不使用参考 scRNA-seq 的空间反卷积

不使用参考 scRNA-seq 的空间反卷积#

这是一个关于在没有参考的情况下进行空间反卷积的教程。

import omicverse as ov

🔬 Starting plot initialization...
Using already downloaded Arial font from: /tmp/omicverse_arial.ttf
Registered as: Arial
🧬 Detecting GPU devices…
✅ NVIDIA CUDA GPUs detected: 1
    • [CUDA 0] NVIDIA H100 80GB HBM3
      Memory: 79.1 GB | Compute: 9.0

   ____            _     _    __                  
  / __ \____ ___  (_)___| |  / /__  _____________ 
 / / / / __ `__ \/ / ___/ | / / _ \/ ___/ ___/ _ \ 
/ /_/ / / / / / / / /__ | |/ /  __/ /  (__  )  __/ 
\____/_/ /_/ /_/_/\___/ |___/\___/_/  /____/\___/                                              

🔖 Version: 1.7.9rc1   📚 Tutorials: https://omicverse.readthedocs.io/
✅ plot_set complete.

第1步：准备空间转录组学 (1 分钟)#

目的：加载 10x Visium (Space Ranger 输出) 或…

# 加载 scRNA-seq 参考
# 示例：人类淋巴结参考
adata_sc = ov.datasets.sc_ref_Lymph()

reading /scratch/users/steorra/analysis/omic_test/data/V1_Human_Lymph_Node/filtered_feature_bc_matrix.h5
 (0:00:00)

步骤 2: Prepare 该 基因 sig markergene_sig means 该 数据框 stored 该 标记 基因 在 each columns. 如果 you don’t have it, you can calculated it using ov.space.calculate_gene_signature#

# 加载 your scRNA-seq 参考# 示例: Human lymph node referenceadata_sc = ov.数据集.sc_ref_Lymph_Node()gene_sig=ov.space.calculate_gene_signature(    adata_sc,    clustertype='Subset',      # 细胞类型 name    rank=True,    key='rank_genes_groups',    foldchange=2,    topgenenumber=20    )gene_sig.head()

🧬 Loading SC reference data for Lymph Node
🔍 Downloading data to ./data/sc_ref_Lymph_Node.h5ad
⚠️ File ./data/sc_ref_Lymph_Node.h5ad already exists
 Loading data from ./data/sc_ref_Lymph_Node.h5ad
✅ Successfully loaded: 73260 cells × 10237 genes
...get cell type marker
ranking genes
WARNING: It seems you use rank_genes_groups on the raw count data. Please logarithmize your data before calling rank_genes_groups.
    finished: added to `.uns['rank_genes_groups']`
    'names', sorted np.recarray to be indexed by group ids
    'scores', sorted np.recarray to be indexed by group ids
    'logfoldchanges', sorted np.recarray to be indexed by group ids
    'pvals', sorted np.recarray to be indexed by group ids
    'pvals_adj', sorted np.recarray to be indexed by group ids (0:01:00)
B_Cycling B_GC_DZ B_GC_LZ B_GC_prePB B_IFN B_activated B_mem B_naive B_plasma B_preGC ... T_CD4+_TfH T_CD4+_TfH_GC T_CD4+_naive T_CD8+_CD161+ T_CD8+_cytotoxic T_CD8+_naive T_TIM3+ T_TfR T_Treg VSMC
0 HMGN2 CD79B CD22 BIK HLA-DPA1 HLA-DPA1 HLA-DPA1 HLA-DPA1 SDF2L1 HLA-DPA1 ... TRAC TCF7 RPS15A ZFP36 CD3E NOSIP TMSB4X CD4 TRAC TAGLN
1 DEK TCEA1 PRPSAP2 PRPSAP2 ISG15 HVCN1 CD52 HVCN1 SSR3 CD72 ... CD69 SRGN NOSIP SRGN CST7 TCF7 LCK SRGN SRGN LGALS3
2 HMGN1 GAPDH HMGN1 VPREB3 STAT1 CD74 SMIM14 CD79B PPIB NME2 ... SRGN PASK TCF7 RGCC GZMK RGCC SRGN CD3E RGCC CST3
3 GAPDH SUGCT GAPDH HMCES CD74 TCL1A CD74 CD72 SEC61G CD74 ... CD3E ITM2A RPS14 CD8A NKG7 DNAJB1 CD3E ZAP70 DNAJB1 LAPTM4A
4 HMGB1 EZR CD74 MEF2B IFIT3 HLA-DPB1 VPREB3 CD74 HERPUD1 HSP90AB1 ... SLC2A3 CD3E LEF1 CCL5 CTSW LEF1 CD27 TCRA_VDJsum FOXP3 IGFBP5

