total = 0
    for row in metadata_rows:
        smiles = str(row.get("smiles", "")).strip()
        if not smiles:
            continue
        total += 1
        seen_smiles.add(smiles)
    duplicate_count = max(0, total - len(seen_smiles))
    duplicate_fraction = (duplicate_count / total) if total else 0.0
    warnings: list[str] = []
    if duplicate_fraction > 0.01:
        warnings.append(
            f"duplicate parent fraction is {duplicate_fraction:.4f} (>1%); inspect ligand source and deduplication settings before scientific benchmarking"
        )
    return {
        "n_input_smiles": total,
        "n_unique_parent_ligands": len(seen_smiles),
        "duplicate_parent_count": duplicate_count,
        "duplicate_parent_fraction": duplicate_fraction,
        "warnings": warnings,
        "deduplication_method": "exact_smiles",
    }


def _infer_existing_dataset_manifest(out: Path, prepared_ligands: int, warnings: list[str] | None = None) -> dict[str, object]:
    warnings = list(warnings or [])
    raw_dir = out / "raw"
    pdb_candidates = sorted(list(raw_dir.glob("*.pdb")) + list(raw_dir.glob("*.cif")) + list(raw_dir.glob("*.mmcif")))
    pdb_id = pdb_candidates[0].stem[:4].upper() if pdb_candidates else ""
    resolved = resolve_known_good_defaults(KNOWN_GOOD, pdb_id or "UNKN", None, None, None)
    ligand_source = "smiles_file"
    reference_smiles = ""
    pubchem_payload: dict[str, object] = {}
    pubchem_path = out / "logs" / "pubchem_diagnostics.json"
    if pubchem_path.exists():
        try:
            pubchem_payload = json.loads(pubchem_path.read_text(encoding="utf-8"))
        except Exception:
            pubchem_payload = {}
        ligand_source = str(pubchem_payload.get("source") or "pubchem_similarity")
        reference_smiles = str(pubchem_payload.get("reference_smiles") or "")
    else:
        ref_smiles_log = out / "logs" / "reference_ligand_to_smiles.stdout.log"
        if ref_smiles_log.exists():
            first = ref_smiles_log.read_text(encoding="utf-8", errors="ignore").splitlines()
            if first:
                reference_smiles = first[0].split()[0].strip()
    manifest = {
        "pdb_id": resolved["pdb_id"] or pdb_id,
        "receptor_chain": resolved["receptor_chain"],
        "reference_ligand_resname": resolved["reference_ligand_resname"],
        "reference_ligand_chain": resolved["reference_ligand_chain"],
        "n_ligands_requested": len(_read_all_smiles(require_file(out / "ligands" / "all_ligands.smi", "all_ligands.smi"))),
        "ligands_prepared": int(prepared_ligands),
        "paths": {
            "target_mol2": str(out / "target" / "target.mol2"),
            "reference_ligand_sdf": str(out / "target" / "reference_ligand.sdf"),
            "all_ligands_sdf": str(out / "ligands" / "all_ligands.sdf"),
            "all_ligands_smi": str(out / "ligands" / "all_ligands.smi"),
            "rdock_prm_dir": str(out / "target" / "rdock_prm"),
            "target_config_yaml": str(out / "target" / "rdock_prm" / "target_config.yaml"),
        },
        "versions": {
            "obabel": probe_version(require_executable("obabel")) if shutil.which("obabel") else "",
            "rbdock": probe_version(require_executable("rbdock")) if shutil.which("rbdock") else "",
            "rbcavity": probe_version(require_executable("rbcavity")) if shutil.which("rbcavity") else "",
        },
        "ligand_source": ligand_source,
        "reference_ligand_smiles": reference_smiles,
        "pubchem_threshold_ladder": pubchem_payload.get("thresholds_tried", []) if pubchem_payload else [],
        "pubchem_diagnostics": pubchem_payload,
        "rbcavity_status": "success" if (out / "target" / "rdock_prm").exists() else "unknown",
        "warnings": warnings,
        "available_hetero_ligands": [],
        "auto_reference_ligand": False,
        "auto_reference_candidates": [],
        "auto_reference_selected": {},
    }
    return manifest


