AttributeError: 'tuple' object has no attribute 'dim'

[Himanshu-at-bayer]

I am currently runnine the disease critical netwrok analysis using the sample code. I have already modified the code for the group function in torch.geometric/nn/conv/hetero_conv.py

import sys from kgwas import KGWAS, KGWAS_Data

data = KGWAS_Data(data_path= "./data/")
All required data files are present.

data.load_kg()
--loading KG---
--using enformer SNP embedding--
--using random go embedding--
--using ESM gene embedding--

data.process_gwas_file()

Using ldsc weight...
ldsc_weight mean: 0.9999999999999993

run= KGWAS(data, exp_name="KGWAS")

run.load_pretrained('./data/model/KGWAS/')

df_network_weight, df_variant_interpretation, disease_critical_network = run.get_disease_critical_network( variant_threshold = 5e-8, magma_path = None, magma_threshold = 0.05, program_threshold = 0.05, K_neighbors = 3, num_cpus = 1)

Retrieving weights...

AttributeError Traceback (most recent call last)
Cell In[17], line 1
----> 1 df_network_weight, df_variant_interpretation, disease_critical_network = run.get_disease_critical_network(variant_threshold = 5e-8, magma_path = None, magma_threshold = 0.05, program_threshold = 0.05,
2 K_neighbors = 3, num_cpus = 1)

File ~/.local/lib/python3.10/site-packages/kgwas/kgwas.py:271, in KGWAS.get_disease_critical_network(self, variant_threshold, magma_path, magma_threshold, program_threshold, K_neighbors, num_cpus)
268 def get_disease_critical_network(self, variant_threshold = 5e-8,
269 magma_path = None, magma_threshold = 0.05, program_threshold = 0.05,
270 K_neighbors = 3, num_cpus = 1):
--> 271 df_network_weight = get_network_weight(self, self.data)
272 df_variant_interpretation, disease_critical_network = generate_viz(self, df_network_weight, self.data_path, variant_threshold, magma_path, magma_threshold, program_threshold, K_neighbors, num_cpus)
273 return df_network_weight, df_variant_interpretation, disease_critical_network

File ~/.local/lib/python3.10/site-packages/kgwas/utils.py:453, in get_network_weight(run, data)
450 x_dict['MolecularFunction'] = model.go_feat_mlp(x_dict['MolecularFunction'])
452 for conv in model.convs:
--> 453 x_dict = conv(x_dict, edge_index_dict,
454 return_attention_weights_dict = dict(zip(list(graph_data.edge_index_dict.keys()),
455 [True] * len(list(graph_data.edge_index_dict.keys())))),
456 return_raw_attention_weights_dict = dict(zip(list(graph_data.edge_index_dict.keys()),
457 [True] * len(list(graph_data.edge_index_dict.keys())))),
458 )
459 attention_layer = {i: j[1] for i,j in x_dict.items()}
460 attention_all_layers.append(attention_layer)

File ~/.local/lib/python3.10/site-packages/torch/nn/modules/module.py:1736, in Module._wrapped_call_impl(self, *args, **kwargs)
1734 return self._compiled_call_impl(*args, **kwargs) # type: ignore[misc]
1735 else:
-> 1736 return self._call_impl(*args, **kwargs)

File ~/.local/lib/python3.10/site-packages/torch/nn/modules/module.py:1747, in Module._call_impl(self, *args, **kwargs)
1742 # If we don't have any hooks, we want to skip the rest of the logic in
1743 # this function, and just call forward.
1744 if not (self._backward_hooks or self._backward_pre_hooks or self._forward_hooks or self._forward_pre_hooks
1745 or _global_backward_pre_hooks or _global_backward_hooks
1746 or _global_forward_hooks or _global_forward_pre_hooks):
-> 1747 return forward_call(*args, **kwargs)
1749 result = None
1750 called_always_called_hooks = set()

File ~/.local/lib/python3.10/site-packages/torch_geometric/nn/conv/hetero_conv.py:160, in HeteroConv.forward(self, *args_dict, **kwargs_dict)
157 if not has_edge_level_arg:
158 continue
--> 160 out = conv(*args, **kwargs)
162 if dst not in out_dict:
163 out_dict[dst] = [out]

File ~/.local/lib/python3.10/site-packages/torch/nn/modules/module.py:1736, in Module._wrapped_call_impl(self, *args, **kwargs)
1734 return self._compiled_call_impl(*args, **kwargs) # type: ignore[misc]
1735 else:
-> 1736 return self._call_impl(*args, **kwargs)

File ~/.local/lib/python3.10/site-packages/torch/nn/modules/module.py:1747, in Module._call_impl(self, *args, **kwargs)
1742 # If we don't have any hooks, we want to skip the rest of the logic in
1743 # this function, and just call forward.
1744 if not (self._backward_hooks or self._backward_pre_hooks or self._forward_hooks or self._forward_pre_hooks
1745 or _global_backward_pre_hooks or _global_backward_hooks
1746 or _global_forward_hooks or _global_forward_pre_hooks):
-> 1747 return forward_call(*args, **kwargs)
1749 result = None
1750 called_always_called_hooks = set()

File ~/.local/lib/python3.10/site-packages/kgwas/conv.py:141, in GATConv.forward(self, x, edge_index, edge_attr, size, pheno_emb, return_attention_weights, return_raw_attention_weights)
139 else: # Tuple of source and target node features:
140 x_src, x_dst = x
--> 141 assert x_src.dim() == 2, "Static graphs not supported in 'GATConv'"
142 x_src = self.lin_src(x_src).view(-1, H, C)
143 if x_dst is not None:

AttributeError: 'tuple' object has no attribute 'dim'
​

Any idea what could be wrong here? I have modified the codes as mentioned in the previous issues but still no luck :(