checkpoint: real-scale 数据链 + 统计引擎 + ABLUP 稀疏性能修复

init 后首次落盘，累计工作： - 数据侧：simulate_breeding_data real-scale 3.5万树重建 + 分层观测 - 引擎侧：ABLUP 稀疏生产档 EM-REML（blup 1.6.0/1.7.0）；性能倒挂修复—— 系谱闭包收口为 BFS 可达祖先 + N_EXACT 3000 对齐 N_SUBSAMPLE（选中 200~3000 树不再顶进数小时精确迹尾），A/B 数值等价实证 + 守卫探针 - 前端：观测过滤 + 表单/HTTP 工具链完善 - 文档：业务链/观测梳理 + 引擎实施记录 + 规划对照总表 - gitignore：排除 Temp/调试脚本/一次性验证输出
2026-08-07 08:03:10 +08:00
parent b95053c52c
commit e937fc930c
16 changed files with 2293 additions and 238 deletions
@@ -26,7 +26,8 @@ import numpy as np
 from datetime import date, datetime
 from typing import Any

-from sqlalchemy import String, case, func, or_, select, text, update
+from sqlalchemy import Integer, String, case, cast, false, func, or_, select, text, update
+from sqlalchemy.dialects.postgresql import ARRAY
 from sqlalchemy.ext.asyncio import AsyncSession

 from app.core.base_schema import AuthSchema
@@ -84,6 +85,19 @@ def _now() -> datetime:
    return datetime.now()


+def _id_in(column, ids):
+    """按 IN (SELECT unnest(:arr)) 绑定大 id 列表为单个数组参数。
+
+    asyncpg 单条查询总绑定参数上限 32767，常规 column.in_(ids) 会在
+    bre_tree 等表 id 规模过万时直接崩（IN 子句分块无济于事，因为是全查询
+    计数）。空列表返回恒假条件，避免 SQLAlchemy 渲染出畸形的 IN (NULL)。
+    """
+    ids = list(ids)
+    if not ids:
+        return false()
+    return column.in_(select(func.unnest(cast(ids, ARRAY(Integer)))))
+
+
 # 交配设计（配合力 GCA/SCA 的算法与 ANOVA 自由度随设计而异；与 combining.solve 分支一致）
 _COMBINING_DESIGNS = combining.DESIGNS
 _DESIGN_LABELS = {
@@ -438,6 +452,27 @@ class StatisticsService:
        )).all()
        return {int(r.tree_id) for r in rows}

+    @staticmethod
+    def _germ_ancestor_closure(tree_germs: set[int],
+                               g_parent_of: dict[int, tuple[int | None, int | None]]) -> set[int]:
+        """种质级祖先闭包：从选中树直系亲本向上取可达祖先（bre_pedigree 只追祖先链）。
+
+        旧实现把全库 bre_pedigree 子代种质无条件并进模型（选中 400 树 → n≈1204），无关种质
+        在 MME 中零贡献（无表型、与选中树无亲缘），EBV/h² 逐位不变，但 n=树数+~800 一越过
+        N_SPARSE 就误入稀疏精确迹二分（~11min/次）。收口到真实系谱规模后，常见「选几百棵
+        候选树」落稠密秒级。与 _germplasm_pedigree 的多源 BFS 思路一致。
+        """
+        out = set(tree_germs)
+        stack = list(tree_germs)
+        while stack:
+            g = stack.pop()
+            gd, gs = g_parent_of.get(g, (None, None))
+            for pp in (gd, gs):
+                if pp is not None and pp not in out:
+                    out.add(pp)
+                    stack.append(pp)
+        return out
+
    async def _build_pedigree(self, tree_ids: list[int]) -> list[dict]:
        """为给定树集合构建系谱（树=子代、种质=亲本，t/g 前缀；含种质级祖先 bre_pedigree）。

@@ -456,7 +491,7 @@ class StatisticsService:
                CrossCombinationModel.male_parent_id,
            )
            .outerjoin(CrossCombinationModel, TreeModel.combination_id == CrossCombinationModel.id)
-            .where(TreeModel.id.in_(tree_ids), TreeModel.is_deleted.is_(False))
+            .where(_id_in(TreeModel.id, tree_ids), TreeModel.is_deleted.is_(False))
        )).all()
        pedigree: list[dict] = []
        tree_germs: set[int] = set()
@@ -489,13 +524,7 @@ class StatisticsService:
                cid = code2g.get(child_code)
                if cid is not None and (gd is not None or gs is not None):
                    g_parent_of[cid] = (gd, gs)
-        all_germ_ids = set(tree_germs)
-        for g, (gd, gs) in g_parent_of.items():
-            all_germ_ids.add(g)
-            if gd is not None:
-                all_germ_ids.add(gd)
-            if gs is not None:
-                all_germ_ids.add(gs)
+        all_germ_ids = self._germ_ancestor_closure(tree_germs, g_parent_of)
        for g in sorted(all_germ_ids):
            if g in g_parent_of:
                gd, gs = g_parent_of[g]
@@ -686,7 +715,7 @@ class StatisticsService:
                    TreeModel.germplasm_id,
                )
                .outerjoin(CrossCombinationModel, TreeModel.combination_id == CrossCombinationModel.id)