def _ensure_dataset_metadata_files(out: Path, prepared_ligands: int, warnings: list[str] | None = None) -> None:
    manifest_path = out / "dataset_manifest.json"
    if manifest_path.exists():
        manifest = json.loads(manifest_path.read_text(encoding="utf-8"))
        manifest["ligands_prepared"] = int(prepared_ligands)
        manifest.setdefault("paths", {})
        manifest["paths"]["all_ligands_sdf"] = str(out / "ligands" / "all_ligands.sdf")
        manifest["paths"]["all_ligands_smi"] = str(out / "ligands" / "all_ligands.smi")
        if warnings:
            manifest["warnings"] = list(dict.fromkeys(list(manifest.get("warnings", [])) + list(warnings)))
        create_dataset_manifest(out, manifest)
    else:
        synth_warnings = ["dataset_manifest.json was missing and was reconstructed from existing dataset files"]
        if warnings:
            synth_warnings.extend(warnings)
        create_dataset_manifest(out, _infer_existing_dataset_manifest(out, prepared_ligands, warnings=synth_warnings))

    report_path = out / "qc" / "preparation_report.md"
    if not report_path.exists():
        report_lines = [
            f"# Dataset Preparation Report: {out.name}",
            "",
            "- report_status: `reconstructed`",
            f"- ligands_prepared: `{prepared_ligands}`",
            f"- target_mol2: `{out / 'target' / 'target.mol2'}`",
            f"- reference_ligand_sdf: `{out / 'target' / 'reference_ligand.sdf'}`",
            f"- all_ligands_sdf: `{out / 'ligands' / 'all_ligands.sdf'}`",
            f"- all_ligands_smi: `{out / 'ligands' / 'all_ligands.smi'}`",
        ]
        if warnings:
            report_lines.extend(["", "## Warnings", *[f"- {item}" for item in warnings]])
        report_path.parent.mkdir(parents=True, exist_ok=True)
        report_path.write_text("\n".join(report_lines) + "\n", encoding="utf-8")


def _list_known_good() -> int:
    complexes = list_known_good_complexes(KNOWN_GOOD)
    print(json.dumps(complexes, indent=2))
    return 0


def _list_hetero_for_pdb(args: argparse.Namespace) -> int:
    temp_root = Path(args.out) if args.out else ROOT / ".tmp_list_hetero"
    _mkdir_or_fail(temp_root)
    pdb_path = download_pdb_structure(str(args.pdb_id), temp_root / "raw", force=False)
    payload = {
        "pdb_id": str(args.pdb_id).upper(),
        "hetero_ligands": list_hetero_ligands(pdb_path, min_reference_ligand_atoms=args.min_reference_ligand_atoms),
    }
    print(json.dumps(payload, indent=2))
    return 0


def _validate_only(dataset_dir: Path) -> int:
    print(json.dumps(validate_dataset_dir(dataset_dir, check_rdock_tools=False), indent=2))
    return 0


def _plan(args: argparse.Namespace, resolved: dict[str, str], smiles_file: Path | None) -> dict[str, object]:
    if smiles_file is None:
        ligand_count = None
        smiles_repr = ""
    elif not smiles_file.exists():
        ligand_count = 0
        smiles_repr = str(smiles_file)
    else:
        ligand_count = len(_read_smiles(smiles_file, args.n_ligands))
        smiles_repr = str(smiles_file)
    missing_tools = [tool for tool in ("obabel", "rbcavity", "rbdock") if shutil.which(tool) is None]
    return {
        "pdb_id": resolved["pdb_id"],
        "receptor_chain": resolved["receptor_chain"],
        "reference_ligand_resname": resolved["reference_ligand_resname"],
        "reference_ligand_chain": resolved["reference_ligand_chain"],
        "n_ligands_requested": args.n_ligands,
        "ligand_source": args.ligand_source,
        "auto_reference_ligand": bool(args.auto_reference_ligand),
        "min_reference_ligand_atoms": int(args.min_reference_ligand_atoms),
        "smiles_file": smiles_repr,
        "detected_smiles_rows": ligand_count,
        "out": str(args.out),
        "missing_dependencies": missing_tools,
        "commands": [
            "download PDB structure if absent",
            "extract receptor chain and reference ligand from PDB",
            "convert reference ligand and receptor with OpenBabel",
            "resolve ligand source via smiles file or PubChem when --smiles-file is not provided",
            "prepare target with rbcavity",
            "convert SMILES to 3D SDF",
            "write dataset_manifest.json and qc/preparation_report.md",
        ],
    }


