Skip to content

Commit 13d80a3

Browse files
rvinasrvinas
committed
Add configs
1 parent 1663270 commit 13d80a3

11 files changed

+764
-0
lines changed

‎configs/default.yaml

Lines changed: 68 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,68 @@
1+
gpu:
2+
value: 0
3+
sweep:
4+
value: False
5+
lr:
6+
desc: Learning rate for optimiser
7+
value: 0.00045568167228053424
8+
batch_size:
9+
value: 63
10+
epochs:
11+
desc: Maximum number of epochs to run
12+
value: 200
13+
patience:
14+
desc: Not yet supported
15+
value: 30
16+
dropout:
17+
desc: Probability of dropping out a hidden unit
18+
value: 0.17385777664292695
19+
bn:
20+
desc: Whether to use batch normalisation (TODO Not yet supported)
21+
value: True
22+
use_demographic_information:
23+
value: True
24+
update_edge_attr:
25+
desc: Whether to update the edge attributes in the hypergraph layer
26+
value: False
27+
n_graph_layers:
28+
desc: Number of hidden layers
29+
value: 2
30+
n_hidden_layers:
31+
desc: Number of hidden layers
32+
value: 1
33+
n_hidden_layers_pred:
34+
desc: Number of hidden layers
35+
value: 2
36+
n_hidden_layers_prior_var:
37+
desc: Number of hidden layers for computing prior var (identifiable VAE)
38+
value: 0
39+
meta_G:
40+
value: 50
41+
d_patient:
42+
value: 71
43+
d_gene:
44+
value: 48
45+
d_edge_attr:
46+
value: 98
47+
d_tissue:
48+
value: 120
49+
d_edge:
50+
value: 28
51+
n_heads:
52+
value: 28
53+
beta:
54+
value: 0
55+
n_top_genes:
56+
value: null
57+
attention_strategy:
58+
value: 'patient'
59+
metagenes_encoder:
60+
value: 'plain'
61+
loss_type:
62+
value: 'normal'
63+
layer:
64+
value: 'gat'
65+
activation:
66+
value: 'swish'
67+
norm:
68+
value: 'batch'

‎configs/default_2.yaml

Lines changed: 62 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,62 @@
1+
gpu:
2+
value: 3
3+
sweep:
4+
value: False
5+
lr:
6+
desc: Learning rate for optimiser
7+
value: 0.001815
8+
batch_size:
9+
value: 18
10+
epochs:
11+
desc: Maximum number of epochs to run
12+
value: 400
13+
patience:
14+
desc: Not yet supported
15+
value: 30
16+
dropout:
17+
desc: Probability of dropping out a hidden unit
18+
value: 0.188
19+
bn:
20+
desc: Whether to use batch normalisation (TODO Not yet supported)
21+
value: True
22+
update_edge_attr:
23+
desc: Whether to update the edge attributes in the hypergraph layer
24+
value: False
25+
use_demographic_information:
26+
value: False
27+
n_graph_layers:
28+
desc: Number of hidden layers
29+
value: 2
30+
n_hidden_layers:
31+
desc: Number of hidden layers
32+
value: 2
33+
n_hidden_layers_pred:
34+
desc: Number of hidden layers
35+
value: 1
36+
n_hidden_layers_prior_var:
37+
desc: Number of hidden layers for computing prior var (identifiable VAE)
38+
value: 1
39+
meta_G:
40+
value: 67
41+
d_patient:
42+
value: 21
43+
d_gene:
44+
value: 117
45+
d_edge_attr:
46+
value: 22
47+
d_tissue:
48+
value: 23
49+
d_edge:
50+
value: 21
51+
n_heads:
52+
value: 12
53+
beta:
54+
value: 1
55+
n_top_genes:
56+
value: null
57+
attention_strategy:
58+
value: 'patient'
59+
metagenes_encoder:
60+
value: 'plain'
61+
loss_type:
62+
value: 'normal'

