总览 · ← 返回运行 20261001-233757-search-t1-abc-r0-C-native

节点 n12

改了什么

方法说明?节点程序自带的 METHOD.md：这个程序做了什么、为什么。

来自 programs.git 2642d11d32:solution/METHOD.md

调研员的计划

代码改动?这个节点的程序和父节点程序的逐行差别：绿色是新增，红色是删除。

对比：父节点版本 01e7938a0f。改动的文件：solution/METHOD.md +9 −9、solution/run.py +69 −45

diff --git a/solution/METHOD.md b/solution/METHOD.mdindex 92e8c97..45fb46a 100644--- a/solution/METHOD.md+++ b/solution/METHOD.md@@ -1,11 +1,11 @@ ## 改了什么-1. **Per-gene empirical Bayes shrinkage 替代统一 ALPHA**：对每个细胞类型的 pseudobulk delta，按基因的信噪比（delta² / (delta² + noise_variance)）做逐基因收缩。噪声方差用两个阶段的 within-type 方差除以各自细胞数估计。信号强的基因（大 delta、低方差）保留更多位移；噪声基因被压向零，减少对原始共变结构的破坏。这直接针对 covariation（当前最弱组 40.74）：统一 alpha 对所有基因一视同仁，噪声基因的随机位移会破坏 gene-gene 相关；逐基因收缩只在有信号的方向移动，保留无信号基因的原始协方差。-2. **ALPHA_GLOBAL 从 0.5 调到 0.6**：在逐基因收缩已经提供了自适应降噪的前提下，全局系数可以稍微放松，让高信噪比基因获得更充分的位移，改善 direction 和 de_recovery。-3. **保留 include_external=False 过滤**（父节点已验证 proxy2 +22.61）。-4. **移除对 baselines.shift_rows / type_deltas 的依赖**，改为自行计算 per-type 均值/方差和逐基因收缩，以获得更细粒度的控制。-5. **单输入阶段退化为 copy_last**（与父节点一致）。+1. **替换统一 EB 收缩为自适应 per-type alpha（SNR 驱动）**：采用实验表中最佳节点（Program 1, score 48.91, covariation 46.18）的机制——对每个类型计算逐基因 SNR（|delta| / noise_se），以 SNR>2 的基因占比决定该类型的 alpha（从 ALPHA_MIN=0.1 线性插值到 ALPHA_MAX=0.75），再乘以逐基因收缩 (snr/(snr+2))^1.5。噪声主导的类型几乎不动（保护协方差），信号一致的类型获得更大位移。+2. **新增事后 per-type 方差重标定**：shift+clip 后，对每个类型逐基因计算残差标准差，与原始（shift 前）标准差比较，按比例缩放残差以恢复原始方差。缩放因子限制在 [0.5, 2.0] 防止极端值。这直接针对 covariation（当前最弱组 41.90）：零截断会压缩低表达基因的方差、破坏 gene-gene 相关；重标定在保持均值位移的同时恢复方差结构。+3. **移除全局 ALPHA_GLOBAL=0.6**，改由自适应机制逐类型决定位移幅度。+4. **保留 include_external=False 过滤**（已验证 proxy2 +22.61）。+5. **单输入阶段退化为 copy_last**（与所有父节点一致）。 ## 用到的知识与出处-- k018（方法卡 T1）：「Optionally shrink per gene (empirical Bayes on the delta's variance) so noisy genes move less」——本实现直接采用此建议。-- k018：「The official constant per-type shift (alpha=1) scores below copy_last on T1, so re-applying the last observed delta must be shrunk by alpha in [0,1]」。-- 父节点 ANALYSIS next_suggestions：「针对 covariation：在 shift_rows 后按 per-type 残差重标定」——本方案从源头（delta 计算）而非事后重标定来解决，因为逐基因收缩等价于在平移前就保护了方差结构。-- James-Stein shrinkage 原理（Efron & Morris 1975, "Data analysis using Stein's estimator and its generalizations"）：观测值向零收缩的最优系数为 signal/(signal+noise)。+- k018（方法卡 T1）：「Optionally shrink per gene (empirical Bayes on the delta's variance) so noisy genes move less」；本实现进一步升级为 SNR 驱动的 per-type alpha。+- 父节点 ANALYSIS next_suggestions：「针对 covariation：改试事后重标定——shift 后按 per-type 残差把方差/相关拉回 last 阶段」——本方案直接实现此建议。+- 实验表 Program 1（score 48.91, covariation 46.18）：验证了 SNR 符号一致性 per-type alpha 在 covariation 上的优势。+- James-Stein shrinkage 原理（Efron & Morris 1975, "Data analysis using Stein's estimator and its generalizations"）：观测值向零收缩的最优系数为 signal/(signal+noise)，此处推广为 (snr/(snr+2))^power。diff --git a/solution/run.py b/solution/run.pyindex ce643fc..e0a6521 100644--- a/solution/run.py+++ b/solution/run.py@@ -1,11 +1,11 @@ #!/usr/bin/env python3-"""Per-gene empirical Bayes shrinkage on pseudobulk deltas.+"""Adaptive per-type shrinkage with post-shift variance restoration. -Uses only official (non-external) input stages. For each cell type's-pseudobulk delta, applies per-gene shrinkage based on the signal-to-noise-ratio: genes with large deltas relative to their within-type variance are-shifted more; noisy genes are shrunk toward zero, preserving the original-covariation structure better than a uniform alpha.+Combines adaptive per-type alpha (SNR-based) from the best-scoring node+with post-shift per-type variance rescaling to protect covariation.+Types with coherent transcriptional shifts are shifted more; noisy types+stay near copy_last. After shifting and clipping, per-type per-gene+residuals are rescaled to restore the original standard deviation.  