def _reference_selection_error(
    exc: RDockPipelineError,
    pdb_id: str,
    receptor_chain: str,
    out: Path,
) -> RDockPipelineError:
    known = _known_good_entry(pdb_id)
    lines = [str(exc)]
    if known:
        lines.append(
            "Known-good suggestion: "
            f"--receptor-chain {known.get('receptor_chain')} "
            f"--reference-ligand-resname {known.get('reference_ligand_resname')} "
            f"--reference-ligand-chain {known.get('reference_ligand_chain')}"
        )
    lines.append(
        "Recommended correction: "
        f"python scripts/prepare_pdb_ligand_dataset.py --pdb-id {pdb_id} --receptor-chain {receptor_chain} "
        f"--auto-reference-ligand --n-ligands 1000 --ligand-source smiles_file "
        f"--smiles-file data/examples/example_smiles_1000.smi --out {out} --force"
    )
    lines.append(
        "You can inspect candidate hetero ligands with: "
        f"python scripts/prepare_pdb_ligand_dataset.py --list-hetero --pdb-id {pdb_id}"
    )
    return RDockPipelineError("\n".join(lines))


def _select_reference_ligand(args: argparse.Namespace, resolved: dict[str, str], pdb_path: Path) -> tuple[dict[str, str], list[dict[str, object]], dict[str, object] | None]:
    if args.auto_reference_ligand or not resolved["reference_ligand_resname"]:
        selected, candidates = auto_detect_reference_ligand(
            pdb_path,
            resolved["receptor_chain"],
            min_reference_ligand_atoms=args.min_reference_ligand_atoms,
        )
        resolved = dict(resolved)
        resolved["reference_ligand_resname"] = str(selected["resname"])
        resolved["reference_ligand_chain"] = str(selected["chain"])
        return resolved, candidates, selected
    return resolved, [], None


def _resolve_ligand_rows(
    args: argparse.Namespace,
    resolved: dict[str, str],
    runner: CommandRunner,
    reference_sdf: Path,
    out: Path,
) -> tuple[list[tuple[str, str]], list[dict[str, object]], str, dict[str, object] | None, str, list[str]]:
    smiles_file = _resolve_smiles_source(args)
    source = args.ligand_source
    if args.smiles_file:
        source = "smiles_file"
    if source == "smiles_file" and smiles_file is None:
        raise RDockPipelineError("`--ligand-source smiles_file` requires --smiles-file /path/to/library.smi.")
    if source in {"pubchem_random_compounds", "zinc_file"}:
        raise RDockPipelineError(
            f"Ligand source `{source}` is not enabled in this portable bundle. "
            "Use --ligand-source smiles_file --smiles-file /path/to/real_library.smi."
        )
    if source == "bundled_example":
        if args.n_ligands > 50:
            raise RDockPipelineError(
                f"`bundled_example` contains only 50 ligands. Requested {args.n_ligands}. "
                f"Use --ligand-source smiles_file --smiles-file data/examples/example_smiles_1000.smi or a larger real library."
            )
    if args.n_ligands > 5000 and source not in {"smiles_file", "pubchem_similarity"}:
        raise RDockPipelineError(
            f"Requested {args.n_ligands} ligands with source `{source}`. For libraries larger than 5000 ligands, "
            "provide a real large library via --ligand-source smiles_file --smiles-file /path/to/library.smi."
        )
    if smiles_file is not None:
        rows = _read_smiles(smiles_file, args.n_ligands)
        if len(rows) < args.n_ligands:
            raise RDockPipelineError(
                f"Requested {args.n_ligands} ligands but {smiles_file} contains only {len(rows)} usable rows. "
                f"Provide a larger --smiles-file. Example: --smiles-file /data/libraries/real_50k_library.smi"
            )
        metadata_rows = [{"ligand_id": ligand_id, "smiles": smiles, "source": "smiles_file"} for smiles, ligand_id in rows]
        return rows, metadata_rows, "", None, source, []

