"""育种统计 数据模型（一期 2 个最基础：ABLUP·EBV + 配合力 GCA/SCA）"""
from datetime import datetime

from sqlalchemy import (
    JSON,
    Boolean,
    Date,
    DateTime,
    Float,
    ForeignKey,
    Index,
    Integer,
    Numeric,
    String,
    Text,
)
from sqlalchemy.orm import Mapped, mapped_column

from app.core.base_model import MappedBase, ModelMixin, UserMixin


class PredictionModel(ModelMixin, UserMixin, MappedBase):
    """育种值模型/批次元数据（§3.12）。"""

    __tablename__ = "bre_prediction"

    model_name: Mapped[str | None] = mapped_column(String(128), nullable=True, comment="模型名称")
    trait_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_trait.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="性状"
    )
    method: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="方法(ABLUP/COMBINING)")
    stage: Mapped[str | None] = mapped_column(String(16), nullable=True, comment="测定阶段(juvenile幼龄/evaluation成株)")
    accuracy: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="模型精度(批次PA=√均值可靠性)")
    train_n: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="训练样本量")
    heritability: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="遗传力 h²")
    predict_date: Mapped[datetime | None] = mapped_column(Date, nullable=True, comment="预测日期")
    data_version: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="数据版本(输入快照版本)")
    input_hash: Mapped[str | None] = mapped_column(String(64), nullable=True, comment="输入快照SHA256(表型+系谱)")
    engine_version: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="统计引擎版本")
    is_active: Mapped[bool | None] = mapped_column(Boolean, nullable=True, default=False, comment="当前生效版本(同性状最近批次，供版本回滚)")
    note: Mapped[str | None] = mapped_column(String(512), nullable=True, comment="备注")

    __table_args__ = (
        Index("ix_bre_prediction_created_deleted", "created_time", "is_deleted"),
    )


class PredictionValueModel(ModelMixin, UserMixin, MappedBase):
    """个体预测值（EBV 持久化，V1.1 建表）。"""

    __tablename__ = "bre_prediction_value"

    prediction_id: Mapped[int] = mapped_column(
        Integer, ForeignKey("bre_prediction.id", ondelete="CASCADE"),
        index=True, nullable=False, comment="关联模型"
    )
    germplasm_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_germplasm.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="种质(亲本级)"
    )
    tree_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_tree.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="单株(树级)"
    )
    trait_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_trait.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="性状"
    )
    predicted_value: Mapped[float | None] = mapped_column(Float, nullable=True, comment="预测值/EBV")
    reliability: Mapped[float | None] = mapped_column(Float, nullable=True, comment="可靠性")
    pa: Mapped[float | None] = mapped_column(Float, nullable=True, comment="预测准确度 PA=√可靠性")
    rank: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="排名")

    __table_args__ = (
        Index("ix_bre_prediction_value_created_deleted", "created_time", "is_deleted"),
    )


class CombiningAbilityModel(ModelMixin, UserMixin, MappedBase):
    """配合力分析结果（§3.12；GCA 按亲本、SCA 按组合）。"""

    __tablename__ = "bre_combining_ability"

    model_name: Mapped[str | None] = mapped_column(String(128), nullable=True, comment="模型名称")
    trait_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_trait.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="性状"
    )
    method: Mapped[str | None] = mapped_column(String(16), nullable=True, comment="方法")
    design_type: Mapped[str | None] = mapped_column(
        String(32), nullable=True,
        comment="分析采用交配设计(full_diallel/partial_diallel/line_tester/nciii，结果快照)",
    )
    gca_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="亲本一般配合力 {parent: gca}")
    sca_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="组合特殊配合力 {combo: {p1,p2,sca}}")
    anova_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="配合力方差分解(F_gca/p_gca)")

    __table_args__ = (
        Index("ix_bre_combining_ability_created_deleted", "created_time", "is_deleted"),
    )


class SelectionIndexModel(ModelMixin, UserMixin, MappedBase):
    """选择指数 批次（加权 Z 综合；可选接入 ABLUP 批次 EBV 与 h² 加权，结果持久化）。"""

