set reasonable defaults

Author: AnneHartebrodt

src/netmap/grn/inferrence.py

@@ -213,7 +213,7 @@ def attribution_one_target(
     return attributions_list
 
 
-def inferrence(models, data_train_full_tensor, gene_names, xai_method):
+def inferrence(models, data_train_full_tensor, gene_names, xai_method='GuidedBackprop'):
 
     """
     The main inferrence function to compute the entire GRN. Computes all
@@ -344,7 +344,7 @@ def attribution_one_model(
     return attributions
 
 
-def inferrence_model_wise(models, data_train_full_tensor, gene_names, xai_method, n_models = [10, 25, 50], background_type = 'zeros'):
+def inferrence_model_wise(models, data_train_full_tensor, gene_names, xai_method = 'GuidedBackprop', n_models = [10, 25, 50], background_type = 'zeros'):
 
     """
     The main inferrence function to compute the entire GRN model wise. Computes all
@@ -430,10 +430,13 @@ def inferrence_model_wise(models, data_train_full_tensor, gene_names, xai_method
 
     # cou = cou.merge(top_egde_collector, left_index = True, right_on='edge_key')
 
-    grn_adata = attribution_to_anndata(attributions[keynames[0]], var=cou)
 
     if len(keynames)>0:
+        grn_adata = attribution_to_anndata(attributions[keynames[0]], var=cou)
+
         for k in keynames[1:len(keynames)]:
             # add remaining versions as masks
             grn_adata.layers[k] = attributions[k]
-    return grn_adata
+        return grn_adata
+    else:
+        return None

src/netmap/model/nbautoencoder.py

@@ -49,7 +49,7 @@ def forward(self, y_true, y_pred, theta):
         return torch.mean(loss)  # Return mean loss over the batch
 
 class NegativeBinomialAutoencoder(nn.Module):
-    def __init__(self, input_dim, latent_dim, dropout_rate=0.0, hidden_dims=[64, 64, 64], activation='relu'):
+    def __init__(self, input_dim, latent_dim, dropout_rate=0.1, hidden_dims=[64], activation='relu'):
         super(NegativeBinomialAutoencoder, self).__init__()
 
         # Mapping for the requested activation functions

src/netmap/model/train_model.py

@@ -7,7 +7,7 @@
 from tqdm import tqdm
 
 
-def create_model_zoo(data_tensor, n_models = 10, n_epochs = 10000, model_type = 'ZINBAutoencoder', dropout_rate = 0.02, latent_dim=8, hidden_dim=[128]):
+def create_model_zoo(data_tensor, n_models = 10, n_epochs = 10000, model_type = 'ZINBAutoencoder', dropout_rate = 0.1, latent_dim=8, hidden_dim=[64]):
     """ Creates a set of Autoencoders of the data using the speicified architecture. The architecture of the encoder can be specified using 
     the `hidden_dim` parameter, the decoder architecture is mirrored. Early stopping is used by default.