-                .where(TreeModel.id.in_(tree_ids))
+                .where(_id_in(TreeModel.id, tree_ids))
            )).all()
            pedigree: list[dict] = []
            tree_germs: set[int] = set()
@@ -739,7 +768,7 @@ class StatisticsService:
                pos_rows = (await self.db.execute(
                    select(TreeModel.id, TreeModel.plot_id, TreeModel.block_no,
                           TreeModel.row_no, TreeModel.col_no)
-                    .where(TreeModel.id.in_(tree_ids), TreeModel.is_deleted.is_(False))
+                    .where(_id_in(TreeModel.id, tree_ids), TreeModel.is_deleted.is_(False))
                )).all()
                grids: dict[tuple[int, int], list[tuple[int, int, int]]] = {}
                for r in pos_rows:
@@ -776,13 +805,7 @@ class StatisticsService:
                    cid = code2g.get(child_code)
                    if cid is not None and (gd is not None or gs is not None):
                        g_parent_of[cid] = (gd, gs)
-            all_germ_ids = set(tree_germs)
-            for g, (gd, gs) in g_parent_of.items():
-                all_germ_ids.add(g)
-                if gd is not None:
-                    all_germ_ids.add(gd)
-                if gs is not None:
-                    all_germ_ids.add(gs)
+            all_germ_ids = self._germ_ancestor_closure(tree_germs, g_parent_of)
            for g in sorted(all_germ_ids):
                if g in g_parent_of:
                    gd, gs = g_parent_of[g]
@@ -846,7 +869,7 @@ class StatisticsService:
                            TreeModel.is_deleted.is_(False),
                            # 与 site 分支对齐：只取本批次选中树，避免他树混入后
                            # clone_of/combo_of 缺失 → 归入虚构基因型组 f{None}
-                            TreeModel.id.in_(tree_ids),
+                            _id_in(TreeModel.id, tree_ids),
                        )
                        .group_by(yr_obs.tree_id, yr_obs.evaluate_year)
                    )
@@ -1093,7 +1116,7 @@ class StatisticsService:
                phenos_keyed = {f"t{tid}": v for tid, v in phenotypes.items()}
                bpos_rows = (await self.db.execute(
                    select(TreeModel.id, TreeModel.block_no)
-                    .where(TreeModel.id.in_(tree_ids), TreeModel.block_no.isnot(None),
+                    .where(_id_in(TreeModel.id, tree_ids), TreeModel.block_no.isnot(None),
                           TreeModel.is_deleted.is_(False))
                )).all()
                block_of: dict[str, int] = {}
@@ -1171,7 +1194,7 @@ class StatisticsService:
                if spatial:
                    pos_rows = (await self.db.execute(
                        select(TreeModel.id, TreeModel.row_no, TreeModel.col_no)
-                        .where(TreeModel.id.in_(tree_ids), TreeModel.is_deleted.is_(False))
+                        .where(_id_in(TreeModel.id, tree_ids), TreeModel.is_deleted.is_(False))
                    )).all()
                    coords: dict[str, tuple[int, int]] = {}
                    missing_coord_ids: list[int] = []
@@ -1391,7 +1414,7 @@ class StatisticsService:
                    TreeModel.germplasm_id,
                )
                .outerjoin(CrossCombinationModel, TreeModel.combination_id == CrossCombinationModel.id)
-                .where(TreeModel.id.in_(tree_ids))
+                .where(_id_in(TreeModel.id, tree_ids))
            )).all()
            pedigree: list[dict] = []
            tree_germs: set[int] = set()
@@ -1434,7 +1457,7 @@ class StatisticsService:
                pos_rows = (await self.db.execute(
                    select(TreeModel.id, TreeModel.plot_id, TreeModel.block_no,
                           TreeModel.row_no, TreeModel.col_no)
-                    .where(TreeModel.id.in_(tree_ids), TreeModel.is_deleted.is_(False))
+                    .where(_id_in(TreeModel.id, tree_ids), TreeModel.is_deleted.is_(False))
