Hugging Face's logo

Datasets:
GAO612
/
SAMI

Languages:
English
Tags:
spatial-omics
multimodal
histology
pytorch
Dataset card Files
xet
 Community
SAMI
File size: 2,084 Bytes
7bd444b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
seed: 42

data:
  train_dir: "./sami/Simulation2/processed_simulation2.h5ad"
  val_dir: ""
  infer_dir: "./data/infer"
  file_pattern: "*.h5ad"
  omics1_obsm_key: "rna_feat_pca"
  omics2_obsm_key: "protein_feat"
  omics1_target_obsm_key: "rna_feat_pca"
  omics2_target_obsm_key: "protein_feat"
  batch_size: 1
  num_workers: 0
  pin_memory: false

model:
  branch_mode: "multimodal_fusion" # Options: "omics1", "omics2", "multimodal_fusion"
  hidden_dim: 64
  omics1_in_dim: 50
  omics2_in_dim: 100
  omics1_target_dim: 50
  omics2_target_dim: 100

  omics1_encoder:
    feature_adapter:
      adapter_type: "mlp_bottleneck"
      adapter_dim: 64
      dropout: 0.1
      force_linear: false

  omics1_augmentation:
    enabled: false
    alpha: 0.5

  omics2_encoder:
    adapter_type: "mlp_bottleneck"
    adapter_dim: 64
    dropout: 0.1
    force_linear: false

  graph:
    top_k: 6
    symmetrize: true
    add_self_loops: false
    chunk_size: 512

  graph_backbone:
    omics1:
      type: "pointnext"
      use_layer_transformer: true
    omics2:
      type: "pointnext"
      use_layer_transformer: true
    pointnext:
      expansion: 4
      dropout: 0.1
      max_neighbors: null
      pooling: "mean"

  cross_modal_attention:
    num_heads: 4
    dropout: 0.1

  gat:
    num_layers: 3
    omics1_num_layers: 3
    omics2_num_layers: 3
    num_heads: 4
    dropout: 0.1
    leaky_relu_negative_slope: 0.2

  layer_transformer:
    num_heads: 4
    ff_dim: 256
    dropout: 0.1
    pooling: "mean"

  fusion:
    mlp_hidden_dim: 256
    dropout: 0.1

train:
  device: "cuda"
  epochs: 100
  lr: 1.0e-3
  weight_decay: 1.0e-4
  grad_clip_norm: 5.0
  amp: false
  log_interval: 10
  output_dir: "./outputs"
  save_every: 20

loss:
  lambda_omics1: 10.0
  lambda_omics2: 10.0
  lambda_contrastive: 0.5
  lambda_consistency: 1.0

  contrastive:
    use_symmetric_loss: false

  consistency:
    temperature: 0.1
    projector_hidden_dim: 128
    projector_out_dim: 64

infer:
  checkpoint_path: ""
  output_path: "./outputs/infer_embeddings.pt"
  save_reconstruction: false