    reference_smiles = _reference_ligand_to_smiles(runner, reference_sdf, out)
    reference_sdf_block = _reference_ligand_sdf_block(reference_sdf)
    threshold_ladder = _resolve_threshold_ladder(args)
    diagnostics_errors: list[str] = []
    combined_diagnostics: list[dict[str, object]] = []
    warnings: list[str] = []

    if source in {"auto", "pubchem_similarity"}:
        try:
            rows, metadata_rows, diagnostics, warning = _pubchem_similarity_rows(
                reference_smiles,
                reference_sdf_block,
                args.n_ligands,
                threshold_ladder,
                timeout=args.pubchem_timeout,
                allow_partial=bool(getattr(args, "allow_partial_ligand_set", False)),
                out=out,
                property_workers=int(getattr(args, "pubchem_property_workers", _default_pubchem_workers())),
                property_chunk_size=int(getattr(args, "pubchem_property_chunk_size", 200)),
                record_multiplier=int(getattr(args, "pubchem_record_multiplier", 8)),
                min_records_per_threshold=int(getattr(args, "pubchem_min_records_per_threshold", 1000)),
                max_records_per_threshold=int(getattr(args, "pubchem_max_records_per_threshold", 5000)),
            )
            combined_diagnostics.extend(diagnostics)
            payload = _pubchem_diagnostics_payload("pubchem_similarity", reference_smiles, threshold_ladder, combined_diagnostics, diagnostics_errors)
            payload["requested_ligands"] = args.n_ligands
            payload["collected_ligands"] = len(rows)
            if warning:
                warnings.append(warning)
            return rows, metadata_rows, reference_smiles, payload, "pubchem_similarity", warnings
        except RDockPipelineError as exc:
            diagnostics_errors.append(str(exc))

    if source in {"auto", "pubchem_compound_search"}:
        queries = [
            resolved["reference_ligand_resname"],
            f"{resolved['pdb_id']} {resolved['reference_ligand_resname']}",
            str((_known_good_entry(resolved["pdb_id"]) or {}).get("target_name", "")),
        ]
        rows, metadata_rows, diagnostics = _pubchem_name_rows(queries, args.n_ligands, timeout=args.pubchem_timeout)
        combined_diagnostics.extend(diagnostics)
        if len(rows) >= args.n_ligands:
            payload = _pubchem_diagnostics_payload("pubchem_compound_search", reference_smiles, threshold_ladder, combined_diagnostics, diagnostics_errors)
            payload["requested_ligands"] = args.n_ligands
            payload["collected_ligands"] = len(rows)
            return rows, metadata_rows, reference_smiles, payload, "pubchem_compound_search", warnings
        diagnostics_errors.append(
            f"PubChem compound search collected only {len(rows)} ligands for queries {queries}. Requested {args.n_ligands}."
        )
        if rows and bool(getattr(args, "allow_partial_ligand_set", False)):
            payload = _pubchem_diagnostics_payload("pubchem_compound_search", reference_smiles, threshold_ladder, combined_diagnostics, diagnostics_errors)
            payload["requested_ligands"] = args.n_ligands
            payload["collected_ligands"] = len(rows)
            warnings.append(
                f"PubChem compound search collected {len(rows)} ligands for queries {queries}. Requested {args.n_ligands}."
            )
            return rows, metadata_rows, reference_smiles, payload, "pubchem_compound_search", warnings

    payload = _pubchem_diagnostics_payload(source, reference_smiles, threshold_ladder, combined_diagnostics, diagnostics_errors)
    payload["requested_ligands"] = args.n_ligands
    payload["collected_ligands"] = 0
    _write_pubchem_diagnostics(out, payload)
    raise RDockPipelineError(
        f"Could not collect {args.n_ligands} ligands from source `{source}` for {resolved['pdb_id']}. "
        f"Diagnostics were written to {out / 'logs' / 'pubchem_diagnostics.json'}. "
        "Use --ligand-source smiles_file --smiles-file /path/to/real_library.smi for a stable large-library workflow."
    )