    __tablename__ = "bre_selection_index"

    model_name: Mapped[str | None] = mapped_column(String(128), nullable=True, comment="批次名称")
    method: Mapped[str | None] = mapped_column(String(16), nullable=True, comment="SI_EBV/SI_PHENO")
    weights_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="{trait_code: 用户权重}")
    batch_refs_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="{trait_code: ABLUP批次id}")
    heritability_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="{trait_code: h²}")
    top_n: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="返回前 N")
    result_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="{traits, top:[{tree_id,index,detail}]}")

    __table_args__ = (
        Index("ix_bre_selection_index_created_deleted", "created_time", "is_deleted"),
    )


class AnovaResultModel(ModelMixin, UserMixin, MappedBase):
    """单因素 ANOVA 结果（家系方差组分 + 广义遗传力 H²）。"""

    __tablename__ = "bre_anova_result"

    model_name: Mapped[str | None] = mapped_column(String(128), nullable=True, comment="批次名称")
    trait_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_trait.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="性状"
    )
    method: Mapped[str | None] = mapped_column(String(16), nullable=True, comment="ANOVA/H2")
    result_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="方差组分/H²/F/p")

    __table_args__ = (
        Index("ix_bre_anova_result_created_deleted", "created_time", "is_deleted"),
    )


class CvResultModel(ModelMixin, UserMixin, MappedBase):
    """k-fold 交叉验证结果（外部验证预测准确度，区别于 PEV 可靠性）。"""

    __tablename__ = "bre_cv_result"

    trait_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_trait.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="性状"
    )
    trait_code: Mapped[str | None] = mapped_column(String(64), nullable=True, comment="性状编码")
    method: Mapped[str | None] = mapped_column(String(16), nullable=True, comment="KFCV/ABLUP 或 KFCV/GXE")
    k: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="折数")
    n_total: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="总记录数")
    n_individuals: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="加性个体数")
    mean_pearson: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="有效折预测-观测 pearson 均值")
    mean_rmse: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="有效折 RMSE 均值")
    pooled_pearson: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="合并全部折 pearson")
    pooled_rmse: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="合并全部折 RMSE")
    cv_accuracy: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="预测准确度(=mean_pearson)")
    h2: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="有效折遗传力均值")
    folds_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="折明细快照")
    data_version: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="数据版本")
    input_hash: Mapped[str | None] = mapped_column(String(64), nullable=True, comment="输入快照SHA256")
    engine_version: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="统计引擎版本")
    note: Mapped[str | None] = mapped_column(String(512), nullable=True, comment="备注")

    __table_args__ = (
        Index("ix_bre_cv_result_created_deleted", "created_time", "is_deleted"),
    )


class CvFoldModel(ModelMixin, UserMixin, MappedBase):
    """k-fold CV 折明细。"""

    __tablename__ = "bre_cv_fold"

    cv_result_id: Mapped[int] = mapped_column(
        Integer, ForeignKey("bre_cv_result.id", ondelete="CASCADE"),
        index=True, nullable=False, comment="关联 CV 结果"
    )
    fold_idx: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="折号(1-based)")
    n_train: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="训练记录数")
    n_test: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="测试个体数")
    n_eval: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="有效评估个体数")
    pearson: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="折内预测-观测 pearson")
    rmse: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="折内 RMSE")
    error: Mapped[str | None] = mapped_column(String(512), nullable=True, comment="折失败原因")

    __table_args__ = (
        Index("ix_bre_cv_fold_created_deleted", "created_time", "is_deleted"),
    )


class StatisticsJobModel(ModelMixin, UserMixin, MappedBase):
    """统计异步任务（R 引擎；ABLUP/COMBINING 提交后异步执行并回写结果）。"""

