Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

fix: set edge weights for normalized adjacency in sctag #194

Merged
merged 5 commits into from
Feb 17, 2023
Merged

fix: set edge weights for normalized adjacency in sctag #194

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
64e6887
fix: specify edge weight to g_n
RemyLau Feb 17, 2023
28fa9c9
to tensor
RemyLau Feb 17, 2023
2e4235e
to tensor
WenzhuoTang Feb 17, 2023
fc28443
Update scTAG results
WenzhuoTang Feb 17, 2023
fc9f552
Merge branch 'hotfix' of github.com:OmicsML/dance into hotfix
WenzhuoTang Feb 17, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -244,7 +244,7 @@ Note: the data split modality of DeepImpute is different from ScGNN and GraphSCI
| scDCC | ARI | 0.82 / 0.81 | 0.98 / N/A | 0.51 / 0.58 | 0.60 / 0.66 |
| scDeepCluster | ARI | 0.81 / 0.78 | 0.98 / 0.97 | 0.51 / 0.52 | 0.56 / 0.58 |
| scDSC | ARI | 0.72 / 0.78 | 0.84 / N/A | 0.46 / 0.65 | 0.65 / 0.72 |
| scTAG | ARI | 0.75 / N/A | 0.96 / N/A | 0.53 / N/A | 0.60 / N/A |
| scTAG | ARI | 0.77 / N/A | 0.96 / N/A | 0.49 / N/A | 0.69 / N/A |

### Multimodality Module

Expand Down
12 changes: 6 additions & 6 deletions dance/modules/single_modality/clustering/sctag.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -95,8 +95,8 @@ def init_model(self, adj: np.ndarray, x: np.ndarray):
src_n, dist_n = np.nonzero(adj_n)

self.g = dgl.graph((src, dist)).to(self.device)
# FIX: set edge weight?...
self.g_n = dgl.graph((src_n, dist_n)).to(self.device)
self.g_n.edata["weight"] = torch.FloatTensor(adj_n[src_n, dist_n]).to(self.device)

self.mu = Parameter(torch.Tensor(self.n_clusters, self.latent_dim).to(self.device))
self.encoder1 = TAGConv(self.in_dim, self.hidden_dim, k=self.k)
Expand Down Expand Up @@ -149,13 +149,13 @@ def preprocessing_pipeline(n_top_genes: int = 3000, n_components: int = 50, n_ne
log_level=log_level,
)

def forward(self, adj_in, x_input):
def forward(self, g, x_input):
"""Forward propagation.

Parameters
----------
adj_in
Input adjacency matrix.
g
Input graph.
x_input
Input features.

Expand All @@ -175,8 +175,8 @@ def forward(self, adj_in, x_input):
Data dropout probability from ZINB.

"""
enc_h = self.encoder1(adj_in, x_input)
z = self.encoder2(adj_in, enc_h)
enc_h = self.encoder1(g, x_input, edge_weight=g.edata["weight"])
z = self.encoder2(g, enc_h, edge_weight=g.edata["weight"])
adj_out = self.decoder_adj(z)
_mean, _disp, _pi = self.decoder_x(z)
q = self.soft_assign(z)
Expand Down