def run(args: argparse.Namespace) -> int:
    if args.list_known_good:
        return _list_known_good()
    if args.list_hetero:
        return _list_hetero_for_pdb(args)
    if args.validate_only:
        return _validate_only(Path(args.out))
    if args.prepare_ligands_only:
        out = Path(args.out)
        _mkdir_or_fail(out)
        for name in ("ligands", "logs", "qc"):
            _mkdir_or_fail(out / name)
        runner = CommandRunner(out / "logs" / "commands.log")
        result = _prepare_ligands_from_existing_smi(
            out,
            runner,
            batch_size=int(args.ligand_batch_size),
            jobs=args.ligand_jobs,
            cpu_fraction=float(args.ligand_cpu_fraction),
            force_rebuild=bool(args.force),
        )
        metadata_warnings: list[str] = []
        if not (out / "dataset_manifest.json").exists():
            metadata_warnings.append("dataset_manifest.json was missing before ligand resume")
        if not (out / "qc" / "preparation_report.md").exists():
            metadata_warnings.append("qc/preparation_report.md was missing before ligand resume")
        _ensure_dataset_metadata_files(out, int(result["prepared_ligands"]), warnings=metadata_warnings)
        print(json.dumps(result, indent=2))
        return 0

    resolved = resolve_known_good_defaults(
        KNOWN_GOOD,
        args.pdb_id,
        args.receptor_chain,
        args.reference_ligand_resname,
        args.reference_ligand_chain,
    )
    if not resolved["receptor_chain"] or (not resolved["reference_ligand_resname"] and not args.auto_reference_ligand):
        raise RDockPipelineError(
            "Missing receptor chain or reference ligand resname. Provide them explicitly, use --auto-reference-ligand, or use a known-good PDB entry."
        )
    smiles_file = _resolve_smiles_source(args)
    plan = _plan(args, resolved, smiles_file)
    plan["ligand_source"] = args.ligand_source
    plan["pubchem_threshold_ladder"] = _resolve_threshold_ladder(args)
    plan["uses_pubchem_similarity"] = smiles_file is None
    out = Path(args.out)
    _mkdir_or_fail(out)
    if args.dry_run:
        (out / "dataset_plan.json").write_text(json.dumps(plan, indent=2), encoding="utf-8")
        print(json.dumps(plan, indent=2))
        return 0

    if out.exists() and args.force:
        shutil.rmtree(out)
        _mkdir_or_fail(out)
    for name in ("raw", "target", "ligands", "logs", "qc"):
        _mkdir_or_fail(out / name)
    runner = CommandRunner(out / "logs" / "commands.log")
    pdb_path = download_pdb_structure(resolved["pdb_id"], out / "raw", force=args.force)
    resolved, auto_candidates, auto_selected = _select_reference_ligand(args, resolved, pdb_path)
    (out / "raw" / "download_metadata.json").write_text(
        json.dumps({"pdb_id": resolved["pdb_id"], "source": str(pdb_path), "status": "downloaded_or_reused"}, indent=2),
        encoding="utf-8",
    )
    try:
        receptor_pdb, ligand_pdb, hetero = extract_receptor_and_reference_ligand(
            pdb_path,
            resolved["receptor_chain"],
            resolved["reference_ligand_resname"],
            resolved["reference_ligand_chain"],
            out / "target",
            min_reference_ligand_atoms=args.min_reference_ligand_atoms,
        )
    except RDockPipelineError as exc:
        raise _reference_selection_error(exc, resolved["pdb_id"], resolved["receptor_chain"], out) from exc
    reference_raw_sdf = out / "target" / "reference_ligand_raw.sdf"
    reference_sdf = out / "target" / "reference_ligand.sdf"
    _obabel_convert(runner, "reference_ligand_raw_to_sdf", ligand_pdb, reference_raw_sdf, [], out)
    shutil.copy2(reference_raw_sdf, reference_sdf)
    rows, metadata_rows, reference_smiles, pubchem_payload, ligand_source_used, ligand_warnings = _resolve_ligand_rows(args, resolved, runner, reference_sdf, out)
    if pubchem_payload is not None:
        _write_pubchem_diagnostics(out, pubchem_payload)
    if ligand_warnings:
        for warning in ligand_warnings:
            print(f"WARNING: {warning}", file=sys.stderr)
    if not rows:
        diagnostics_path = out / "logs" / "pubchem_diagnostics.json"
        raise RDockPipelineError(
            f"PubChem returned zero usable ligands for {resolved['pdb_id']} with source `{ligand_source_used}` and thresholds "
            f"{_resolve_threshold_ladder(args)}. No docking dataset can be created from zero ligands. "
            f"See diagnostics: {diagnostics_path}. "
            "Lower --identity-threshold-stop, try a different reference complex, or provide --smiles-file with a real library."
        )
    metadata_rows, analog_summary = _annotate_near_duplicate_analogs(metadata_rows)
    metadata_by_id = {str(row.get("ligand_id", "")): row for row in metadata_rows}
    rows = [(smiles, ligand_id) for smiles, ligand_id in rows if str(ligand_id) in metadata_by_id]

