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

节点 n28

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 28cb57d3b8: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：这个提交的上一版（种子程序：相对空仓库）。改动的文件：solution/EXECUTION.json +1 −0、solution/METHOD.md +12 −0、solution/README.md +4 −0、solution/g37_common.py +101 −0、solution/run.py +143 −0

diff --git a/solution/EXECUTION.json b/solution/EXECUTION.jsonnew file mode 100644index 0000000..9d5125c--- /dev/null+++ b/solution/EXECUTION.json@@ -0,0 +1 @@+{"gpu": false}diff --git a/solution/METHOD.md b/solution/METHOD.mdnew file mode 100644index 0000000..6b9b4f9--- /dev/null+++ b/solution/METHOD.md@@ -0,0 +1,12 @@+## 改了什么+从 ot_moscot（Waddington-OT 耦合 + 重心位移外推）改为 graph_fate 家族的最小实现：在两个输入阶段的联合 PCA 上建 kNN 图，用 DPT 伪时间（根选为增殖得分最高的细胞）驱动 CellRank PseudotimeKernel 得到转移矩阵，每个输出细胞沿其图后继（最新阶段细胞）的加权均值方向移动一步。与 OT 方法的区别：后继状态来自图上的局部转移，同型细胞可走向不同方向；不依赖全局耦合矩阵。保留了生长率加权重抽样和 addnz 解码。机制对照：MECHANISM_ON=False 时不做前推，等价于生长加权的 copy_last。单输入阶段（proxy）退化为生长加权重抽样。++第 1 轮修复：CellRank transition_matrix 返回稀疏矩阵，.sum(axis=1, keepdims=True) 不支持，改为 .todense() 转稠密再操作。++第 2 轮修复：第 1 轮 cell_state=3.5、covariation=3.26 极低，原因是 succ_mean 是全群加权均值（过于平滑），LAMBDA=1 时步长过大导致细胞身份和共变结构被破坏。修改：(1) LAMBDA 从 1.0 降到 0.15，保留方向信号但不过度平滑；(2) 对每个细胞的转移行只保留 top-30 后继（稀疏化），使后继均值更局部、更少平滑效应。方向分（56.9）已高于父节点（49.14），说明图转移方向正确，只需控制步长幅度。+## 用到的知识与出处+- k041：scanpy DPT + CellRank PseudotimeKernel 的离线工具链用法（Wolf et al., Genome Biol 2019, 10.1186/s13059-019-1663-x; Weiler et al., Nat Methods 2024, 10.1038/s41592-024-02303-9）。+- k038：T1 无 spliced/unspliced，不能用 velocity kernel，用 PseudotimeKernel 替代。+- 根选择：用增殖得分（moscot 小鼠增殖基因列表，来自 WOT 论文 Schiebinger et al., Cell 2019, 10.1016/j.cell.2019.01.006）最高的细胞作 DPT 根，不依赖目标数据。+- 方向库 graph_fate 条目的最小实现说明。+- addnz 解码与步长控制来自父节点 ot_moscot 的经验（G37 候选）。diff --git a/solution/README.md b/solution/README.mdnew file mode 100644index 0000000..dc1cbaf--- /dev/null+++ b/solution/README.md@@ -0,0 +1,4 @@+# ot_moscot++Waddington-OT / moscot：两个最新输入阶段在联合 PCA 上做非平衡熵 OT 耦合（WOT 增殖/凋亡先验生长），最新阶段按 g^Δt 重抽样，每个细胞沿“kNN 平滑位置 − 耦合祖先均值”再走一个等比例步长，只加在非零基因上，夹到 ≥0。+来自 G37 候选 modeling/candidates/T1/ot_moscot；addnz 解码是在 X3 上选的（详见 METHOD.md）。纯 CPU，final 约 70 s、6.7 GB。diff --git a/solution/g37_common.py b/solution/g37_common.pynew file mode 100644index 0000000..8c991b3--- /dev/null+++ b/solution/g37_common.py@@ -0,0 +1,101 @@+"""Helpers of the ot_moscot seed (copy of modeling/candidates/T1/ot_moscot/g37_common.py, G37; stage_pair reads every+input stage of the view).++- ``stage_pair``: the two latest input stages (official, or external in proxy2 / test views) and the gene mask on+  which both are really measured (external stages lack part of the panel; filled columns must not drive dynamics);+- ``embed``: HVG -> z-score (clip 10) -> PCA, fitted on the input stages only (scanpy-style preprocessing);+- ``knn_mean``: per-cell kNN average in the embedding (smooths single-cell noise before taking displacements);+- ``growth_rates``: Waddington-OT / moscot prior growth from proliferation and apoptosis gene scores.+"""++from __future__ import annotations++import numpy as np+from scipy import sparse++from src.task1_temporal.view_io import covered_mask, inputs_by_time, read_stage+++def stage_pair(view, manifest: dict, genes: list[str]):+    """(prev, last, mask, dt_in) for the two latest inputs; prev None with a single input stage."""+    stages = inputs_by_time(manifest, include_external=True)  # every input stage, whatever its source+    last_e = stages[-1]+    last = read_stage(view, last_e, genes)+    mask = covered_mask(view, last_e, genes)+    if len(stages) < 2:+        return None, last, mask, None, last_e+    prev_e = stages[-2]+    prev = read_stage(view, prev_e, genes)+    mask &= covered_mask(view, prev_e, genes)+    return prev, last, mask, float(last_e["time"] - prev_e["time"]), last_e+++def _gene_moments(X: sparse.csr_matrix):+    n = X.shape[0]+    mean = np.asarray(X.mean(axis=0)).ravel()+    sq = np.asarray(X.multiply(X).sum(axis=0)).ravel() / n+    return mean, np.maximum(sq - mean ** 2, 0.0)+++def embed(mats: list, mask: np.ndarray, n_hvg: int, n_pcs: int, seed: int):+    """PCA coordinates of each matrix in ``mats`` (same order), fitted on all of them together.