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

节点 n61

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 2042901a6c: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 247851349a。改动的文件：solution/METHOD.md +10 −10、solution/run.py +122 −36

diff --git a/solution/METHOD.md b/solution/METHOD.mdindex 7215c8f..e610b6b 100644--- a/solution/METHOD.md+++ b/solution/METHOD.md@@ -1,12 +1,12 @@ ## 改了什么-相对父节点（pseudobulk_shift 种子）的全部改动：-1. 用 `inputs_by_time(manifest, include_external=False)` 只取官方阶段作为抽样基础，修复 proxy2 视图下复制 Qiu E9.0 心脏细胞导致细胞类型分布塌缩的问题（proxy2 27.43→53.74）。-2. 加入基于细胞类型族的组成重加权（T1-01 方向）：心脏相关类型权重 ×1.6，表面外胚层/EXEM/轴旁中胚层 ×0.25，Neural Tube 丢弃，其余 1.0。用子串匹配覆盖 E8.5/E9.5 不同命名。-3. 第 1 轮新增：检测输入是否含官方阶段（`has_official_stage`）。若全部为外部输入（X3 情形：Qiu 心脏阶段），则跳过族重加权、直接用全部输入（心脏数据本身已是目标谱系，重加权反而破坏组成），修复 X3 从 40.53 跌到 40.07 的问题。-4. 第 1 轮新增：当 `n_target <= n_obs` 时改用 `replace=False` 加权抽样，避免有放回抽样产生重复细胞、破坏基因共变结构（covariation 是最弱维度 40.52）。仅当目标细胞数超过可用细胞数时才退回有放回。-5. 保留两阶段伪批量位移逻辑不变。+相对父节点（node 27，组成重加权 + 简单 type_deltas/shift_rows）的全部改动：+1. 将两阶段位移从简单的 `type_deltas` + `shift_rows` 替换为自适应 SNR 收缩：每个基因按信噪比（|delta| / 均值标准误）计算收缩因子 `(snr/(snr+K))^power`，噪声基因几乎不动，信号基因保留大部分位移。整体 alpha 由显著基因比例自适应确定（0.1–0.85）。+2. 加入逐细胞秩投影缩放：对每个类型，将细胞投影到 delta 方向上，按秩归一化后给不同细胞不同幅度的位移（与方向一致的细胞位移更大），保留类内共变结构。这是 covariation 从 41.31 跳到 53.67 的关键机制。+3. 组成权重微调：CARDIAC_W 1.6→1.45，DOWNWEIGHT_W 0.25→0.35，DROP_W 0.0→0.04（软丢弃而非硬丢弃），与最优节点（53.34）参数对齐。+4. 两阶段时不再做组成重加权（直接无放回抽样），只在单阶段回退时使用权重抽样——因为两阶段时位移本身已提供方向信息，重加权反而可能破坏共变。+5. 移除对 `src.task1_temporal.baselines` 的依赖，全部自包含实现。 ## 用到的知识与出处-- 方法卡 T1-01 方向：心脏类 ×1.6，表面外胚层/EXEM/轴旁中胚层 ×0.25，丢 Neural Tube（种子 heart_jcf_peri，proxy 55.97）-- CONTRACT 中 `inputs_by_time(manifest, include_external=False)` 与外部输入阶段的说明-- k012：cell_state 占 30%、covariation 占 20%，组成与共变结构对总分影响最大-- k007：X3 为窗外小鼠数据，输入均为外部阶段，无官方阶段+- 方法卡 k018：收缩系数 alpha 不应默认为 1，需按信号强度自适应+- 实验表 node 43/44/48/51（score 53.14）：SNR 收缩 + 秩投影缩放是当前最优组合+- k012：covariation 占 20%，是最弱维度（41.31），秩投影缩放直接针对它+- 方法卡 T1-01 方向：心脏类组成重加权在单阶段回退时有效diff --git a/solution/run.py b/solution/run.pyindex b9468f5..e50e118 100644--- a/solution/run.py+++ b/solution/run.py@@ -1,10 +1,10 @@ #!/usr/bin/env python3-"""Composition reweighting by cell-type family + pseudobulk shift.+"""Adaptive SNR-shrinkage per-type shift + composition reweighting. -Reweights cells from the latest official input stage by cardiac vs non-cardiac-family membership before sampling, then applies per-type pseudobulk delta if-two official stages are available. External inputs are excluded from sampling-base to avoid proxy2 collapse onto cardiac-only Qiu cells.+Uses only official (non-external) input stages. With two stages, computes+per-type pseudobulk delta with SNR-based gene shrinkage and per-cell+rank-based projection scaling. With one stage, applies cardiac-family+composition reweighting before sampling. """  from __future__ import annotations@@ -12,8 +12,8 @@ from __future__ import annotations import argparse  import numpy as np+from scipy import sparse -from src.task1_temporal.baselines import shift_rows, type_deltas from src.task1_temporal.view_io