5 rows × 34 columns

gene_sig.to_csv('data/gene_sig_Lymph_Node.csv')

gene_sig=ov.pd.read_csv('data/gene_sig_Lymph_Node.csv',index_col=0)gene_sig.head()
B_Cycling B_GC_DZ B_GC_LZ B_GC_prePB B_IFN B_activated B_mem B_naive B_plasma B_preGC ... T_CD4+_TfH T_CD4+_TfH_GC T_CD4+_naive T_CD8+_CD161+ T_CD8+_cytotoxic T_CD8+_naive T_TIM3+ T_TfR T_Treg VSMC
0 HMGN2 CD79B CD22 BIK HLA-DPA1 HLA-DPA1 HLA-DPA1 HLA-DPA1 SDF2L1 HLA-DPA1 ... TRAC TCF7 RPS15A ZFP36 CD3E NOSIP TMSB4X CD4 TRAC TAGLN
1 DEK TCEA1 PRPSAP2 PRPSAP2 ISG15 HVCN1 CD52 HVCN1 SSR3 CD72 ... CD69 SRGN NOSIP SRGN CST7 TCF7 LCK SRGN SRGN LGALS3
2 HMGN1 GAPDH HMGN1 VPREB3 STAT1 CD74 SMIM14 CD79B PPIB NME2 ... SRGN PASK TCF7 RGCC GZMK RGCC SRGN CD3E RGCC CST3
3 GAPDH SUGCT GAPDH HMCES CD74 TCL1A CD74 CD72 SEC61G CD74 ... CD3E ITM2A RPS14 CD8A NKG7 DNAJB1 CD3E ZAP70 DNAJB1 LAPTM4A
4 HMGB1 EZR CD74 MEF2B IFIT3 HLA-DPB1 VPREB3 CD74 HERPUD1 HSP90AB1 ... SLC2A3 CD3E LEF1 CCL5 CTSW LEF1 CD27 TCRA_VDJsum FOXP3 IGFBP5

5 rows × 34 columns

步骤 3: 解卷积 使用 starfyshWe perform n_repeat random restarts 和 选择 该 best model 使用 lowest loss 对于 参数 c (inferred 细胞类型 proportions):STARFYSH is integrated into 该 omicverse.space.Deconvolution 类. Simply set method='starfysh'.### Key 参数- n_repeats: Number 的 restart 到 run STARFYSH.- epochs: number 的 iterations- device: 该 trainning device#

# Initialize 该 解卷积 objectdecov_obj = ov.space.解卷积(    adata_sp=adata_sp)

decov_obj.preprocess_sp(    mode='pearsonr',n_svgs=3000,target_sum=1e4,    subset_genes=False,)

🔍 [2026-01-11 23:07:13] Running preprocessing in 'cpu' mode...
Begin robust gene identification
    After filtration, 25187/36601 genes are kept.
    Among 25187 genes, 22411 genes are robust.
✅ Robust gene identification completed successfully.
Begin size normalization: shiftlog and HVGs selection pearson

🔍 Count Normalization:
   Target sum: 10000.0
   Exclude highly expressed: True
   Max fraction threshold: 0.2
   ⚠️ Excluding 1 highly-expressed genes from normalization computation
   Excluded genes: ['IGKC']

✅ Count Normalization Completed Successfully!
   ✓ Processed: 4,035 cells × 22,411 genes
   ✓ Runtime: 0.47s

🔍 Highly Variable Genes Selection (Experimental):
   Method: pearson_residuals
   Target genes: 3,000
   Theta (overdispersion): 100