++    Returns (list of Z, info) with info = {hvg, mean, std, components} so displacements can be decoded."""+    from sklearn.decomposition import PCA++    allX = sparse.vstack(mats).tocsr()+    mean, var = _gene_moments(allX)+    var = np.where(mask, var, -1.0)+    hvg = np.sort(np.argsort(var)[::-1][:n_hvg])+    sub = allX[:, hvg].toarray().astype(np.float32)+    mu, sd = mean[hvg].astype(np.float32), np.sqrt(var[hvg]).astype(np.float32)+    sd[sd == 0] = 1.0+    sub -= mu+    sub /= sd+    np.clip(sub, -10, 10, out=sub)+    pca = PCA(n_components=n_pcs, svd_solver="randomized", random_state=seed)+    Z = pca.fit_transform(sub).astype(np.float32)+    out, start = [], 0+    for M in mats:+        out.append(Z[start:start + M.shape[0]])+        start += M.shape[0]+    return out, {"hvg": hvg, "mu": mu, "sd": sd, "components": pca.components_.astype(np.float32)}+++def knn_mean(Z: np.ndarray, X: sparse.csr_matrix, rows: np.ndarray, k: int) -> np.ndarray:+    """Dense (len(rows), n_genes): mean expression of the k nearest cells (in Z, the cell itself included)."""+    from sklearn.neighbors import NearestNeighbors++    nn = NearestNeighbors(n_neighbors=k).fit(Z)+    idx = nn.kneighbors(Z[rows], return_distance=False)+    W = sparse.csr_matrix((np.full(idx.size, 1.0 / k, dtype=np.float32), idx.ravel(),+                           np.arange(0, idx.size + 1, k)), shape=(len(rows), Z.shape[0]))+    return np.asarray((W @ X).todense(), dtype=np.float32)+++# Waddington-OT prior growth (Schiebinger et al. 2019, Cell; as implemented in moscot.base.problems.birth_death):+# birth = generalised logistic of the proliferation score, death = of the apoptosis score, g = exp(birth - death) / day.+def _gen_logistic(p, sup, inf, center, width):+    return inf + (sup - inf) / (1 + np.exp(-(p - center) / width))+++def growth_rates(adata_list: list, genes: list[str], mask: np.ndarray, seed: int) -> list[np.ndarray]:+    """Per-day growth rate of every cell of each AnnData (scores computed on all of them jointly)."""+    import anndata as ad+    import scanpy as sc+    from moscot.utils.data import apoptosis_markers, proliferation_markers++    ok = set(np.asarray(genes)[mask])+    joint = ad.concat(adata_list, join="inner")+    prolif = [g for g in proliferation_markers("mouse") if g in ok]+    apopt = [g for g in apoptosis_markers("mouse") if g in ok]+    sc.tl.score_genes(joint, prolif, score_name="proliferation", random_state=seed)+    sc.tl.score_genes(joint, apopt, score_name="apoptosis", random_state=seed)+    birth = _gen_logistic(joint.obs["proliferation"].to_numpy(float), 1.7, 0.3, 0.25, 0.5)+    death = _gen_logistic(joint.obs["apoptosis"].to_numpy(float), 1.7, 0.3, 0.1, 0.2)+    g = np.exp(birth - death)+    out, start = [], 0+    for a in adata_list:+        out.append(g[start:start + a.n_obs])+        start += a.n_obs+    return outdiff --git a/solution/run.py b/solution/run.pynew file mode 100644index 0000000..88c1d85--- /dev/null+++ b/solution/run.py@@ -0,0 +1,143 @@+#!/usr/bin/env python3+"""graph_fate draft: kNN graph + CellRank PseudotimeKernel transition, fate-weighted displacement extrapolation.++Two input stages:+  1. joint PCA of both stages (HVG, z-score, 30 PCs; fitted on the inputs only);+  2. kNN graph on the joint embedding, DPT pseudotime rooted at the most progenitor-like cell+     (highest proliferation score), CellRank PseudotimeKernel -> transition matrix;+  3. output cells = latest-stage cells resampled with weights g^dt_out (WOT birth-death prior growth);+  4. each output cell j moves toward its graph-successor weighted mean expression in the latest stage:+     step = LAMBDA * dt_out / dt_in * (successor_mean_j - x_j), applied to non-zero entries only, clipped >= 0.+     This differs from the OT approach: successor states come from local graph transitions, so cells of the+     same type can move in different directions based on their neighborhood.+One input stage (proxy): only growth-weighted resampling runs (same as parent).