    __tablename__ = "bre_statistics_job"

    job_type: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="ABLUP/COMBINING")
    params_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="入参")
    status: Mapped[str] = mapped_column(String(16), nullable=False, default="PENDING", comment="PENDING/RUNNING/SUCCESS/FAILED")
    result_ref: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="结果主表 id")
    error_msg: Mapped[str | None] = mapped_column(Text, nullable=True, comment="错误信息")
    started_time: Mapped[datetime | None] = mapped_column(DateTime, nullable=True)
    finished_time: Mapped[datetime | None] = mapped_column(DateTime, nullable=True)

    __table_args__ = (
        Index("ix_bre_statistics_job_created_deleted", "created_time", "is_deleted"),
    )


class GeneticCorrResultModel(ModelMixin, UserMixin, MappedBase):
    """成对双性状 BLUP 遗传相关矩阵结果（MT-BLUP，§8.18）。

    逐对 bivariate 估计 σ²a1/σ²a2/σ²a12（固定单性状方差、仅对遗传相关 ρ 作
    1-D 剖面 REML golden-max），输出遗传相关矩阵 r_g 供 Smith-Hazel 指数
    G 矩阵非对角替换 Calo 近似。
    """

    __tablename__ = "bre_genetic_corr_result"

    trait_ids_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="{trait_code: trait_id}")
    matrix_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="遗传相关矩阵 {code: {code: r_g}}")
    sigma_a_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="加性方差矩阵 {code: {code: σa}}")
    heritability_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="{code: h²}")
    pairs_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="逐对详情 {pair: {r_g, va1, va2, ve1, ve2, n_common, converged, warning}}")
    n_common: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="成对共有树数(最小)")
    data_version: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="数据版本")
    input_hash: Mapped[str | None] = mapped_column(String(64), nullable=True, comment="输入快照SHA256(表型+系谱)")
    engine_version: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="统计引擎版本")
    note: Mapped[str | None] = mapped_column(String(512), nullable=True, comment="备注")

    __table_args__ = (
        Index("ix_bre_genetic_corr_result_created_deleted", "created_time", "is_deleted"),
    )


class TypeBResultModel(ModelMixin, UserMixin, MappedBase):
    """Type-B 多环境遗传相关结果（§8.27）。

    同一性状在多个环境（site=研究点 / year=年份）的表型，以环境互为"性状"逐对
    REML 双性状 BLUP 估计遗传相关 r_g（复用 mtblup.solve_bivariate）；r_g 均值
    即 Type-B 遗传相关，衡量基因型×环境互作强度与跨环境遗传稳定性。
    """

    __tablename__ = "bre_type_b_result"

    trait_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_trait.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="性状"
    )
    trait_code: Mapped[str | None] = mapped_column(String(64), nullable=True, comment="性状编码")
    env_dim: Mapped[str | None] = mapped_column(String(16), nullable=True, comment="环境维度(site/year)")
    method: Mapped[str | None] = mapped_column(String(16), nullable=True, comment="方法(reml/calo，calo 无环境级EBV时回退reml)")
    envs_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="{env_code: env_label}")
    matrix_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="环境间遗传相关矩阵 {env: {env: r_g}}")
    pairs_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="逐对详情 {pair: {r_g, va1, va2, ve1, ve2, n_common, converged, warning}}")
    n_common: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="成对共有树数(最小)")
    data_version: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="数据版本")
    input_hash: Mapped[str | None] = mapped_column(String(64), nullable=True, comment="输入快照SHA256")
    engine_version: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="统计引擎版本")
    note: Mapped[str | None] = mapped_column(String(512), nullable=True, comment="备注")

    __table_args__ = (
        Index("ix_bre_type_b_result_created_deleted", "created_time", "is_deleted"),
    )


class StabilityResultModel(ModelMixin, UserMixin, MappedBase):
    """AMMI / Finlay-Wilkinson 稳定性分析结果（§8.18）。

    两因素(基因型×环境)均值表 → AMMI SVD(IPC1/IPC2/ASV/ecovalence) 与
    Finlay-Wilkinson 回归斜率 b，输出稳定性排名。
    """