                )).all()
                grids: dict[tuple[int, int], list[tuple[int, int, int]]] = {}
                for r in pos_rows:
@@ -1470,13 +1493,7 @@ class StatisticsService:
                    cid = code2g.get(child_code)
                    if cid is not None and (gd is not None or gs is not None):
                        g_parent_of[cid] = (gd, gs)
-            all_germ_ids = set(tree_germs)
-            for g, (gd, gs) in g_parent_of.items():
-                all_germ_ids.add(g)
-                if gd is not None:
-                    all_germ_ids.add(gd)
-                if gs is not None:
-                    all_germ_ids.add(gs)
+            all_germ_ids = self._germ_ancestor_closure(tree_germs, g_parent_of)
            for g in sorted(all_germ_ids):
                if g in g_parent_of:
                    gd, gs = g_parent_of[g]
@@ -1766,7 +1783,7 @@ class StatisticsService:
        rel = {r.tree_id: float(r.reliability or 0.0) for r in rows}
        trees = (await self.db.execute(
            select(TreeModel.id, TreeModel.trial_study_id, TreeModel.combination_id)
-            .where(TreeModel.id.in_(tree_ids), TreeModel.is_deleted.is_(False))
+            .where(_id_in(TreeModel.id, tree_ids), TreeModel.is_deleted.is_(False))
        )).all()
        site_of = {t.id: t.trial_study_id for t in trees}
        combo_of = {t.id: t.combination_id for t in trees}
@@ -2015,7 +2032,7 @@ class StatisticsService:
        if tree_ids:
            rows = (await self.db.execute(
                select(TreeModel.id, TreeModel.clone_id).where(
-                    TreeModel.id.in_(tree_ids), TreeModel.is_deleted.is_(False)
+                    _id_in(TreeModel.id, tree_ids), TreeModel.is_deleted.is_(False)
                )
            )).all()
            tree_map = {t.id: t.clone_id for t in rows}
@@ -2023,7 +2040,7 @@ class StatisticsService:
        clone_map: dict[int, CloneModel] = {}
        if clone_ids:
            clones = (await self.db.execute(
-                select(CloneModel).where(CloneModel.id.in_(clone_ids), CloneModel.is_deleted.is_(False))
+                select(CloneModel).where(_id_in(CloneModel.id, clone_ids), CloneModel.is_deleted.is_(False))
            )).scalars().all()
            clone_map = {c.id: c for c in clones}
        agg: dict[int, dict[str, float | int]] = {}
@@ -2533,7 +2550,7 @@ class StatisticsService:
            pid2code = {pid: c for c, pid in batch_ids.items()}
            pv_rows = (await self.db.execute(
                select(PredictionValueModel.prediction_id, PredictionValueModel.tree_id, PredictionValueModel.reliability).where(
-                    PredictionValueModel.prediction_id.in_(list(batch_ids.values())),
+                    _id_in(PredictionValueModel.prediction_id, list(batch_ids.values())),
                    PredictionValueModel.tree_id.isnot(None),
                )
            )).all()
@@ -2677,7 +2694,7 @@ class StatisticsService:
        """
        rows = (await self.db.execute(
            select(TreeModel.id, TreeModel.clone_id, TreeModel.tree_no).where(
-                TreeModel.id.in_(tree_ids), TreeModel.is_deleted.is_(False)
+                _id_in(TreeModel.id, tree_ids), TreeModel.is_deleted.is_(False)
            )
        )).all()
        clone_of = {t.id: t.clone_id for t in rows}
@@ -2687,7 +2704,7 @@ class StatisticsService:
        if clone_ids:
            clones = (await self.db.execute(
                select(CloneModel.id, CloneModel.clone_code).where(
-                    CloneModel.id.in_(clone_ids), CloneModel.is_deleted.is_(False)
+                    _id_in(CloneModel.id, clone_ids), CloneModel.is_deleted.is_(False)
                )
            )).all()
            clone_label = {c.id: c.clone_code for c in clones}
@@ -3011,7 +3028,7 @@ class StatisticsService:
        if refs:
