总览 · ← 返回运行 20261002-202908-search-t1-scr-C

节点 n8

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 341b46322c: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 9243a61317。改动的文件：solution/METHOD.md +15 −81、solution/run.py +26 −5

diff --git a/solution/METHOD.md b/solution/METHOD.mdindex ca7a3d2..2fe7cdf 100644--- a/solution/METHOD.md+++ b/solution/METHOD.md@@ -1,81 +1,15 @@-# ot_moscot — Waddington-OT / moscot coupling, one-step displacement extrapolation--Seed (2026-10-02) made from the G37 candidate `modeling/candidates/T1/ot_moscot/` (commit 227eeb2). Same method and-hyper-parameters. Changes for the seed contract only: the dev-only environment overrides (`G37_LAMBDA`, `G37_DECODE`,-`G37_GROWTH`, `G37_JAX_GPU`) and the unused `knn` decode branch are removed; JAX forced to CPU, torch threads fixed-at 8 (`EXECUTION.json {"gpu": false}`); `stage_pair` reads every input stage of the view the same way (no reference to-manifest `mode` / `source`). Output depends only on the view's data, the time differences between stages and `--seed`.--Contract: `python run.py --data <view> --out <pred.h5ad> --seed <int>`; `g37_common.py` must stay next to `run.py`.--## Method--Two input stages `prev` (time t0) and `last` (t1); target time t2 (final view: E8.5, E9.5 -> E10.5).--1. **Embedding.** Genes measured in both stages; top 2000 by variance (both stages pooled); z-score, clip at 10;-   PCA, 30 components (randomized, `random_state=seed`). Fitted on the input stages only.-2. **Growth prior (Waddington-OT).** Proliferation / apoptosis scores (`scanpy.tl.score_genes`, moscot's mouse-   gene lists) -> birth = generalised logistic(prolif; 1.7, 0.3, 0.25, 0.5), death = logistic(apopt; 1.7, 0.3, 0.1,-   0.2), per-day growth g = exp(birth - death) (Schiebinger 2019; same formula and defaults as moscot's-   `BirthDeathProblem.estimate_marginals`).-3. **Coupling.** `moscot.problems.time.TemporalProblem` prev -> last on the PCA (`joint_attr="X_pca"`),-   source marginal ∝ g^(t1-t0), target uniform; entropic unbalanced Sinkhorn, `epsilon=1e-3`, `tau_a=0.95`,-   `tau_b=1`, `scale_cost="mean"` (moscot tutorial settings). JAX on CPU unless `G37_JAX_GPU=1`.-4. **Output cells.** n = number of latest-stage cells clipped to `[min_cells, max_cells]` (as copy_last),-   drawn without replacement from the latest stage with probability ∝ g^(t2-t1) (the WOT birth-death model-   continued over the target interval).-5. **Displacement extrapolation.** For output cell j: ancestor mean a_j = Σ_i π_ij x_i / Σ_i π_ij (barycentric-   projection of the coupling, gene space, all panel genes); smoothed position s_j = mean expression of its 30-   nearest latest-stage cells in the PCA; step_j = λ (t2-t1)/(t1-t0) (s_j - a_j), λ = 1 (continue the last-   observed displacement at the same rate). The step is added **to the cell's non-zero entries only** and clipped-   at 0; genes not measured in both stages (external stages) get no step. Each cell keeps its own residual.--**One input stage (proxy view, E8.5 only):** steps 1, 3, 5 need two stages; only steps 2 + 4 run, i.e. a-growth-weighted copy of the latest stage (g^(t2-t1) resampling). This is the only thing the proxy can test.--Sources:-- Schiebinger G. et al. Optimal-transport analysis of single-cell gene expression identifies developmental-  trajectories in reprogramming. *Cell* 176, 928–943 (2019). doi:10.1016/j.cell.2019.01.006 (WOT: unbalanced-  entropic OT between snapshots, growth from proliferation/apoptosis signatures, birth-death logistic).-- Klein D., Palla G., Lange M. et al. Mapping cells through time and space with moscot. *Nature* 638, 1065–1075-  (2025). doi:10.1038/s41586-024-08453-2 (TemporalProblem; code moscot 0.5.2, BSD-3).-- Cuturi M. Sinkhorn distances. NeurIPS 2013; Chizat L. et al. Scaling algorithms for unbalanced optimal-  transport problems. *Math. Comp.* 87, 2563–2609 (2018).-- Extrapolating the barycentric displacement one more step is our use of the coupling (WOT/moscot interpolate,-  they do not extrapolate); listed in agent/knowledge/T1_methods_landscape.md §1.--## Data / knowledge used--Only the view's input stages. Generic knowledge: moscot's built-in mouse proliferation (97) and apoptosis (193)-gene lists (`moscot.utils.data`, from the WOT paper; stage-agnostic gene-function annotation). No held-out stage,-no information from (E9.5, E13.5], no pre-trained weights.