    __tablename__ = "bre_stability_result"

    trait_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_trait.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="性状"
    )
    trait_code: Mapped[str | None] = mapped_column(String(64), nullable=True, comment="性状编码")
    env_dim: Mapped[str | None] = mapped_column(String(16), nullable=True, comment="环境维度(site/year)")
    methods: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="方法(ammi,finlay)")
    detail_json: Mapped[dict | None] = mapped_column(JSON, nullable=True, comment="稳定性分析明细快照")
    data_version: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="数据版本")
    input_hash: Mapped[str | None] = mapped_column(String(64), nullable=True, comment="输入快照SHA256")
    engine_version: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="统计引擎版本")
    note: Mapped[str | None] = mapped_column(String(512), nullable=True, comment="备注")

    __table_args__ = (
        Index("ix_bre_stability_result_created_deleted", "created_time", "is_deleted"),
    )


class GwasResultModel(ModelMixin, UserMixin, MappedBase):
    """GWAS 关联分析批次结果（§8.22）。

    GLM+PC 逐标记单标记回归（纯 numpy + fdist），输出显著标记集与 QTL 区间。
    """

    __tablename__ = "bre_gwas_result"

    dataset_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_genotyping_dataset.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="基因型数据集"
    )
    trait_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_trait.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="性状"
    )
    trait_code: Mapped[str | None] = mapped_column(String(64), nullable=True, comment="性状编码")
    method: Mapped[str | None] = mapped_column(String(16), nullable=True, comment="方法(gwas/ssgwas)")
    n_individuals: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="样本数(表型∩基因型)")
    n_markers: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="标记总数")
    m_after_maf: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="MAF 过滤后标记数")
    maf_min: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="MAF 过滤下限")
    n_pc: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="群体结构主成分数")
    sig_level: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="显著性水平")
    threshold_bonf: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="Bonferroni 阈值")
    n_sig_bonf: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="Bonferroni 显著标记数")
    n_sig_fdr: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="FDR q<α 标记数")
    n_qtl: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="QTL 区间数")
    data_version: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="数据版本")
    input_hash: Mapped[str | None] = mapped_column(String(64), nullable=True, comment="输入快照SHA256(表型+系谱+基因型)")
    engine_version: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="统计引擎版本")
    remark: Mapped[str | None] = mapped_column(String(512), nullable=True, comment="备注")

    __table_args__ = (
        Index("ix_bre_gwas_result_created_deleted", "created_time", "is_deleted"),
    )


class GwasSnpModel(ModelMixin, UserMixin, MappedBase):
    """GWAS 逐标记关联结果（§8.22）。"""

    __tablename__ = "bre_gwas_snp"

    gwas_result_id: Mapped[int] = mapped_column(
        Integer, ForeignKey("bre_gwas_result.id", ondelete="CASCADE"),
        index=True, nullable=False, comment="关联 GWAS 批次"
    )
    marker_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_marker.id", ondelete="SET NULL"),
        index=True, nullable=True, comment="标记"
    )
    marker_name: Mapped[str | None] = mapped_column(String(64), nullable=True, comment="标记名称")
    chromosome: Mapped[str | None] = mapped_column(String(16), nullable=True, comment="染色体")
    position: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="物理位置(bp)")
    maf: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="次要等位频率")
    effect: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="加性效应")
    se: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="效应标准误")
    t_value: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="t 值")
    p_value: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="p 值")
    neg_log10p: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="−log10(p)")
    q_value: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="BH-FDR q 值")
    sig_bonf: Mapped[bool | None] = mapped_column(Boolean, nullable=True, comment="Bonferroni 显著")
    sig_fdr: Mapped[bool | None] = mapped_column(Boolean, nullable=True, comment="FDR 显著")