            pv_rows = (await self.db.execute(
                select(PredictionValueModel.tree_id, PredictionValueModel.reliability).where(
-                    PredictionValueModel.prediction_id.in_(list(refs.values())),
+                    _id_in(PredictionValueModel.prediction_id, list(refs.values())),
                    PredictionValueModel.tree_id.isnot(None),
                )
            )).all()
@@ -3048,11 +3065,11 @@ class StatisticsService:
        all_tids = {t for u in units for t in u["tree_ids"]}
        tree_rows = (await self.db.execute(
            select(TreeModel.id, TreeModel.combination_id, TreeModel.tree_no, TreeModel.stage)
-            .where(TreeModel.id.in_(all_tids))
+            .where(_id_in(TreeModel.id, all_tids))
        )).all()
        tree_map = {t.id: t for t in tree_rows}
        existing = (await self.db.execute(
-            select(SelectionResultModel).where(SelectionResultModel.tree_id.in_(all_tids))
+            select(SelectionResultModel).where(_id_in(SelectionResultModel.tree_id, all_tids))
        )).scalars().all()
        existing_by_tree = {e.tree_id: e for e in existing}

@@ -3144,7 +3161,7 @@ class StatisticsService:
        """
        stmt = select(TreeModel.id, TreeModel.dam_id, TreeModel.sire_id).where(TreeModel.is_deleted.is_(False))
        if tree_ids:
-            stmt = stmt.where(TreeModel.id.in_(tree_ids))
+            stmt = stmt.where(_id_in(TreeModel.id, tree_ids))
        rows = (await self.db.execute(stmt)).all()
        if not rows:
            raise CustomException(msg="无活动单株")
@@ -3203,11 +3220,35 @@ class StatisticsService:
            raise CustomException(msg=f"性状不可用于统计(需数值型且存在): {trait_code}")
        direction = (trait_row.direction or "desc").lower()
        desirable_high = direction != "asc"  # desc/None：表型越高越好；asc：越低越好（如病指）
-        # 1) 单株系谱 → 深层系谱 A 矩阵 → 近交系数 F
+        # 0) 先取目标性状每树表型（同株多次取均值）——近交只针对有该性状表型的单株；
+        #    全库 3.6 万树建稠密 A 会 OOM（36990²×8B≈10 GiB），必须按表型树收口。
+        obs = TraitObservationModel
+        stmt = (
+            select(obs.tree_id, func.avg(obs.value_numeric).label("v"))
+            .join(TreeModel, obs.tree_id == TreeModel.id)
+            .where(
+                obs.trait_id == trait_id,
+                obs.value_numeric.isnot(None),
+                obs.tree_id.isnot(None),
+                obs.is_deleted.is_(False),
+                TreeModel.is_deleted.is_(False),
+            )
+            .group_by(obs.tree_id)
+        )
+        if year is not None:
+            stmt = stmt.where(obs.evaluate_year == year)
+        if trial_study_id is not None:
+            stmt = stmt.where(obs.trial_study_id == trial_study_id)
+        agg = (await self.db.execute(stmt)).all()
+        if not agg:
+            raise CustomException(msg="无可用表型数据（该性状为空）")
+        # 1) 单株系谱（仅表型树）→ 深层系谱 A 矩阵 → 近交系数 F
        # 单株 dam/sire 引用种质；种质系谱经 _germplasm_pedigree 多源 BFS 解析（bre_pedigree
        # 名称反查 + tree FK 兜底 + 祖先展开，与 OCS/mating 同源）——多世代项目（半同胞共享
        # 祖亲）不再低估 F；被引用但未解析（软删等）种质补 founder 兜底。
-        stmt = select(TreeModel.id, TreeModel.dam_id, TreeModel.sire_id).where(TreeModel.is_deleted.is_(False))
+        stmt = (select(TreeModel.id, TreeModel.dam_id, TreeModel.sire_id)
+                .where(TreeModel.is_deleted.is_(False),
+                       _id_in(TreeModel.id, [r.tree_id for r in agg])))
        rows = (await self.db.execute(stmt)).all()
        if not rows:
            raise CustomException(msg="无活动单株")
@@ -3249,27 +3290,7 @@ class StatisticsService:
        for k, v in rmat_res["inbreeding"].items():
            if k.startswith("t") and k[1:].isdigit():
                f_by_tree[int(k[1:])] = float(v)
-        # 2) 目标性状每树表型（同株多次取均值）
-        obs = TraitObservationModel
-        stmt = (
-            select(obs.tree_id, func.avg(obs.value_numeric).label("v"))