++Mechanism control: MECHANISM_ON=False replaces the transition with identity (no forward push),+equivalent to growth-weighted copy_last.+"""++from __future__ import annotations++import argparse+import os+import sys+from pathlib import Path++os.environ.setdefault("XLA_PYTHON_CLIENT_PREALLOCATE", "false")+os.environ.setdefault("XLA_PYTHON_CLIENT_MEM_FRACTION", "0.1")+os.environ["JAX_PLATFORMS"] = "cpu"  # CPU only: deterministic, no GPU slot++import numpy as np++sys.path.insert(0, str(Path(__file__).resolve().parent))+from g37_common import embed, growth_rates, knn_mean, stage_pair  # noqa: E402++from src.task1_temporal.view_io import load_manifest, panel_genes, target_n_cells, write_prediction  # noqa: E402++N_HVG = 2000+N_PCS = 30+K_GRAPH = 30+K_SMOOTH = 30+LAMBDA = 0.15+TOP_K_SUCC = 30+GROWTH = True+N_THREADS = 8+MECHANISM_ON = True+++def weighted_rows(w: np.ndarray, n: int, rng: np.random.Generator) -> np.ndarray:+    p = w / w.sum()+    return np.sort(rng.choice(len(w), size=n, replace=n > len(w), p=p))+++def main() -> None:+    parser = argparse.ArgumentParser()+    parser.add_argument("--data", required=True)+    parser.add_argument("--out", required=True)+    parser.add_argument("--seed", type=int, default=0)+    args = parser.parse_args()++    manifest = load_manifest(args.data)+    genes = panel_genes(args.data, manifest)+    prev, last, mask, dt_in, last_e = stage_pair(args.data, manifest, genes)+    dt_out = float(manifest["target"]["time"] - last_e["time"])+    rng = np.random.default_rng(args.seed)+    n = target_n_cells(manifest, last.n_obs)++    if prev is None:+        (g_last,) = growth_rates([last], genes, mask, args.seed)+        rows = weighted_rows(g_last ** dt_out if GROWTH else np.ones(last.n_obs), n, rng)+        write_prediction(last.X[rows], genes, args.out, seed=args.seed)+        return++    import anndata as ad+    import scanpy as sc+    import torch++    torch.set_num_threads(N_THREADS)++    (Zp, Zl), _ = embed([prev.X, last.X], mask, N_HVG, N_PCS, args.seed)+    g_prev, g_last = growth_rates([prev, last], genes, mask, args.seed)++    n_prev = prev.n_obs+    n_last = last.n_obs+    Z_all = np.vstack([Zp, Zl])++    adata = ad.AnnData(+        X=np.zeros((n_prev + n_last, 1), dtype=np.float32),+        obs={"time": np.r_[np.zeros(n_prev), np.ones(n_last)]},+    )+    adata.obsm["X_pca"] = Z_all++    sc.pp.neighbors(adata, use_rep="X_pca", n_neighbors=K_GRAPH, random_state=args.seed)++    prolif_all = np.r_[g_prev, g_last]+    root_idx = int(np.argmax(prolif_all))+    adata.uns["iroot"] = root_idx+    sc.tl.dpt(adata, n_branchings=0)++    if not MECHANISM_ON:+        rows = weighted_rows(g_last ** dt_out if GROWTH else np.ones(n_last), n, rng)+        write_prediction(last.X[rows], genes, args.out, seed=args.seed)+        return++    import cellrank as cr++    adata.obs["dpt_pseudotime"] = adata.obs["dpt_pseudotime"].values.astype(np.float64)+    vk = cr.kernels.PseudotimeKernel(adata, time_key="dpt_pseudotime")+    vk.compute_transition_matrix()+    T = vk.transition_matrix++    T_last = T[n_prev:, :]++    rows = weighted_rows(g_last ** dt_out if GROWTH else np.ones(n_last), n, rng)++    T_sel = np.asarray(T_last[rows].todense(), dtype=np.float32)+    T_sel_sum = T_sel.sum(axis=1, keepdims=True)+    T_sel_sum[T_sel_sum == 0] = 1.0+    T_sel = T_sel / T_sel_sum++    last_to_all = T_sel[:, n_prev:]+    for i in range(last_to_all.shape[0]):+        row = last_to_all[i]+        if np.count_nonzero(row) > TOP_K_SUCC:+            thresh = np.partition(row, -TOP_K_SUCC)[-TOP_K_SUCC]+            row[row < thresh] = 0.0+            s = row.sum()+            if s > 0:+                last_to_all[i] = row / s++    Xl_dense = torch.from_numpy(last.X.toarray())+    succ_mean = (torch.from_numpy(last_to_all.astype(np.float32)) @ Xl_dense).numpy()++    X = last.X[rows].toarray()+    step = (succ_mean - X) * (LAMBDA * dt_out / dt_in)+    step[:, ~mask] = 0.0+    step *= X > 0+    X += step+    np.maximum(X, 0.0, out=X)+    write_prediction(X, genes, args.out, seed=args.seed)+++if __name__ == "__main__":+    main()