    __table_args__ = (
        Index("ix_bre_gwas_snp_created_deleted", "created_time", "is_deleted"),
    )


class QtlModel(ModelMixin, UserMixin, MappedBase):
    """QTL 区间（§8.22）。

    source=known 为人工录入已知 QTL（桃成熟期/果重等位点）；source=gwas 为 GWAS
    显著标记聚类自动定位（gwas_result_id 关联批次）。
    """

    __tablename__ = "bre_qtl"

    trait_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_trait.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="性状"
    )
    chromosome: Mapped[str | None] = mapped_column(String(16), nullable=True, comment="染色体")
    start_bp: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="区间起点(bp)")
    end_bp: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="区间终点(bp)")
    peak_marker_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_marker.id", ondelete="SET NULL"),
        index=True, nullable=True, comment="峰标记"
    )
    peak_marker_name: Mapped[str | None] = mapped_column(String(64), nullable=True, comment="峰标记名称")
    peak_p: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="峰标记 p 值")
    n_markers: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="区间显著标记数")
    effect: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="区间峰效应")
    source: Mapped[str | None] = mapped_column(String(8), nullable=True, default="known", comment="来源(known/gwas)")
    gwas_result_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_gwas_result.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="来源 GWAS 批次(source=gwas 时)"
    )
    remark: Mapped[str | None] = mapped_column(String(512), nullable=True, comment="备注")

    __table_args__ = (
        Index("ix_bre_qtl_created_deleted", "created_time", "is_deleted"),
    )


class MasPanelModel(ModelMixin, UserMixin, MappedBase):
    """MAS 标记辅助选择面板（§8.22）。

    一组与目标性状关联的标记（QTL 峰标记），供决策预览按有利剂量命中童期幼苗。
    """

    __tablename__ = "bre_mas_panel"

    panel_name: Mapped[str | None] = mapped_column(String(128), nullable=True, comment="面板名称")
    trait_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_trait.id", ondelete="CASCADE"),
        index=True, nullable=True, comment="目标性状"
    )
    n_markers: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="标记数")
    remark: Mapped[str | None] = mapped_column(String(512), nullable=True, comment="备注")

    __table_args__ = (
        Index("ix_bre_mas_panel_created_deleted", "created_time", "is_deleted"),
    )


class MasPanelMarkerModel(ModelMixin, UserMixin, MappedBase):
    """MAS 面板标记（§8.22）。"""

    __tablename__ = "bre_mas_panel_marker"

    panel_id: Mapped[int] = mapped_column(
        Integer, ForeignKey("bre_mas_panel.id", ondelete="CASCADE"),
        index=True, nullable=False, comment="关联面板"
    )
    marker_id: Mapped[int | None] = mapped_column(
        Integer, ForeignKey("bre_marker.id", ondelete="SET NULL"),
        index=True, nullable=True, comment="标记"
    )
    favorable_dose: Mapped[int | None] = mapped_column(Integer, nullable=True, comment="有利剂量(0/1/2)")
    effect: Mapped[float | None] = mapped_column(Numeric, nullable=True, comment="效应(绝对值参考)")
    direction: Mapped[str | None] = mapped_column(String(8), nullable=True, default="high", comment="方向(high=剂量≥favorable_dose命中 / low=剂量≤命中)")
    mode: Mapped[str | None] = mapped_column(
        String(16), nullable=True, default="additive",
        comment="基因作用模式：additive加性(剂量≥fav)/dominance显性(剂量≥1高/≤1低)/recessive隐性(剂量=2高/0低)/allele等位存在(有利等位)/haplotype单倍型(同组全命中)")
    favorable_allele: Mapped[str | None] = mapped_column(String(16), nullable=True, comment="有利等位(allele/haplotype 模式用，SSR 等位索引)")
    haplotype_group: Mapped[str | None] = mapped_column(String(32), nullable=True, comment="单倍型组(haplotype 模式：同组标记全命中才计 1)")

    __table_args__ = (
        Index("ix_bre_mas_panel_marker_created_deleted", "created_time", "is_deleted"),
    )