✅ Experimental HVG Selection Completed Successfully!
   ✓ Selected: 3,000 highly variable genes out of 22,411 total (13.4%)
   ✓ Results added to AnnData object:
     • 'highly_variable': Boolean vector (adata.var)
     • 'highly_variable_rank': Float vector (adata.var)
     • 'highly_variable_nbatches': Int vector (adata.var)
     • 'highly_variable_intersection': Boolean vector (adata.var)
     • 'means': Float vector (adata.var)
     • 'variances': Float vector (adata.var)
     • 'residual_variances': Float vector (adata.var)
    Time to analyze data in cpu: 3.05 seconds.
computing PCA
    with n_comps=50
    finished (0:00:01)
✅ Preprocessing completed successfully.
    Added:
        'highly_variable_features', boolean vector (adata.var)
        'means', float vector (adata.var)
        'variances', float vector (adata.var)
        'residual_variances', float vector (adata.var)
        'counts', raw counts layer (adata.layers)
    End of size normalization: shiftlog and HVGs selection pearson
✓ spatial transcriptomics data is preprocessed

%time# Run FlashDeconv deconvolutiondecov_obj.解卷积(    方法='STARFYSH',    gene_sig=gene_sig,      starfysh_kwargs={        'n_repeats':3,        'epochs':200,        'patience':50,        'device':None,        'batch_size':32,        'alpha_mul':50,        'lr':1e-4,        'poe':False,        'verbose':True,        'n_anchors':60,        'window_size':3,    })