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

Looking at the scores: direction is excellent (56.9) but cell_state (3.5) and covariation (3.26) collapsed. The problem is that `succ_mean` is a weighted average over ALL latest-stage cells, which is extremely smooth and destroys individual cell identity when used as a full step.

Fix: (1) sparsify the transition to keep only top-k successors per cell (more local), (2) reduce LAMBDA significantly to preserve cell identity while keeping the directional signal.

<<<<<<< SEARCH
LAMBDA = 1.0
=======
LAMBDA = 0.15
TOP_K_SUCC = 30
>>>>>>> REPLACE

<<<<<<< SEARCH
    T_sel = np.asarray(T_last[rows].todense(), dtype=np.float32)
    T_sel_sum = T_sel.sum(axis=1, keepdims=True)
    T_sel_sum[T_sel_sum == 0] = 1.0
    T_sel = T_sel / T_sel_sum

    last_to_all = T_sel[:, n_prev:]

    Xl_dense = torch.from_numpy(last.X.toarray())
    succ_mean = (torch.from_numpy(last_to_all.astype(np.float32)) @ Xl_dense).numpy()
=======
    T_sel = np.asarray(T_last[rows].todense(), dtype=np.float32)
    T_sel_sum = T_sel.sum(axis=1, keepdims=True)
    T_sel_sum[T_sel_sum == 0] = 1.0
    T_sel = T_sel / T_sel_sum

    last_to_all = T_sel[:, n_prev:]
    for i in range(last_to_all.shape[0]):
        row = last_to_all[i]
        if np.count_nonzero(row) > TOP_K_SUCC:
            thresh = np.partition(row, -TOP_K_SUCC)[-TOP_K_SUCC]
            row[row < thresh] = 0.0
            s = row.sum()
            if s > 0:
                last_to_all[i] = row / s

    Xl_dense = torch.from_numpy(last.X.toarray())
    succ_mean = (torch.from_numpy(last_to_all.astype(np.float32)) @ Xl_dense).numpy()
>>>>>>> REPLACE