‎configs/default_celltypesig.yaml

Lines changed: 70 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,70 @@
1+
gpu:
2+
value: 0
3+
sweep:
4+
value: False
5+
lr:
6+
desc: Learning rate for optimiser
7+
value: 0.00045568167228053424
8+
batch_size:
9+
value: 63
10+
epochs:
11+
desc: Maximum number of epochs to run
12+
value: 200
13+
patience:
14+
desc: Not yet supported
15+
value: 30
16+
dropout:
17+
desc: Probability of dropping out a hidden unit
18+
value: 0.17385777664292695
19+
bn:
20+
desc: Whether to use batch normalisation (TODO Not yet supported)
21+
value: True
22+
use_demographic_information:
23+
value: True
24+
update_edge_attr:
25+
desc: Whether to update the edge attributes in the hypergraph layer
26+
value: False
27+
n_graph_layers:
28+
desc: Number of hidden layers
29+
value: 2
30+
n_hidden_layers:
31+
desc: Number of hidden layers
32+
value: 1
33+
n_hidden_layers_pred:
34+
desc: Number of hidden layers
35+
value: 2
36+
n_hidden_layers_prior_var:
37+
desc: Number of hidden layers for computing prior var (identifiable VAE)
38+
value: 0
39+
meta_G:
40+
value: 50
41+
d_patient:
42+
value: 71
43+
d_gene:
44+
value: 48
45+
d_edge_attr:
46+
value: 98
47+
d_tissue:
48+
value: 120
49+
d_edge:
50+
value: 28
51+
n_heads:
52+
value: 28
53+
beta:
54+
value: 0
55+
n_top_genes:
56+
value: null
57+
attention_strategy:
58+
value: 'patient'
59+
metagenes_encoder:
60+
value: 'plain'
61+
loss_type:
62+
value: 'normal'
63+
layer:
64+
value: 'gat'
65+
activation:
66+
value: 'swish'
67+
norm:
68+
value: 'batch'
69+
gene_selection_strategy:
70+
value: 'highly variable v9' # 'highly variable v9' # 'differentially expressed'

‎configs/sweep_ablation/sweep_gat_m1.yaml

Lines changed: 72 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,72 @@
1+
program: train_gtex.py
2+
method: bayes
3+
project: multitissue_imputation
4+
entity: multitissue_imputation_project
5+
metric:
6+
goal: minimize
7+
name: val_loss
8+
parameters:
9+
gpu:
10+
value: 0
11+
sweep:
12+
value: True
13+
epochs:
14+
value: 100
15+
patience:
16+
value: 30
17+
lr:
18+
min: 0.0001
19+
max: 0.005
20+
batch_size:
21+
min: 16
22+
max: 64
23+
dropout:
24+
min: 0.0
25+
max: 0.5
26+
norm:
27+
values: ['batch', 'layer', 'none']
28+
activation:
29+
values: ['relu', 'swish']
30+
update_edge_attr:
31+
values: [True, False]
32+
use_demographic_information:
33+
value: True
34+
n_graph_layers:
35+
min: 1
36+
max: 3
37+
n_hidden_layers:
38+
min: 1
39+
max: 2
40+
n_hidden_layers_pred:
41+
min: 1
42+
max: 2
43+
n_hidden_layers_prior_var:
44+
value: 0
45+
d_patient:
46+
min: 16
47+
max: 128
48+
d_gene:
49+
min: 16
50+
max: 128
51+
d_edge_attr:
52+
min: 16
53+
max: 128
54+
d_tissue:
55+
min: 16
56+
max: 128
57+
d_edge:
58+
min: 16
59+
max: 128
60+
n_heads:
61+
min: 4
62+
max: 32
63+
beta:
64+
value: 0.
65+
attention_strategy:
66+
value: 'patient'
67+
loss_type:
68+
value: 'normal'
69+
layer:
70+
value: 'gat'
71+
meta_G:
72+
value: 1

‎configs/sweep_ablation/sweep_gat_m10.yaml

Lines changed: 72 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,72 @@
1+
program: train_gtex.py
2+
method: bayes
3+
project: multitissue_imputation
4+
entity: multitissue_imputation_project
5+
metric:
6+
goal: minimize
7+
name: val_loss
8+
parameters:
9+
gpu:
10+
value: 0
11+
sweep:
12+
value: True
13+
epochs:
14+
value: 100
15+
patience:
16+
value: 30
17+
lr:
18+
min: 0.0001
19+
max: 0.005
20+
batch_size:
21+
min: 16
22+
max: 64
23+
dropout:
24+
min: 0.0
25+
max: 0.5
26+
norm:
27+
values: ['batch', 'layer', 'none']
28+
activation:
29+
values: ['relu', 'swish']
30+
update_edge_attr:
31+
values: [True, False]
32+
use_demographic_information:
33+
value: True
34+
n_graph_layers:
35+
min: 1
36+
max: 3
37+
n_hidden_layers:
38+
min: 1
39+
max: 2
40+
n_hidden_layers_pred:
41+
min: 1
42+
max: 2
43+
n_hidden_layers_prior_var:
44+
value: 0
45+
d_patient:
46+
min: 16
47+
max: 128
48+
d_gene:
49+
min: 16
50+
max: 128
51+
d_edge_attr:
52+
min: 16
53+
max: 128
54+
d_tissue:
55+
min: 16
56+
max: 128
57+
d_edge:
58+
min: 16
59+
max: 128
60+
n_heads:
61+
min: 4
62+
max: 32
63+
beta:
64+
value: 0.
65+
attention_strategy:
66+
value: 'patient'
67+
loss_type:
68+
value: 'normal'
69+
layer:
70+
value: 'gat'
71+
meta_G:
72+
value: 10

0 commit comments

Comments
 (0)