-            .join(TreeModel, obs.tree_id == TreeModel.id)
-            .where(
-                obs.trait_id == trait_id,
-                obs.value_numeric.isnot(None),
-                obs.tree_id.isnot(None),
-                obs.is_deleted.is_(False),
-                TreeModel.is_deleted.is_(False),
-            )
-            .group_by(obs.tree_id)
-        )
-        if year is not None:
-            stmt = stmt.where(obs.evaluate_year == year)
-        if trial_study_id is not None:
-            stmt = stmt.where(obs.trial_study_id == trial_study_id)
-        agg = (await self.db.execute(stmt)).all()
-        if not agg:
-            raise CustomException(msg="无可用表型数据（该性状为空）")
+        # 2) 表型已在上方取好（agg），直接配对 F
        pairs: list[tuple[int, float, float]] = []
        for r in agg:
            f = f_by_tree.get(int(r.tree_id))
@@ -3329,7 +3350,7 @@ class StatisticsService:
            interpretation = f"未检出近交衰退（F 对表型无{trend}影响）"
        tree_ids = [p[0] for p in pairs]
        trows = (await self.db.execute(
-            select(TreeModel.id, TreeModel.tree_no).where(TreeModel.id.in_(tree_ids))
+            select(TreeModel.id, TreeModel.tree_no).where(_id_in(TreeModel.id, tree_ids))
        )).all()
        tree_map = {tid: no for tid, no in trows}
        base.update({
@@ -3470,7 +3491,7 @@ class StatisticsService:
                if flagged_ids:
                    for tid, tno in (await self.db.execute(
                            select(TreeModel.id, TreeModel.tree_no)
-                            .where(TreeModel.id.in_(flagged_ids)))).all():
+                            .where(_id_in(TreeModel.id, flagged_ids)))).all():
                        tree_no_map[int(tid)] = tno or ""
                excluded_ids = await self._excluded_tree_ids()
                mendelian_section = {
@@ -3665,7 +3686,7 @@ class StatisticsService:
        gs = (await self.db.execute(
            select(BreedingGermplasmModel.id, BreedingGermplasmModel.cultivar_name,
                   BreedingGermplasmModel.accession_no)
-            .where(BreedingGermplasmModel.id.in_(gids))
+            .where(_id_in(BreedingGermplasmModel.id, gids))
        )).all()
        cand_ids = [int(g.id) for g in gs]
        if not cand_ids:
@@ -3793,7 +3814,7 @@ class StatisticsService:
            pvs = (await self.db.execute(
                select(PredictionValueModel).where(
                    PredictionValueModel.prediction_id == ebv_pred.id,
-                    PredictionValueModel.germplasm_id.in_(cand_ids),
+                    _id_in(PredictionValueModel.germplasm_id, cand_ids),
                    PredictionValueModel.predicted_value.isnot(None),
                )
            )).scalars().all()
@@ -3819,7 +3840,7 @@ class StatisticsService:
                mset = set(missing)
                off_tree = (await self.db.execute(
                    select(TreeModel.id, TreeModel.dam_id, TreeModel.sire_id, TreeModel.combination_id).where(
-                        or_(TreeModel.dam_id.in_(missing), TreeModel.sire_id.in_(missing)),
+                        or_(_id_in(TreeModel.dam_id, missing), _id_in(TreeModel.sire_id, missing)),
                        TreeModel.is_deleted.is_(False),
                    )
                )).all()
@@ -3828,7 +3849,7 @@ class StatisticsService:
                    off_pvs = (await self.db.execute(
                        select(PredictionValueModel.tree_id, PredictionValueModel.predicted_value).where(
                            PredictionValueModel.prediction_id == ebv_pred.id,
-                            PredictionValueModel.tree_id.in_(off_ids),
+                            _id_in(PredictionValueModel.tree_id, off_ids),
                            PredictionValueModel.predicted_value.isnot(None),
                        )
                    )).all()
@@ -3887,7 +3908,7 @@ class StatisticsService:
        cand_ids = list(dict.fromkeys(candidate_germplasm_ids))
        gs = (await self.db.execute(
            select(BreedingGermplasmModel).where(
-                BreedingGermplasmModel.id.in_(cand_ids),
+                _id_in(BreedingGermplasmModel.id, cand_ids),
                BreedingGermplasmModel.is_deleted.is_(False),
            )
        )).scalars().all()
@@ -4048,7 +4069,7 @@ class StatisticsService:
            raise CustomException(msg=f"选择数量 n_select={n_select} 超过候选亲本数 {len(cand_ids)}")
        gs = (await self.db.execute(
            select(BreedingGermplasmModel).where(
-                BreedingGermplasmModel.id.in_(cand_ids),
+                _id_in(BreedingGermplasmModel.id, cand_ids),
                BreedingGermplasmModel.is_deleted.is_(False),
            )
        )).scalars().all()
@@ -4210,7 +4231,7 @@ class StatisticsService:

        cand_rows = (await self.db.execute(
            select(TreeModel.id, TreeModel.tree_no).where(
-                TreeModel.id.in_(list(dict.fromkeys(candidate_tree_ids))),
+                _id_in(TreeModel.id, list(dict.fromkeys(candidate_tree_ids))),
                TreeModel.is_deleted.is_(False),
            )
        )).all()
@@ -4241,7 +4262,7 @@ class StatisticsService:
                select(GenotypeCallModel.sample_id, GenotypeCallModel.marker_id,
                       GenotypeCallModel.allele)
                .where(
-                    GenotypeCallModel.marker_id.in_(bg_marker_ids),
+                    _id_in(GenotypeCallModel.marker_id, bg_marker_ids),
                    GenotypeCallModel.allele.isnot(None),
                    GenotypeCallModel.is_deleted.is_(False),
                )
@@ -4249,7 +4270,7 @@ class StatisticsService:
            if calls:
                sample_ids = sorted({c.sample_id for c in calls})
                samples = (await self.db.execute(
-                    select(GenotypeSampleModel).where(GenotypeSampleModel.id.in_(sample_ids))
+                    select(GenotypeSampleModel).where(_id_in(GenotypeSampleModel.id, sample_ids))
                )).scalars().all()
                sample_tree, _ = await self._sample_to_tree_map(samples)
                for c in calls:
@@ -4412,7 +4433,7 @@ class StatisticsService:
                block_of[ent.id] = blk
            await self.db.execute(
                update(TrialStudyEntryModel)
-                .where(TrialStudyEntryModel.id.in_(list(block_of)))
+                .where(_id_in(TrialStudyEntryModel.id, list(block_of)))
                .values(block_no=case(block_of, value=TrialStudyEntryModel.id, else_=None))
            )
            await self.db.flush()
@@ -4468,7 +4489,7 @@ class StatisticsService:
        if block_of:
            await self.db.execute(
                update(TrialStudyEntryModel)
-                .where(TrialStudyEntryModel.id.in_(list(block_of)))
+                .where(_id_in(TrialStudyEntryModel.id, list(block_of)))
                .values(block_no=case(block_of, value=TrialStudyEntryModel.id, else_=None))
            )
            await self.db.flush()
@@ -4577,7 +4598,7 @@ class StatisticsService:
        if block_of:
            await self.db.execute(
                update(TrialStudyEntryModel)
-                .where(TrialStudyEntryModel.id.in_(list(block_of)))
+                .where(_id_in(TrialStudyEntryModel.id, list(block_of)))
                .values(block_no=case(block_of, value=TrialStudyEntryModel.id, else_=None))
            )
            await self.db.flush()
@@ -4634,7 +4655,7 @@ class StatisticsService:
        if block:
            tree_ids = [int(r.tree_id) for r in agg]
            block_rows = (await self.db.execute(
-                select(TreeModel.id, TreeModel.block_no).where(TreeModel.id.in_(tree_ids))
+                select(TreeModel.id, TreeModel.block_no).where(_id_in(TreeModel.id, tree_ids))
            )).all()