CPU times: user 3 μs, sys: 0 ns, total: 3 μs
Wall time: 7.87 μs
30 components are retained using conditional_number=40.00
Epoch[10/200], train_loss: 1964.5555, train_reconst: 1717.3358, train_u: 17.7731,train_z: 12.2491,train_c: 216.1686,train_l: 1.0289
Epoch[20/200], train_loss: 1836.5867, train_reconst: 1631.5397, train_u: 17.3562,train_z: 8.2599,train_c: 177.4633,train_l: 1.9675
Epoch[30/200], train_loss: 1803.6228, train_reconst: 1611.8570, train_u: 17.0200,train_z: 6.8935,train_c: 165.4288,train_l: 2.4234
Epoch[40/200], train_loss: 1784.1136, train_reconst: 1599.9807, train_u: 16.7433,train_z: 6.1974,train_c: 158.4979,train_l: 2.6943
Epoch[50/200], train_loss: 1771.6689, train_reconst: 1591.6377, train_u: 16.5159,train_z: 5.8556,train_c: 154.8143,train_l: 2.8453
Epoch[60/200], train_loss: 1761.9189, train_reconst: 1585.5430, train_u: 16.3333,train_z: 5.7242,train_c: 151.3896,train_l: 2.9288
Epoch[70/200], train_loss: 1755.7707, train_reconst: 1580.6499, train_u: 16.1848,train_z: 5.6212,train_c: 150.3357,train_l: 2.9791
Epoch[80/200], train_loss: 1748.3032, train_reconst: 1575.9084, train_u: 16.0641,train_z: 5.6020,train_c: 147.7620,train_l: 2.9667
Epoch[90/200], train_loss: 1745.4689, train_reconst: 1573.8276, train_u: 15.9672,train_z: 5.5890,train_c: 147.1002,train_l: 2.9849
Epoch[100/200], train_loss: 1741.4059, train_reconst: 1571.2467, train_u: 15.8885,train_z: 5.5138,train_c: 145.7818,train_l: 2.9751
Epoch[110/200], train_loss: 1739.2005, train_reconst: 1569.5755, train_u: 15.8249,train_z: 5.5090,train_c: 145.3119,train_l: 2.9792
Epoch[120/200], train_loss: 1737.3613, train_reconst: 1568.0982, train_u: 15.7730,train_z: 5.5225,train_c: 144.9962,train_l: 2.9714
Epoch[130/200], train_loss: 1735.5456, train_reconst: 1567.2170, train_u: 15.7308,train_z: 5.5234,train_c: 144.1060,train_l: 2.9684
Epoch[140/200], train_loss: 1733.8153, train_reconst: 1565.6691, train_u: 15.6964,train_z: 5.5186,train_c: 143.9690,train_l: 2.9622
Epoch[150/200], train_loss: 1734.0409, train_reconst: 1565.6646, train_u: 15.6686,train_z: 5.5101,train_c: 144.2291,train_l: 2.9684
Epoch[160/200], train_loss: 1733.5203, train_reconst: 1564.8707, train_u: 15.6460,train_z: 5.5183,train_c: 144.5110,train_l: 2.9744
Epoch[170/200], train_loss: 1731.6396, train_reconst: 1563.7964, train_u: 15.6274,train_z: 5.5189,train_c: 143.7256,train_l: 2.9713
Epoch[180/200], train_loss: 1730.6814, train_reconst: 1563.4632, train_u: 15.6123,train_z: 5.4934,train_c: 143.1347,train_l: 2.9779
Epoch[190/200], train_loss: 1730.8846, train_reconst: 1563.2024, train_u: 15.6000,train_z: 5.5424,train_c: 143.5656,train_l: 2.9743
Epoch[200/200], train_loss: 1730.3862, train_reconst: 1563.3851, train_u: 15.5899,train_z: 5.4530,train_c: 142.9817,train_l: 2.9766
Epoch[10/200], train_loss: 1964.5555, train_reconst: 1717.3358, train_u: 17.7731,train_z: 12.2491,train_c: 216.1686,train_l: 1.0289
Epoch[20/200], train_loss: 1836.5867, train_reconst: 1631.5397, train_u: 17.3562,train_z: 8.2599,train_c: 177.4633,train_l: 1.9675
Epoch[30/200], train_loss: 1803.6228, train_reconst: 1611.8570, train_u: 17.0200,train_z: 6.8935,train_c: 165.4288,train_l: 2.4234
Epoch[40/200], train_loss: 1784.1136, train_reconst: 1599.9807, train_u: 16.7433,train_z: 6.1974,train_c: 158.4979,train_l: 2.6943
Epoch[50/200], train_loss: 1771.6689, train_reconst: 1591.6377, train_u: 16.5159,train_z: 5.8556,train_c: 154.8143,train_l: 2.8453
Epoch[60/200], train_loss: 1761.9189, train_reconst: 1585.5430, train_u: 16.3333,train_z: 5.7242,train_c: 151.3896,train_l: 2.9288
Epoch[70/200], train_loss: 1755.7707, train_reconst: 1580.6499, train_u: 16.1848,train_z: 5.6212,train_c: 150.3357,train_l: 2.9791
Epoch[80/200], train_loss: 1748.3032, train_reconst: 1575.9084, train_u: 16.0641,train_z: 5.6020,train_c: 147.7620,train_l: 2.9667
Epoch[90/200], train_loss: 1745.4689, train_reconst: 1573.8276, train_u: 15.9672,train_z: 5.5890,train_c: 147.1002,train_l: 2.9849
Epoch[100/200], train_loss: 1741.4059, train_reconst: 1571.2467, train_u: 15.8885,train_z: 5.5138,train_c: 145.7818,train_l: 2.9751