    target_prm_dir = out / "target" / "rdock_prm"
    prep = prepare_dataset_target_with_rdock(receptor_pdb, reference_sdf, target_prm_dir, jobs="auto", cpu_fraction=0.85)
    target_mol2 = require_file(target_prm_dir / "receptor.mol2", "prepared receptor mol2")
    shutil.copy2(target_mol2, out / "target" / "target.mol2")

    smi_path = _write_smi(rows, out / "ligands" / "all_ligands.smi")
    _write_csv(out / "ligands" / "ligand_metadata.csv", metadata_rows)
    _write_csv(out / "ligands" / "invalid_ligands.csv", [])
    if args.stop_after_ligand_collection:
        collection_summary = {
            "status": "ligand_collection_complete",
            "pdb_id": resolved["pdb_id"],
            "ligands_requested": args.n_ligands,
            "ligands_collected": len(rows),
            "reference_ligand_smiles": reference_smiles,
            "ligand_source": ligand_source_used,
            "all_ligands_smi": str(smi_path),
            "ligand_metadata_csv": str(out / "ligands" / "ligand_metadata.csv"),
            "analog_grouping": analog_summary,
            "pubchem_diagnostics_json": str(out / "logs" / "pubchem_diagnostics.json"),
            "next_stage": "openbabel_ligand_preparation",
        }
        (out / "qc" / "collection_summary.json").write_text(json.dumps(collection_summary, indent=2), encoding="utf-8")
        print(
            f"Collected {len(rows)}/{args.n_ligands} ligands and wrote {smi_path}. "
            "Stopping before OpenBabel ligand preparation as requested.",
            file=sys.stderr,
        )
        print(json.dumps(collection_summary, indent=2))
        return 0
    ligand_prep = _prepare_ligands_from_existing_smi(
        out,
        runner,
        batch_size=int(args.ligand_batch_size),
        jobs=args.ligand_jobs,
        cpu_fraction=float(args.ligand_cpu_fraction),
        force_rebuild=bool(args.force),
        metadata_rows=metadata_rows,
    )
    ligands_sdf = require_file(Path(str(ligand_prep["final_sdf"])), "prepared ligand sdf")