            block_of = {int(tid): bn for tid, bn in block_rows}
            rows = [
@@ -4998,10 +5019,10 @@ class StatisticsService:
        all_tree_ids = sorted(set(by_tree) | set(ebv_map) | mas["tree_ids"])
        if all_tree_ids:
            tree_no_map = dict((await self.db.execute(
-                select(TreeModel.id, TreeModel.tree_no).where(TreeModel.id.in_(all_tree_ids))
+                select(TreeModel.id, TreeModel.tree_no).where(_id_in(TreeModel.id, all_tree_ids))
            )).all())
            combo_map = dict((await self.db.execute(
-                select(TreeModel.id, TreeModel.combination_id).where(TreeModel.id.in_(all_tree_ids))
+                select(TreeModel.id, TreeModel.combination_id).where(_id_in(TreeModel.id, all_tree_ids))
            )).all())
        else:
            tree_no_map, combo_map = {}, {}
@@ -5168,7 +5189,7 @@ class StatisticsService:
            tree_ids = sorted({r.tree_id for r in rows})
            trows = (await self.db.execute(
                select(TreeModel.id, TreeModel.clone_id, TreeModel.combination_id).where(
-                    TreeModel.id.in_(tree_ids), TreeModel.is_deleted.is_(False))
+                    _id_in(TreeModel.id, tree_ids), TreeModel.is_deleted.is_(False))
            )).all()
            clone_of = {t.id: t.clone_id for t in trows}
            combo_of = {t.id: t.combination_id for t in trows}
@@ -5545,7 +5566,7 @@ class StatisticsService:
        else:
            tree_rows = (await self.db.execute(
                select(TreeModel.id, TreeModel.clone_id, TreeModel.tree_no).where(
-                    TreeModel.id.in_(list(by_tree)), TreeModel.is_deleted.is_(False)
+                    _id_in(TreeModel.id, list(by_tree)), TreeModel.is_deleted.is_(False)
                )
            )).all()
            clone_of = {t.id: t.clone_id for t in tree_rows}
@@ -5555,7 +5576,7 @@ class StatisticsService:
            if clone_ids:
                clones = (await self.db.execute(
                    select(CloneModel.id, CloneModel.clone_code).where(
-                        CloneModel.id.in_(clone_ids), CloneModel.is_deleted.is_(False)
+                        _id_in(CloneModel.id, clone_ids), CloneModel.is_deleted.is_(False)
                    )
                )).all()
                clone_label = {c.id: c.clone_code for c in clones}
@@ -5807,7 +5828,7 @@ class StatisticsService:
            rel = result["reliability"]
            ebv = result["ebv"]
            germ_of = dict((await self.db.execute(
-                select(TreeModel.id, TreeModel.germplasm_id).where(TreeModel.id.in_(tree_ids))
+                select(TreeModel.id, TreeModel.germplasm_id).where(_id_in(TreeModel.id, tree_ids))
            )).all())
            direction = (trait_row.direction or "desc").lower()
            tree_ebv = [(tid, float(ebv.get(f"t{tid}", 0.0))) for tid in tree_ids]
@@ -5898,7 +5919,7 @@ class StatisticsService:
            select(GenotypeCallModel.sample_id, GenotypeCallModel.marker_id,
                   GenotypeCallModel.allele)
            .where(
-                GenotypeCallModel.sample_id.in_(sample_ids),
+                _id_in(GenotypeCallModel.sample_id, sample_ids),
                GenotypeCallModel.marker_id.isnot(None),
                GenotypeCallModel.allele.isnot(None),
                GenotypeCallModel.is_deleted.is_(False),
@@ -5909,7 +5930,7 @@ class StatisticsService:
        marker_ids = sorted({r.marker_id for r in call_rows})
        markers = (await self.db.execute(
            select(MarkerModel.id, MarkerModel.marker_name, MarkerModel.marker_type).where(
-                MarkerModel.id.in_(marker_ids), MarkerModel.is_deleted.is_(False))
+                _id_in(MarkerModel.id, marker_ids), MarkerModel.is_deleted.is_(False))
        )).all()
        if not markers:
            raise CustomException(msg="标记不存在")