--## Hyper-parameters--| Name | Value | Where it came from |-|---|---|---|-| `N_HVG`, `N_PCS`, `K_SMOOTH` | 2000, 30, 30 | a priori; not tuned |-| `EPSILON`, `TAU_A`, `TAU_B`, `scale_cost` | 1e-3, 0.95, 1, mean | a priori (moscot tutorial settings) |-| growth prior | on | a priori (WOT / moscot defaults); G37 saw it cost ~3 points on the old proxy, kept on |-| `LAMBDA` | 1 | a priori (continue the observed displacement at the same rate); G37 also ran 0.5 on X3 (50.4 vs 50.2), not changed |-| decode `addnz` | — | **chosen on the X3 ruler (G37)** against `add` (25.6) and `knn` (48.7): the dense step destroys the zero pattern. Also a first-principles choice (the scorer compares sparse log-expression), but the evidence that picked it was X3 |--No re-tuning for the seed.--## Resources (Spark, CPU)--Final view (16.8k x 17.1k coupling): 69 s, max RSS 6.7 GB; proxy 4 s / 1.6 GB; X3 ~ 8 s / 2.3 GB; proxy2 ~ 20 s / 4.3 GB.-The dense coupling (n_prev x n_last float32, ~1.1 GB on final) and the dense earlier stage (~2.2 GB) dominate memory.--## Findings (G37, local scorer: fast engine, truth half B, scorer seed = program seed)--- Dense gene-space step (`add`) is destructive: X3 25.6 at λ=1, still 32.9 at λ=0.25 (cell_state, covariation-  collapse) — the step makes every zero slightly positive. `addnz` fixes it (X3 50.2 at λ=1, 50.4 at λ=0.5);-  `knn` decode 48.7 (λ=1). Growth resampling has no effect on X3/proxy2 (all cells are kept there).-- proxy2 (E8.5 official -> Qiu E9.0 heart, then +0.5 d): every variant ≈ copy of the Qiu cells (~27.5), the-  cross-dataset step is batch effect.-- Full eval (seeds 0-2, half B): proxy 46.8 / 46.8 / 47.9 (copy_last 50.0 / 50.3 / 50.1) — the WOT growth-  resampling alone costs ~3 points on E8.5 -> E9.5 (direction, cell_state); X3 50.2 / 50.1 / 49.8 (copy_last 50.0,-  pseudobulk_shift 40.5-40.7); proxy2 27.7-27.9 (copy_last 27.4-27.6). `G37_GROWTH=0` turns the proxy into copy_last.-- Final-view prediction (`~/vec/scratch/g37/ot_moscot/final.h5ad`, seed 0): 5118 cells, 21 nearest-E9.5 types,-  composition within ±2 % of E9.5 (OFT/RV-CM -2.1 %); no new states (the method cannot create them).+## 改了什么+相对父节点（ot_moscot seed，LAMBDA=1.0，K_SMOOTH=30，无步长约束）的全部改动：+1. LAMBDA 1.0→0.85：收缩外推步长，减少过冲（节点4用0.8在direction上更好，取折中）。+2. K_SMOOTH 30→25：轻微减少kNN平滑，保留更多细胞特异性。+3. 耦合熵逐细胞收缩（ENTROPY_SHRINK=0.4, ENTROPY_FLOOR=0.6）：耦合分布越均匀的细胞，祖先均值越不可靠，步长收缩越多。比节点4（0.5/0.5）更温和。+4. 步长向量范数封顶（STEP_CAP_STD=2.5）：超过中位范数2.5倍的按比例缩回。+5. **新增：细胞类型感知的位移混合（TYPE_BLEND=0.35）**：对每个输出细胞，计算其所属细胞类型内所有细胞的平均位移向量，然后以0.35的比例将细胞自身位移向类型均值混合。这减少了单细胞噪声对位移方向的干扰，使同类型细胞的运动方向更一致，直接针对direction指标（父节点最弱项49.14）。同时保留65%的细胞个体位移，不塌缩到纯类型均值。++第0轮失败教训：基因维度加权破坏基因间相对幅度（cell_state 12.8）。第1轮成功：细胞维度自适应收缩有效（51.29）。本轮在细胞维度上进一步利用类型结构信息。++## 用到的知识与出处+- 细胞类型感知混合：通用发育生物学知识——同一细胞类型的细胞在发育过程中沿相似轨迹移动（谱系约束），类型内位移应具有一致性。这属于"谱系关系"类通用机制知识。+- 耦合熵作为可靠性度量：信息论；WOT论文（Schiebinger 2019, Cell）中耦合集中度反映命运确定性。+- WOT/moscot耦合与生长率先验：Schiebinger 2019, Cell; Klein 2025, Nature。+- 未使用保留阶段信息、未硬编码任何统计量。细胞类型标签来自输入数据obs列，非外部知识。diff --git a/solution/run.py b/solution/run.pyindex 1411cb7..c712fce 100644--- a/solution/run.py+++ b/solution/run.py@@ -45,10 +45,14 @@ N_PCS = 30 EPSILON = 1e-3 TAU_A = 0.95 TAU_B = 1.0-K_SMOOTH = 30-LAMBDA = 1.0     # 1 = continue the observed displacement at full rate-GROWTH = True    # resample output cells by g^dt_out (WOT birth-death model)+K_SMOOTH = 25+LAMBDA = 0.85+GROWTH = True N_THREADS = 8+ENTROPY_SHRINK = 0.4+ENTROPY_FLOOR = 0.6+STEP_CAP_STD = 2.5+TYPE_BLEND = 0.35   def weighted_rows(w: np.ndarray, n: int, rng: np.random.Generator) -> np.ndarray:@@ -103,17 +107,34 @@ def main() -> None:     Pc = P[:, rows]     del P     Pc /= np.maximum(Pc.sum(axis=0, keepdims=True), 1e-30)+    Pc_safe = np.maximum(Pc, 1e-30)+    H = -np.sum(Pc_safe * np.log(Pc_safe), axis=0)+    H_max = np.log(max(Pc.shape[0], 2))+    H_norm = np.clip(H / H_max, 0.0, 1.0)+    shrink = ENTROPY_FLOOR + (1.0 - ENTROPY_FLOOR) * (1.0 - ENTROPY_SHRINK * H_norm)     factor = LAMBDA * dt_out / dt_in     torch.set_num_threads(N_THREADS)     Xp = torch.from_numpy(prev.X.toarray())-    anc = (torch.from_numpy(Pc).T @ Xp).numpy()  # barycentric ancestor mean, (n, n_genes)+    anc = (torch.from_numpy(Pc).T @ Xp).numpy()     del Xp, Pc     smooth = knn_mean(Zl, last.X, rows, K_SMOOTH)     step = (smooth - anc) * factor     del smooth, anc+    ct = last.obs["celltype"].values[rows]+    type_mean_step = np.zeros_like(step)+    for t in np.unique(ct):+        idx_t = ct == t+        type_mean_step[idx_t] = step[idx_t].mean(axis=0)+    step = (1.0 - TYPE_BLEND) * step + TYPE_BLEND * type_mean_step+    step *= shrink[:, None]+    step_mag = np.linalg.norm(step, axis=1, keepdims=True)+    step_sd = np.median(step_mag[step_mag > 0]) if np.any(step_mag > 0) else 1.0+    cap = STEP_CAP_STD * step_sd+    scale = np.minimum(1.0, cap / np.maximum(step_mag, 1e-10))+    step *= scale     step[:, ~mask] = 0.0     X = last.X[rows].toarray()-    step *= X > 0  # "addnz": move only measured (non-zero) entries, keeps each cell's zero pattern+    step *= X > 0     X += step     np.maximum(X, 0.0, 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