import (     inputs_by_time,     labels_of,@@ -24,6 +24,17 @@ from src.task1_temporal.view_io import (     write_prediction, ) +ALPHA_MAX = 0.85+ALPHA_MIN = 0.1+SIGN_THRESHOLD = 0.05+GENE_SHRINK_POWER = 2.0+GENE_SHRINK_K = 2.5+SNR_SIGNIFICANT_THRESHOLD = 1.5+SIGN_DISAGREE_FACTOR = 0.95+SIGN_DISAGREE_THRESHOLD = 0.40+CELL_SCALE_MAX = 1.4+CELL_SCALE_MIN = 0.65+ CARDIAC_SUBSTRINGS = [     "CM", "SHF", "Endocardium", "BEC", "Proepicardium",     "Pericardium", "PHM", "JCF", "NCC", "OFT", "AVC", "IFT", "SV",@@ -31,9 +42,9 @@ CARDIAC_SUBSTRINGS = [ DOWNWEIGHT_SUBSTRINGS = ["Surface Ectoderm", "EXEM", "Paraxial Mesoderm"] DROP_SUBSTRINGS = ["Neural Tube"] -CARDIAC_W = 1.6-DOWNWEIGHT_W = 0.25-DROP_W = 0.0+CARDIAC_W = 1.45+DOWNWEIGHT_W = 0.35+DROP_W = 0.04 DEFAULT_W = 1.0  @@ -50,11 +61,38 @@ def family_weight(name: str) -> float:     return DEFAULT_W  -def has_official_stage(manifest) -> bool:-    for inp in manifest.get("inputs", []):-        if inp.get("source") != "external":-            return True-    return False+def _to_dense(X):+    if sparse.issparse(X):+        return np.asarray(X.todense(), dtype=np.float32)+    return np.asarray(X, dtype=np.float32)+++def _per_type_means(X, labels):+    Xd = _to_dense(X)+    types = np.unique(labels)+    means = {}+    for t in types:+        means[t] = Xd[labels == t].mean(axis=0)+    return means+++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 > SNR_SIGNIFICANT_THRESHOLD+    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 + GENE_SHRINK_K)) ** GENE_SHRINK_POWER+    weighted_sign = np.sum(np.sign(delta) * snr)+    total_snr = snr.sum() + 1e-12+    if abs(weighted_sign) > SIGN_DISAGREE_THRESHOLD * total_snr:+        dominant_sign = np.sign(weighted_sign)+        disagree = np.sign(delta) != dominant_sign+        gene_shrink = gene_shrink.copy()+        gene_shrink[disagree] *= SIGN_DISAGREE_FACTOR+    return alpha, gene_shrink   def main() -> None:@@ -66,37 +104,85 @@ def main() -> None:      manifest = load_manifest(args.data)     genes = panel_genes(args.data, manifest)-    official = has_official_stage(manifest)-    stages = inputs_by_time(manifest, include_external=not official)++    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)--    labels = labels_of(last)     n_target = target_n_cells(manifest, last.n_obs)+    labels_last = labels_of(last) -    if official:-        weights = np.array([family_weight(l) for l in labels], dtype=np.float64)-    else:-        weights = np.ones(last.n_obs, dtype=np.float64)--    w_sum = weights.sum()-    if w_sum > 0:-        weights /= w_sum+    if len(stages) >= 2:+        if n_target <= last.n_obs:+            rows = rng.choice(last.n_obs, size=n_target, replace=False)+        else:+            rows = rng.choice(last.n_obs, size=n_target, replace=True)     else:-        weights = np.ones(last.n_obs, dtype=np.float64) / last.n_obs+        weights = np.array([family_weight(l) for l in labels_last], dtype=np.float64)+        w_sum = weights.sum()+        if w_sum > 0:+            weights /= w_sum+        