    uniqueness = _dataset_uniqueness_summary(metadata_rows)
    manifest_warnings = list(ligand_warnings) + [str(item) for item in uniqueness.get("warnings", [])]
    manifest = {
        "pdb_id": resolved["pdb_id"],
        "receptor_chain": resolved["receptor_chain"],
        "reference_ligand_resname": resolved["reference_ligand_resname"],
        "reference_ligand_chain": resolved["reference_ligand_chain"],
        "reference_ligand_pubchem_cid": (pubchem_payload or {}).get("reference_pubchem_cid", ""),
        "n_ligands_requested": args.n_ligands,
        "n_collected_raw": len(rows),
        "n_input_smiles": int(uniqueness["n_input_smiles"]),
        "n_unique_parent_ligands": int(uniqueness["n_unique_parent_ligands"]),
        "ligands_prepared": len(rows),
        "n_prepared_ligands": len(rows),
        "n_invalid": 0,
        "deduplication_method": str(uniqueness["deduplication_method"]),
        "source": ligand_source_used,
        "synthetic_expansion": False,
        "synthetic_stress_test_only": False,
        "similarity_thresholds": _resolve_threshold_ladder(args) if pubchem_payload is not None else [],
        "min_similarity": float(getattr(args, "min_similarity", 0.70)),
        "max_similarity": float(getattr(args, "max_similarity", 0.99)),
        "deduplicate_canonical_smiles": _bool_arg(getattr(args, "deduplicate_canonical_smiles", "true"), True),
        "deduplicate_inchikey": _bool_arg(getattr(args, "deduplicate_inchikey", "true"), True),
        "paths": {
            "target_mol2": str(out / "target" / "target.mol2"),
            "reference_ligand_sdf": str(reference_sdf),
            "all_ligands_sdf": str(ligands_sdf),
            "rdock_prm_dir": str(target_prm_dir),
            "target_config_yaml": str(target_prm_dir / "target_config.yaml"),
        },
        "versions": {
            "obabel": probe_version(require_executable("obabel")),
            "rbdock": probe_version(require_executable("rbdock")),
            "rbcavity": probe_version(require_executable("rbcavity")),
        },
        "ligand_source": ligand_source_used,
        "analog_grouping": analog_summary,
        "reference_ligand_smiles": reference_smiles,
        "pubchem_threshold_ladder": _resolve_threshold_ladder(args) if pubchem_payload is not None else [],
        "pubchem_diagnostics": pubchem_payload or {},
        "rbcavity_status": "success",
        "warnings": manifest_warnings,
        "ligand_preparation": ligand_prep,
        "available_hetero_ligands": hetero[:50],
        "auto_reference_ligand": bool(args.auto_reference_ligand),
        "auto_reference_candidates": auto_candidates[:50],
        "auto_reference_selected": auto_selected or {},
        "prepared_target": prep,
        "copied_target_bundle": {path.name: str(path) for path in target_prm_dir.iterdir() if path.is_file()},
    }
    create_dataset_manifest(out, manifest)
    _write_csv(
        out / "qc" / "deduplication_report.tsv",
        [
            {
                "n_input_smiles": int(uniqueness["n_input_smiles"]),
                "n_unique_parent_ligands": int(uniqueness["n_unique_parent_ligands"]),
                "duplicate_parent_fraction": float(uniqueness["duplicate_parent_fraction"]),
                "deduplication_method": str(uniqueness["deduplication_method"]),
                "synthetic_expansion": "false",
            }
        ],
    )
    qc_report = [
        f"# Dataset Preparation Report: {resolved['pdb_id']}",
        "",
        f"- receptor_chain: `{resolved['receptor_chain']}`",
        f"- reference_ligand: `{resolved['reference_ligand_resname']}` chain `{resolved['reference_ligand_chain']}`",
        f"- ligands_prepared: `{len(rows)}`",
        f"- ligands_requested: `{args.n_ligands}`",
        f"- ligand_source: `{ligand_source_used}`",
        f"- n_input_smiles: `{uniqueness['n_input_smiles']}`",
        f"- n_unique_parent_ligands: `{uniqueness['n_unique_parent_ligands']}`",
        f"- duplicate_parent_fraction: `{float(uniqueness['duplicate_parent_fraction']):.4f}`",
        f"- analog_group_count: `{analog_summary['analog_group_count']}`",
        f"- analog_grouped_ligands: `{analog_summary['analog_grouped_ligands']}`",
        f"- analog_largest_group_size: `{analog_summary['analog_largest_group_size']}`",
        f"- deduplication_method: `{uniqueness['deduplication_method']}`",
        "- synthetic_expansion: `false`",
        f"- reference_ligand_smiles: `{reference_smiles}`",
        f"- pubchem_threshold_ladder: `{_resolve_threshold_ladder(args) if pubchem_payload is not None else []}`",
        f"- target_mol2: `{out / 'target' / 'target.mol2'}`",
        f"- reference_ligand_sdf: `{reference_sdf}`",
        f"- all_ligands_sdf: `{ligands_sdf}`",
        f"- target_config_yaml: `{target_prm_dir / 'target_config.yaml'}`",
    ]
    if auto_selected:
        qc_report.extend(
            [
                "",
                "## Auto reference ligand selection",
                f"- selected_resname: `{auto_selected.get('resname', '')}`",
                f"- selected_chain: `{auto_selected.get('chain', '')}`",
                f"- selected_residue_id: `{auto_selected.get('residue_id', '')}`",
                f"- heavy_atom_count: `{auto_selected.get('heavy_atom_count', '')}`",
                f"- min_distance_to_receptor: `{auto_selected.get('min_distance_to_receptor', '')}`",
            ]
        )
    if manifest_warnings:
        qc_report.extend(["", "## Ligand Collection Warnings", *[f"- {warning}" for warning in manifest_warnings]])
    (out / "qc" / "preparation_report.md").write_text("\n".join(qc_report) + "\n", encoding="utf-8")
    print(json.dumps(validate_dataset_dir(out, check_rdock_tools=False), indent=2))
    return 0