With a single official input stage, falls back to copy_last. """@@ -28,7 +28,12 @@ from src.task1_temporal.view_io import (     write_prediction, ) -ALPHA_GLOBAL = 0.6+ALPHA_MAX = 0.75+ALPHA_MIN = 0.1+SIGN_THRESHOLD = 0.05+GENE_SHRINK_POWER = 1.5+SCALE_CLIP_LO = 0.5+SCALE_CLIP_HI = 2.0   def _to_dense(X):@@ -37,27 +42,16 @@ def _to_dense(X):     return np.asarray(X, dtype=np.float32)  -def _per_type_stats(X, labels):-    Xd = _to_dense(X)-    types = np.unique(labels)-    means = {}-    variances = {}-    counts = {}-    for t in types:-        mask = labels == t-        sub = Xd[mask]-        means[t] = sub.mean(axis=0)-        variances[t] = sub.var(axis=0)-        counts[t] = mask.sum()-    return means, variances, counts---def _shrink_deltas_eb(delta, var_last, n_last, var_prev, n_prev):-    noise_var = var_last / max(n_last, 1) + var_prev / max(n_prev, 1)-    signal = delta ** 2-    denom = signal + noise_var + 1e-10-    shrink = signal / denom-    return delta * shrink+def _adaptive_alpha(delta, gene_var_last, gene_var_prev, n_last, n_prev):+    noise_se = np.sqrt(gene_var_last / max(n_last, 1) + gene_var_prev / max(n_prev, 1) + 1e-12)+    snr = np.abs(delta) / (noise_se + 1e-12)+    significant = snr > 2.0+    frac_significant = significant.mean()+    if frac_significant < SIGN_THRESHOLD:+        return ALPHA_MIN, np.ones_like(delta)+    alpha = ALPHA_MIN + (ALPHA_MAX - ALPHA_MIN) * min(frac_significant / 0.3, 1.0)+    gene_shrink = (snr / (snr + 2.0)) ** GENE_SHRINK_POWER+    return alpha, gene_shrink   def main() -> None:@@ -69,46 +63,76 @@ def main() -> None:      manifest = load_manifest(args.data)     genes = panel_genes(args.data, manifest)+     stages = inputs_by_time(manifest, include_external=False)+    if not stages:+        stages = inputs_by_time(manifest)+     last = read_stage(args.data, stages[-1], genes)     rng = np.random.default_rng(args.seed)     rows = sample_rows(last.n_obs, target_n_cells(manifest, last.n_obs), rng)-    X = last.X[rows]-    if hasattr(X, "toarray"):-        X = X.toarray()-    X = np.asarray(X, dtype=np.float32)+    X = _to_dense(last.X[rows])      if len(stages) >= 2:         prev = read_stage(args.data, stages[-2], genes)         labels_last = labels_of(last)         labels_prev = labels_of(prev) -        means_last, vars_last, counts_last = _per_type_stats(last.X, labels_last)-        means_prev, vars_prev, counts_prev = _per_type_stats(prev.X, labels_prev)-        del prev+        Xd_last = _to_dense(last.X)+        Xd_prev = _to_dense(prev.X)++        types_last = np.unique(labels_last)+        types_prev_set = set(np.unique(labels_prev))++        means_last = {}+        for t in types_last:+            means_last[t] = Xd_last[labels_last == t].mean(axis=0)+        means_prev = {}+        for t in types_prev_set:+            means_prev[t] = Xd_prev[labels_prev == t].mean(axis=0)++        orig_stds = {}+        for t in types_last:+            orig_stds[t] = Xd_last[labels_last == t].std(axis=0)          deltas = {}-        for t in means_last:+        for t in types_last:             if t in means_prev:                 raw_delta = means_last[t] - means_prev[t]-                shrunk = _shrink_deltas_eb(-                    raw_delta,-                    vars_last[t], counts_last[t],-                    vars_prev[t], counts_prev[t],-                )-                deltas[t] = shrunk * ALPHA_GLOBAL+                mask_last = labels_last == t+                mask_prev = labels_prev == t+                n_last = int(mask_last.sum())+                n_prev = int(mask_prev.sum())+                var_last = Xd_last[mask_last].var(axis=0)+                var_prev = Xd_prev[mask_prev].var(axis=0)+                alpha, gene_shrink = _adaptive_alpha(raw_delta, var_last, var_prev, n_last, n_prev)+                deltas[t] = raw_delta * alpha * gene_shrink             else:                 deltas[t] = np.zeros(len(genes), dtype=np.float32) +        del prev, Xd_prev+         labels_sampled = labels_last[rows]-        delta_matrix = np.zeros((X.shape[0], X.shape[1]), dtype=np.float32)         for t in np.unique(labels_sampled):             mask = labels_sampled == t             if t in deltas:-                delta_matrix[mask] = deltas[t]-        X = X + delta_matrix+                X[mask] += deltas[t]+         np.clip(X, 0.0, None, out=X) +        for t in np.unique(labels_sampled):+            mask = labels_sampled == t+            if t in orig_stds:+                shifted_mean = X[mask].mean(axis=0)+                residuals = X[mask] - shifted_mean+                curr_std = residuals.std(axis=0)+                scale = np.where(curr_std > 1e-8, orig_stds[t] / (curr_std + 1e-8), 1.0)+                scale = np.clip(scale, SCALE_CLIP_LO, SCALE_CLIP_HI)+                X[mask] = shifted_mean + residuals * scale++        del Xd_last++    np.clip(X, 0.0, None, out=X)     write_prediction(X, genes, args.out, seed=args.seed)  

