Skip to content

Fix protein Conv1d dimensions and align CNN architecture with paper #21

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

Open
matija-marijan wants to merge 2 commits into
base: master
Choose a base branch
from
Open

Fix protein Conv1d dimensions and align CNN architecture with paper #21

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
92d3a02
Transpose protein embedding matrix, apply three Conv1d with max pooling
matija-marijan Dec 19, 2025
4c567bc
Add ReLU activations between Conv1d layers
matija-marijan Dec 19, 2025
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
20 changes: 13 additions & 7 deletions models/gat.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,8 +18,10 @@ def __init__(self, num_features_xd=78, n_output=1, num_features_xt=25,

# 1D convolution on protein sequence
self.embedding_xt = nn.Embedding(num_features_xt + 1, embed_dim)
self.conv_xt1 = nn.Conv1d(in_channels=1000, out_channels=n_filters, kernel_size=8)
self.fc_xt1 = nn.Linear(32*121, output_dim)
self.conv_xt_1 = nn.Conv1d(in_channels=embed_dim, out_channels=n_filters, kernel_size=8)
self.conv_xt_2 = nn.Conv1d(in_channels=n_filters, out_channels=2*n_filters, kernel_size=8)
self.conv_xt_3 = nn.Conv1d(in_channels=2*n_filters, out_channels=3*n_filters, kernel_size=8)
self.fc1_xt = nn.Linear(3*n_filters, output_dim)

# combined layers
self.fc1 = nn.Linear(256, 1024)
Expand All @@ -46,12 +48,16 @@ def forward(self, data):
# protein input feed-forward:
target = data.target
embedded_xt = self.embedding_xt(target)
conv_xt = self.conv_xt1(embedded_xt)
conv_xt = self.relu(conv_xt)
embedded_xt = torch.permute(embedded_xt, (0, 2, 1))

# flatten
xt = conv_xt.view(-1, 32 * 121)
xt = self.fc_xt1(xt)
conv_xt = self.conv_xt_1(embedded_xt)
conv_xt = self.relu(conv_xt)
conv_xt = self.conv_xt_2(conv_xt)
conv_xt = self.relu(conv_xt)
conv_xt = self.conv_xt_3(conv_xt)
conv_xt = self.relu(conv_xt)
xt = torch.max(conv_xt, dim = -1)[0]
xt = self.fc1_xt(xt)

# concat
xc = torch.cat((x, xt), 1)
Expand Down
16 changes: 12 additions & 4 deletions models/gat_gcn.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,8 +23,10 @@ def __init__(self, n_output=1, num_features_xd=78, num_features_xt=25,

# 1D convolution on protein sequence
self.embedding_xt = nn.Embedding(num_features_xt + 1, embed_dim)
self.conv_xt_1 = nn.Conv1d(in_channels=1000, out_channels=n_filters, kernel_size=8)
self.fc1_xt = nn.Linear(32*121, output_dim)
self.conv_xt_1 = nn.Conv1d(in_channels=embed_dim, out_channels=n_filters, kernel_size=8)
self.conv_xt_2 = nn.Conv1d(in_channels=n_filters, out_channels=2*n_filters, kernel_size=8)
self.conv_xt_3 = nn.Conv1d(in_channels=2*n_filters, out_channels=3*n_filters, kernel_size=8)
self.fc1_xt = nn.Linear(3*n_filters, output_dim)

# combined layers
self.fc1 = nn.Linear(256, 1024)
Expand All @@ -46,9 +48,15 @@ def forward(self, data):
x = self.fc_g2(x)

embedded_xt = self.embedding_xt(target)
embedded_xt = torch.permute(embedded_xt, (0, 2, 1))

conv_xt = self.conv_xt_1(embedded_xt)
# flatten
xt = conv_xt.view(-1, 32 * 121)
conv_xt = self.relu(conv_xt)
conv_xt = self.conv_xt_2(conv_xt)
conv_xt = self.relu(conv_xt)
conv_xt = self.conv_xt_3(conv_xt)
conv_xt = self.relu(conv_xt)
xt = torch.max(conv_xt, dim = -1)[0]
xt = self.fc1_xt(xt)

# concat
Expand Down
16 changes: 12 additions & 4 deletions models/gcn.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,8 +22,10 @@ def __init__(self, n_output=1, n_filters=32, embed_dim=128,num_features_xd=78, n

# protein sequence branch (1d conv)
self.embedding_xt = nn.Embedding(num_features_xt + 1, embed_dim)
self.conv_xt_1 = nn.Conv1d(in_channels=1000, out_channels=n_filters, kernel_size=8)
self.fc1_xt = nn.Linear(32*121, output_dim)
self.conv_xt_1 = nn.Conv1d(in_channels=embed_dim, out_channels=n_filters, kernel_size=8)
self.conv_xt_2 = nn.Conv1d(in_channels=n_filters, out_channels=2*n_filters, kernel_size=8)
self.conv_xt_3 = nn.Conv1d(in_channels=2*n_filters, out_channels=3*n_filters, kernel_size=8)
self.fc1_xt = nn.Linear(3*n_filters, output_dim)

# combined layers
self.fc1 = nn.Linear(2*output_dim, 1024)
Expand Down Expand Up @@ -54,9 +56,15 @@ def forward(self, data):

# 1d conv layers
embedded_xt = self.embedding_xt(target)
embedded_xt = torch.permute(embedded_xt, (0, 2, 1))

conv_xt = self.conv_xt_1(embedded_xt)
# flatten
xt = conv_xt.view(-1, 32 * 121)
conv_xt = self.relu(conv_xt)
conv_xt = self.conv_xt_2(conv_xt)
conv_xt = self.relu(conv_xt)
conv_xt = self.conv_xt_3(conv_xt)
conv_xt = self.relu(conv_xt)
xt = torch.max(conv_xt, dim = -1)[0]
xt = self.fc1_xt(xt)

# concat
Expand Down
16 changes: 12 additions & 4 deletions models/ginconv.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -41,8 +41,10 @@ def __init__(self, n_output=1,num_features_xd=78, num_features_xt=25,

# 1D convolution on protein sequence
self.embedding_xt = nn.Embedding(num_features_xt + 1, embed_dim)
self.conv_xt_1 = nn.Conv1d(in_channels=1000, out_channels=n_filters, kernel_size=8)
self.fc1_xt = nn.Linear(32*121, output_dim)
self.conv_xt_1 = nn.Conv1d(in_channels=embed_dim, out_channels=n_filters, kernel_size=8)
self.conv_xt_2 = nn.Conv1d(in_channels=n_filters, out_channels=2*n_filters, kernel_size=8)
self.conv_xt_3 = nn.Conv1d(in_channels=2*n_filters, out_channels=3*n_filters, kernel_size=8)
self.fc1_xt = nn.Linear(3*n_filters, output_dim)

# combined layers
self.fc1 = nn.Linear(256, 1024)
Expand All @@ -68,9 +70,15 @@ def forward(self, data):
x = F.dropout(x, p=0.2, training=self.training)

embedded_xt = self.embedding_xt(target)
embedded_xt = torch.permute(embedded_xt, (0, 2, 1))

conv_xt = self.conv_xt_1(embedded_xt)
# flatten
xt = conv_xt.view(-1, 32 * 121)
conv_xt = self.relu(conv_xt)
conv_xt = self.conv_xt_2(conv_xt)
conv_xt = self.relu(conv_xt)
conv_xt = self.conv_xt_3(conv_xt)
conv_xt = self.relu(conv_xt)
xt = torch.max(conv_xt, dim = -1)[0]
xt = self.fc1_xt(xt)

# concat
Expand Down