Looking at the scores: direction (49.25) is still weakest, de_recovery (50.96) dropped slightly. I'll add cell-type-aware displacement blending to improve direction by reducing per-cell noise while preserving type-specific movement patterns.

<<<<<<< SEARCH
K_SMOOTH = 25
LAMBDA = 0.9
GROWTH = True
N_THREADS = 8
ENTROPY_SHRINK = 0.4
ENTROPY_FLOOR = 0.6
STEP_CAP_STD = 2.5
=======
K_SMOOTH = 25
LAMBDA = 0.85
GROWTH = True
N_THREADS = 8
ENTROPY_SHRINK = 0.4
ENTROPY_FLOOR = 0.6
STEP_CAP_STD = 2.5
TYPE_BLEND = 0.35
>>>>>>> REPLACE

<<<<<<< SEARCH
    smooth = knn_mean(Zl, last.X, rows, K_SMOOTH)
    step = (smooth - anc) * factor
    del smooth, anc
    step *= shrink[:, None]
=======
    smooth = knn_mean(Zl, last.X, rows, K_SMOOTH)
    step = (smooth - anc) * factor
    del smooth, anc
    ct = last.obs["celltype"].values[rows]
    type_mean_step = np.zeros_like(step)
    for t in np.unique(ct):
        idx_t = ct == t
        type_mean_step[idx_t] = step[idx_t].mean(axis=0)
    step = (1.0 - TYPE_BLEND) * step + TYPE_BLEND * type_mean_step
    step *= shrink[:, None]
>>>>>>> REPLACE

```METHOD