@@ -6068,7 +6089,7 @@ class StatisticsService:
            select(GenotypeCallModel.sample_id, GenotypeCallModel.marker_id,
                   GenotypeCallModel.allele)
            .where(
-                GenotypeCallModel.sample_id.in_(sample_ids),
+                _id_in(GenotypeCallModel.sample_id, sample_ids),
                GenotypeCallModel.marker_id.isnot(None),
                GenotypeCallModel.allele.isnot(None),
                GenotypeCallModel.is_deleted.is_(False),
@@ -6081,7 +6102,7 @@ class StatisticsService:
            select(MarkerModel.id, MarkerModel.marker_name,
                   MarkerModel.chromosome, MarkerModel.position,
                   MarkerModel.marker_type)
-            .where(MarkerModel.id.in_(marker_ids), MarkerModel.is_deleted.is_(False))
+            .where(_id_in(MarkerModel.id, marker_ids), MarkerModel.is_deleted.is_(False))
        )).all()
        if not markers:
            raise CustomException(msg="标记不存在")
@@ -6594,7 +6615,7 @@ class StatisticsService:
            select(GenotypeCallModel.sample_id, GenotypeCallModel.marker_id,
                   GenotypeCallModel.allele)
            .where(
-                GenotypeCallModel.marker_id.in_(marker_ids),
+                _id_in(GenotypeCallModel.marker_id, marker_ids),
                GenotypeCallModel.allele.isnot(None),
                GenotypeCallModel.is_deleted.is_(False),
            )
@@ -6603,7 +6624,7 @@ class StatisticsService:
            return out
        sample_ids = sorted({c.sample_id for c in calls})
        samples = (await self.db.execute(
-            select(GenotypeSampleModel).where(GenotypeSampleModel.id.in_(sample_ids))
+            select(GenotypeSampleModel).where(_id_in(GenotypeSampleModel.id, sample_ids))
        )).scalars().all()
        sample_tree, _ = await self._sample_to_tree_map(samples)
        tree_geno: dict[int, dict[int, str]] = {}
@@ -6744,7 +6765,7 @@ class StatisticsService:
            # GXE 分支：环境维度记录 → rec_phenos（与 run_ablup 同构）
            grp_rows = (await self.db.execute(
                select(TreeModel.id, TreeModel.clone_id, TreeModel.combination_id).where(
-                    TreeModel.id.in_(tree_ids), TreeModel.is_deleted.is_(False))
+                    _id_in(TreeModel.id, tree_ids), TreeModel.is_deleted.is_(False))
            )).all()
            clone_of = {t.id: t.clone_id for t in grp_rows}
            combo_of = {t.id: t.combination_id for t in grp_rows}
@@ -6764,7 +6785,7 @@ class StatisticsService:
            else:
                site_rows = (await self.db.execute(
                    select(TreeModel.id, TreeModel.trial_study_id).where(
-                        TreeModel.id.in_(tree_ids), TreeModel.is_deleted.is_(False))
+                        _id_in(TreeModel.id, tree_ids), TreeModel.is_deleted.is_(False))
                )).all()
                site_of = {t.id: t.trial_study_id for t in site_rows}
                rec_phenos = {}
@@ -12,6 +12,7 @@ from app.utils.dict_util import DictLabelResolver, dict_value_to_label
 from app.utils.number_gen import NumberGenService

 from .crud import BreedingTreeCRUD
+from .model import TreeModel
 from .schema import (
    TreeCreateSchema,
    TreeOutSchema,
@@ -370,8 +371,10 @@ class TreeService:
        await BreedingTreeCRUD(self.auth, self.db).delete(ids=ids)

    async def list_options(self) -> list[dict[str, Any]]:
-        """供前端下拉选择使用：返回 [{value, label}]。"""
-        obj_list = await BreedingTreeCRUD(self.auth, self.db).get_list(order_by=[{"id": "asc"}])
+        """供前端下拉选择使用：返回 [{value, label}]。只取 id/tree_no 两列，避免全表列加载拖慢下拉。"""
+        obj_list = await BreedingTreeCRUD(self.auth, self.db).get_list(
+            order_by=[{"id": "asc"}], load_columns=[TreeModel.id, TreeModel.tree_no]
+        )
        return [{"value": o.id, "label": o.tree_no} for o in obj_list]

    @staticmethod