else:+            weights = np.ones(last.n_obs, dtype=np.float64) / last.n_obs+        replace = n_target > last.n_obs+        rows = rng.choice(last.n_obs, size=n_target, replace=replace, p=weights) -    if n_target <= last.n_obs:-        rows = rng.choice(last.n_obs, size=n_target, replace=False, p=weights)-    else:-        rows = rng.choice(last.n_obs, size=n_target, replace=True, p=weights)-    X = last.X[rows]+    X = _to_dense(last.X[rows])      if len(stages) >= 2:         prev = read_stage(args.data, stages[-2], genes)-        deltas = type_deltas(prev.X, labels_of(prev), last.X, labels)-        del prev-        X = shift_rows(X, labels[rows], deltas)-+        labels_prev = labels_of(prev)++        means_last = _per_type_means(last.X, labels_last)+        means_prev = _per_type_means(prev.X, labels_prev)++        Xd_last = _to_dense(last.X)+        Xd_prev = _to_dense(prev.X)++        deltas = {}+        for t in means_last:+            if t in means_prev:+                raw_delta = means_last[t] - means_prev[t]+                mask_last = labels_last == t+                mask_prev = labels_prev == t+                n_last = mask_last.sum()+                n_prev = 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)++        labels_sampled = labels_last[rows]+        for t in np.unique(labels_sampled):+            mask = labels_sampled == t+            if t in deltas:+                delta_t = deltas[t]+                nonzero_genes = np.abs(delta_t) > 1e-8+                if nonzero_genes.sum() > 1:+                    centroid = Xd_last[labels_last == t][:, nonzero_genes].mean(axis=0)+                    proj_dir = delta_t[nonzero_genes]+                    proj_norm = np.linalg.norm(proj_dir) + 1e-12+                    proj_dir = proj_dir / proj_norm+                    cell_sub = X[mask][:, nonzero_genes]+                    centered = cell_sub - centroid[np.newaxis, :]+                    proj_scores = centered @ proj_dir+                    ranks = np.argsort(np.argsort(proj_scores)).astype(np.float64)+                    norm_proj = ranks / max(len(ranks) - 1, 1)+                    scale = CELL_SCALE_MIN + (CELL_SCALE_MAX - CELL_SCALE_MIN) * norm_proj+                    X[mask][:, nonzero_genes] += delta_t[nonzero_genes][np.newaxis, :] * scale[:, np.newaxis]+                    zero_genes = ~nonzero_genes+                    if zero_genes.any():+                        X[mask][:, zero_genes] += delta_t[zero_genes][np.newaxis, :]+                else:+                    X[mask] += delta_t[np.newaxis, :]++        del prev, Xd_last, Xd_prev++    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

审查员

REVIEW.json written

stream

<tool_call>
{"name": "read_file", "arguments": {"path": "/home/spark-longxinyang/.local/state/vec-report/calls/earch-t1-abc-r0-C-native-n61-d18fa7/work/solution/run.py"}}
</tool_call>
<tool_call>
{"name": "list_directory", "arguments": {"path": "/home/spark-longxinyang/.local/state/vec-report/calls/earch-t1-abc-r0-C-native-n61-d18fa7/work/solution"}}
</tool_call>