Epoch[110/200], train_loss: 1739.2005, train_reconst: 1569.5755, train_u: 15.8249,train_z: 5.5090,train_c: 145.3119,train_l: 2.9792
Epoch[120/200], train_loss: 1737.3613, train_reconst: 1568.0982, train_u: 15.7730,train_z: 5.5225,train_c: 144.9962,train_l: 2.9714
Epoch[130/200], train_loss: 1735.5456, train_reconst: 1567.2170, train_u: 15.7308,train_z: 5.5234,train_c: 144.1060,train_l: 2.9684
Epoch[140/200], train_loss: 1733.8153, train_reconst: 1565.6691, train_u: 15.6964,train_z: 5.5186,train_c: 143.9690,train_l: 2.9622
Epoch[150/200], train_loss: 1734.0409, train_reconst: 1565.6646, train_u: 15.6686,train_z: 5.5101,train_c: 144.2291,train_l: 2.9684
Epoch[160/200], train_loss: 1733.5203, train_reconst: 1564.8707, train_u: 15.6460,train_z: 5.5183,train_c: 144.5110,train_l: 2.9744
Epoch[170/200], train_loss: 1731.6396, train_reconst: 1563.7964, train_u: 15.6274,train_z: 5.5189,train_c: 143.7256,train_l: 2.9713
Epoch[180/200], train_loss: 1730.6814, train_reconst: 1563.4632, train_u: 15.6123,train_z: 5.4934,train_c: 143.1347,train_l: 2.9779
Epoch[190/200], train_loss: 1730.8846, train_reconst: 1563.2024, train_u: 15.6000,train_z: 5.5424,train_c: 143.5656,train_l: 2.9743
Epoch[200/200], train_loss: 1730.3862, train_reconst: 1563.3851, train_u: 15.5899,train_z: 5.4530,train_c: 142.9817,train_l: 2.9766
Epoch[10/200], train_loss: 1964.5555, train_reconst: 1717.3358, train_u: 17.7731,train_z: 12.2491,train_c: 216.1686,train_l: 1.0289
Epoch[20/200], train_loss: 1836.5867, train_reconst: 1631.5397, train_u: 17.3562,train_z: 8.2599,train_c: 177.4633,train_l: 1.9675
Epoch[30/200], train_loss: 1803.6228, train_reconst: 1611.8570, train_u: 17.0200,train_z: 6.8935,train_c: 165.4288,train_l: 2.4234
Epoch[40/200], train_loss: 1784.1136, train_reconst: 1599.9807, train_u: 16.7433,train_z: 6.1974,train_c: 158.4979,train_l: 2.6943
Epoch[50/200], train_loss: 1771.6689, train_reconst: 1591.6377, train_u: 16.5159,train_z: 5.8556,train_c: 154.8143,train_l: 2.8453
Epoch[60/200], train_loss: 1761.9189, train_reconst: 1585.5430, train_u: 16.3333,train_z: 5.7242,train_c: 151.3896,train_l: 2.9288
Epoch[70/200], train_loss: 1755.7707, train_reconst: 1580.6499, train_u: 16.1848,train_z: 5.6212,train_c: 150.3357,train_l: 2.9791
Epoch[80/200], train_loss: 1748.3032, train_reconst: 1575.9084, train_u: 16.0641,train_z: 5.6020,train_c: 147.7620,train_l: 2.9667
Epoch[90/200], train_loss: 1745.4689, train_reconst: 1573.8276, train_u: 15.9672,train_z: 5.5890,train_c: 147.1002,train_l: 2.9849
Epoch[100/200], train_loss: 1741.4059, train_reconst: 1571.2467, train_u: 15.8885,train_z: 5.5138,train_c: 145.7818,train_l: 2.9751
Epoch[110/200], train_loss: 1739.2005, train_reconst: 1569.5755, train_u: 15.8249,train_z: 5.5090,train_c: 145.3119,train_l: 2.9792
Epoch[120/200], train_loss: 1737.3613, train_reconst: 1568.0982, train_u: 15.7730,train_z: 5.5225,train_c: 144.9962,train_l: 2.9714
Epoch[130/200], train_loss: 1735.5456, train_reconst: 1567.2170, train_u: 15.7308,train_z: 5.5234,train_c: 144.1060,train_l: 2.9684
Epoch[140/200], train_loss: 1733.8153, train_reconst: 1565.6691, train_u: 15.6964,train_z: 5.5186,train_c: 143.9690,train_l: 2.9622
Epoch[150/200], train_loss: 1734.0409, train_reconst: 1565.6646, train_u: 15.6686,train_z: 5.5101,train_c: 144.2291,train_l: 2.9684
Epoch[160/200], train_loss: 1733.5203, train_reconst: 1564.8707, train_u: 15.6460,train_z: 5.5183,train_c: 144.5110,train_l: 2.9744
Epoch[170/200], train_loss: 1731.6396, train_reconst: 1563.7964, train_u: 15.6274,train_z: 5.5189,train_c: 143.7256,train_l: 2.9713
Epoch[180/200], train_loss: 1730.6814, train_reconst: 1563.4632, train_u: 15.6123,train_z: 5.4934,train_c: 143.1347,train_l: 2.9779
Epoch[190/200], train_loss: 1730.8846, train_reconst: 1563.2024, train_u: 15.6000,train_z: 5.5424,train_c: 143.5656,train_l: 2.9743
Epoch[200/200], train_loss: 1730.3862, train_reconst: 1563.3851, train_u: 15.5899,train_z: 5.4530,train_c: 142.9817,train_l: 2.9766
✓ starfysh is done
The starfysh result is saved in self.adata_cell2location
The starfysh model is saved in self.starfysh_model