调研来源?调研员查到并用到的知识条目和文献检索结果（只列标题和编号）。

没有记录调研来源。

分析结果?分析员写的 ANALYSIS.json：改了什么、各组分数怎么变、假设是否成立、经验和下一步建议。

对话摘要?每个角色和大模型对话的统计：轮数、工具调用、用时、token 数和最后的回答摘录；原始记录只给路径。

只给统计和最后回答的摘录；完整对话请到原始记录位置里列出的文件看。

分析员

ANALYSIS.json written

native

stream

<<<<<<< SEARCH
ALPHA_MAX = 0.75
ALPHA_MIN = 0.1
SIGN_THRESHOLD = 0.05
GENE_SHRINK_POWER = 1.5
SCALE_CLIP_LO = 0.5
SCALE_CLIP_HI = 2.0
=======
ALPHA_MAX = 0.85
ALPHA_MIN = 0.1
SIGN_THRESHOLD = 0.04
GENE_SHRINK_POWER = 1.0
SCALE_CLIP_LO = 0.5
SCALE_CLIP_HI = 2.0
>>>>>>> REPLACE

<<<<<<< SEARCH
    noise_se = np.sqrt(gene_var_last / max(n_last, 1) + gene_var_prev / max(n_prev, 1) + 1e-12)
    snr = np.abs(delta) / (noise_se + 1e-12)
    significant = snr > 2.0
    frac_significant = significant.mean()
    if frac_significant < SIGN_THRESHOLD:
        return ALPHA_MIN, np.ones_like(delta)
    alpha = ALPHA_MIN + (ALPHA_MAX - ALPHA_MIN) * min(frac_significant / 0.3, 1.0)
    gene_shrink = (snr / (snr + 2.0)) ** GENE_SHRINK_POWER
    return alpha, gene_shrink
=======
    noise_se = np.sqrt(gene_var_last / max(n_last, 1) + gene_var_prev / max(n_prev, 1) + 1e-12)
    snr = np.abs(delta) / (noise_se + 1e-12)
    significant = snr > 1.5
    frac_significant = significant.mean()
    if frac_significant < SIGN_THRESHOLD:
        return ALPHA_MIN, np.ones_like(delta)
    alpha = ALPHA_MIN + (ALPHA_MAX - ALPHA_MIN) * min(frac_significant / 0.25, 1.0)
    gene_shrink = (snr / (snr + 2.0)) ** GENE_SHRINK_POWER
    return alpha, gene_shrink
>>>>>>> REPLACE

```METHOD
## 改了什么