def build_parser() -> argparse.ArgumentParser:
    parser = argparse.ArgumentParser(description="Prepare a portable PDB + ligand dataset for rDock production runs.")
    parser.add_argument("--pdb-id")
    parser.add_argument("--receptor-chain")
    parser.add_argument("--reference-ligand-resname")
    parser.add_argument("--reference-ligand-chain")
    parser.add_argument("--auto-reference-ligand", action="store_true")
    parser.add_argument("--min-reference-ligand-atoms", type=int, default=8)
    parser.add_argument("--n-ligands", type=int, default=1000)
    parser.add_argument(
        "--ligand-source",
        default="auto",
        choices=[
            "auto",
            "smiles_file",
            "pubchem_similarity",
            "pubchem_compound_search",
            "pubchem_random_compounds",
            "zinc_file",
            "bundled_example",
        ],
    )
    parser.add_argument("--source", dest="ligand_source")
    parser.add_argument("--smiles-file")
    parser.add_argument("--similarity-thresholds", default="")
    parser.add_argument("--min-similarity", type=float, default=0.70)
    parser.add_argument("--max-similarity", type=float, default=0.99)
    parser.add_argument("--deduplicate-canonical-smiles", default="true")
    parser.add_argument("--deduplicate-inchikey", default="true")
    parser.add_argument("--pubchem-threshold-ladder", default="95,90,85,80,75,70")
    parser.add_argument("--pubchem-threshold-start", "--identity-threshold-start", dest="pubchem_threshold_start", type=int, default=None)
    parser.add_argument("--pubchem-threshold-stop", "--identity-threshold-stop", dest="pubchem_threshold_stop", type=int, default=None)
    parser.add_argument("--pubchem-threshold-step", type=int, default=1)
    parser.add_argument("--pubchem-timeout", type=int, default=60)
    parser.add_argument("--pubchem-property-workers", type=int, default=_default_pubchem_workers())
    parser.add_argument("--pubchem-property-chunk-size", type=int, default=200)
    parser.add_argument("--pubchem-record-multiplier", type=int, default=8)
    parser.add_argument("--pubchem-min-records-per-threshold", type=int, default=1000)
    parser.add_argument("--pubchem-max-records-per-threshold", type=int, default=5000)
    parser.add_argument("--allow-partial-ligand-set", action="store_true")
    parser.add_argument("--prepare-ligands-only", action="store_true")
    parser.add_argument("--stop-after-ligand-collection", action="store_true")
    parser.add_argument("--ligand-batch-size", type=int, default=250)
    parser.add_argument("--ligand-jobs", default=str(_default_ligand_jobs()))
    parser.add_argument("--ligand-cpu-fraction", type=float, default=0.85)
    parser.add_argument("--out")
    parser.add_argument("--ph", type=float, default=7.4)
    parser.add_argument("--seed", type=int, default=42)
    parser.add_argument("--force", action="store_true")
    parser.add_argument("--dry-run", action="store_true")
    parser.add_argument("--list-known-good", action="store_true")
    parser.add_argument("--list-hetero", action="store_true")
    parser.add_argument("--validate-only", action="store_true")
    return parser


def main() -> int:
    parser = build_parser()
    args = parser.parse_args()
    if not args.list_known_good and not args.validate_only and not args.list_hetero and not args.prepare_ligands_only and not args.pdb_id:
        parser.error("--pdb-id is required unless --list-known-good, --list-hetero, or --validate-only is used")
    if not args.list_known_good and not args.list_hetero and not args.out:
        parser.error("--out is required unless --list-known-good or --list-hetero is used")
    return run(args)


if __name__ == "__main__":
    raise SystemExit(main())