步骤 4: 可视化 (5–15 min)We provide multiple views: single-target spatial heatmaps, multi-target overlays, 和 local pie charts. Start 使用 global inspection, then zoom into biologically relevant regions 对于 higher-resolution assessment.### 4.1 Spatial value dotplot#

annotation_list=['B_Cycling', 'B_GC_LZ', 'T_CD4+_TfH_GC', 'FDC',                'B_naive', 'T_CD4+_naive', 'B_plasma', 'Endo']sc.pl.spatial(    decov_obj.adata_cell2location,     cmap='magma',    # 显示 first 8 细胞类型    颜色=annotation_list,    ncols=4, 大小=1.3,    img_key='hires',    # limit 颜色 缩放 在 99.2% quantile 的 cell abundance    #vmin=0, vmax='p99.2')
../_images/b1d5fc0d7bb871568e83671c4ce2946249678548500abe6bea469019c2245140.png

4.2 Multi-target overlay#

import matplotlib as mplclust_labels = annotation_list[:5]clust_col = ['' + str(i) for i in clust_labels] # 在 case column names differ 从 labelswith mpl.rc_context({'图.figsize': (6, 6),'轴.grid': False}):    fig = ov.pl.plot_spatial(        adata=decov_obj.adata_cell2location,        # 标签 到 显示 在 a 绘制        颜色=clust_col, 标签=clust_labels,        show_img=True,        # 'fast' (white background) 或 'dark_background'        style='fast',        # limit 颜色 缩放 在 99.2% quantile 的 cell abundance        max_color_quantile=0.992,        # 大小 的 locations (adjust depending 在 图 大小)        circle_diameter=4,        reorder_cmap = [#0,            1,2,3,4,6], #['yellow', 'orange', 'blue', 'green', 'purple', 'grey', 'white'],        colorbar_position='right',        #palette=color_dict    )
../_images/59736f9724635714c6321c7562d553dda3b0bd1569db1cc536a3e9388dac94c8.png

4.3 Pie plotWe recommend cropping a region 的 interest before plotting 到 avoid overly 密集 pie charts 在 whole slides.#

adata_s = ov.space.crop_space_visium(    decov_obj.adata_cell2location,     crop_loc=(0, 0),          crop_area=(500, 1000),     library_id=list(decov_obj.adata_cell2location.uns['spatial'].keys())[0] ,     scale=1)

sc.pl.spatial(adata_s, cmap='magma',                  # 显示 first 8 细胞类型                  颜色=annotation_list[0],                  ncols=4, 大小=1.3,                  img_key='hires',                  # limit 颜色 缩放 在 99.2% quantile 的 cell abundance                  #vmin=0, vmax='p99.2'                 )
../_images/5c44e6fbe5fc2eb2c2820a7fd197366209273f1a3c3e4d0d228e69a832e3598d.png 
color_dict=dict(zip(annotation_list,                   ov.pl.sc_color))

fig, ax = ov.plt.subplots(figsize=(8, 4))sc.pl.spatial(    adata_s,     basis='spatial',    color=None,      size=1.3,    img_key='hires',    ax=ax,          show=False)ov.pl.add_pie2spatial(    adata_s,    img_key='hires',    cell_type_columns=annotation_list[:],    ax=ax,    colors=color_dict,    pie_radius=10,    remainder='gap',    legend_loc=(0.5, -0.25),    ncols=4,    alpha=0.8)# plt.显示()

<Axes: xlabel='spatial1', ylabel='spatial2'>
../_images/13ade77e2ff9339d5135e9fa7071c672249acbf0a3b3c3beeb70a5d8bf1be0ac.png
目录