Skip to content

align

BatchAlign 

BatchAlign(
    n_epochs: int = 10,
    batch_size: int = 128,
    learning_rate: float = 0.0003,
    n_dis: int = 3,
    GPU: Union[bool, str] = True,
    random_state: Optional[int] = None,
    weight: Optional[Dict[str, float]] = None,
)
Source code in src\stands\align.py 
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
def __init__(self, 
             n_epochs: int = 10, 
             batch_size: int = 128,
             learning_rate: float = 3e-4, 
             n_dis: int = 3,
             GPU: Union[bool, str] = True, 
             random_state: Optional[int] = None,
             weight: Optional[Dict[str, float]] = None):

    self.n_epochs = n_epochs
    self.batch_size = batch_size
    self.lr = learning_rate
    self.n_dis = n_dis
    self.device = select_device(GPU)
    self.GPU = GPU

    self.seed = random_state
    if random_state is not None:
        seed_everything(random_state)

    if weight is None:
        weight = {'w_rec': 30, 'w_adv': 1, 'w_gp': 10}
    self.weight = weight

UpdateD 

UpdateD(blocks)

Updating discriminator

Source code in src\stands\align.py 
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
def UpdateD(self, blocks):
    '''Updating discriminator'''
    self.opt_D.zero_grad()

    # generate fake data
    batchid = blocks[-1].dstdata['batch']
    fake_g = self.G.STforward(blocks, blocks[0].srcdata['gene'], batchid)

    # get real data from blocks
    real_g = blocks[1].dstdata['gene']

    d1 = torch.mean(self.D.SCforward(real_g))
    d2 = torch.mean(self.D.SCforward(fake_g.detach()))
    gp = calculate_gradient_penalty(self.D, real_g, fake_g.detach())         

    # store discriminator loss for printing training information
    self.D_loss = - d1 + d2 + gp * self.weight['w_gp']
    self.D_loss.backward()
    self.opt_D.step()

UpdateG 

UpdateG(blocks)

Updating generator

Source code in src\stands\align.py 
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
def UpdateG(self, blocks):
    '''Updating generator'''
    self.opt_G.zero_grad()

    # generate fake data
    batchid = blocks[-1].dstdata['batch']
    fake_g = self.G.STforward(blocks, blocks[0].srcdata['gene'], batchid)

    # get real data from blocks
    real_g = blocks[1].dstdata['gene']

    # discriminator provides feedback
    d = self.D.SCforward(fake_g)

    Loss_rec = self.L1(real_g, fake_g)
    Loss_adv = - torch.mean(d)

    # store generator loss for printing training information and backward
    self.G_loss = self.weight['w_rec']*Loss_rec + self.weight['w_adv']*Loss_adv
    self.G_loss.backward()
    self.opt_G.step()

fit 

fit(
    raw: Dict[str, Any],
    generator: GeneratorAD,
    **alignerkwargs
)

Remove batch effects

Source code in src\stands\align.py 
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
def fit(self, raw: Dict[str, Any], generator: GeneratorAD, **alignerkwargs):
    '''Remove batch effects'''
    adatas = raw['adata']
    adata_ref = adatas[0]
    adata_tgt = ad.concat(adatas[1:])

    # find Kin Pairs
    Aligner = FindPairs(GPU=self.GPU, random_state=self.seed, **alignerkwargs)
    _, tgt_g = Aligner.fit(generator, raw)

    self.sampler = dgl.dataloading.MultiLayerFullNeighborSampler(2)
    self.dataset = dgl.dataloading.DataLoader(
        tgt_g, tgt_g.nodes(), self.sampler, batch_size=self.batch_size, 
        shuffle=True, drop_last=False, num_workers=0, device=self.device
    )

    self.init_model(raw, generator)

    tqdm.write('Begin to correct spatial transcriptomics datasets...')
    self.G.train()
    self.D.train()
    with tqdm(total=self.n_epochs) as t:
        for _ in range(self.n_epochs):
            t.set_description(f'Train Epochs')

            for _, _, blocks in self.dataset:

                # Update discriminator for n_dis times
                for _ in range(self.n_dis):
                    self.UpdateD(blocks)

                # Update generator for one time
                self.UpdateG(blocks)

            # Update learning rate for G and D
            self.D_sch.step()
            self.G_sch.step()
            t.set_postfix(G_Loss = self.G_loss.item(),
                          D_Loss = self.D_loss.item())
            t.update(1)

    self.dataset = dgl.dataloading.DataLoader(
        tgt_g, tgt_g.nodes(), self.sampler, batch_size=self.batch_size, 
        shuffle=False, drop_last=False, num_workers=0, device=self.device
    )

    self.G.eval()
    corrected = []
    with torch.no_grad():
        for _, _, blocks in self.dataset:
            fake_g = self.G.STforward(
                blocks, blocks[0].srcdata['gene'], blocks[-1].dstdata['batch']
            )
            corrected.append(fake_g.cpu().detach())

    corrected = torch.cat(corrected, dim=0).numpy()
    adata_tgt.X = corrected
    adata = ad.concat([adata_ref, adata_tgt])
    tqdm.write('Datasets have been corrected.\n')
    return adata

FindPairs 

FindPairs(
    n_epochs: int = 1000,
    learning_rate: float = 0.0002,
    GPU: Union[bool, str] = True,
    random_state: Optional[int] = None,
    weight: Optional[Dict[str, float]] = None,
)
Source code in src\stands\align.py 
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
def __init__(self, 
             n_epochs: int = 1000, 
             learning_rate: float = 2e-4,
             GPU: Union[bool, str] = True, 
             random_state: Optional[int] = None,
             weight: Optional[Dict[str, float]] = None):

    self.n_epochs = n_epochs
    self.lr = learning_rate
    self.device = select_device(GPU)

    if random_state is not None:
        seed_everything(random_state)

    if weight is None:
        weight = {'w_rec': 30, 'w_adv': 1, 'w_gp': 10}
    self.weight = weight

UpdateD 

UpdateD(z_ref, z_tgt)

Updating discriminator

Source code in src\stands\align.py 
105
106
107
108
109
110
111
112
113
114
115
116
117
def UpdateD(self, z_ref, z_tgt):
    '''Updating discriminator'''
    self.opt_D.zero_grad()

    z_ref, z_tgt, _ = self.M(z_ref, z_tgt)
    d1 = torch.mean(self.D.Zforward(z_ref.detach()))
    d2 = torch.mean(self.D.Zforward(z_tgt.detach()))
    gp = calculate_gradient_penalty(self.D, z_ref.detach(), z_tgt.detach(), Zforward=True)

    # store discriminator loss for printing training information
    self.D_loss = - d1 + d2 + gp * self.weight['w_gp']
    self.D_loss.backward()
    self.opt_D.step()

UpdateM 

UpdateM(z_ref, z_tgt)

Updating mapping matrix

Source code in src\stands\align.py 
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
def UpdateM(self, z_ref, z_tgt):
    '''Updating mapping matrix'''
    self.opt_M.zero_grad()

    # reconstruct z_tgt with z_ref
    fake_z_tgt, z_tgt, _ = self.M(z_ref, z_tgt)
    d = self.D.Zforward(fake_z_tgt)

    Loss_rec = self.Loss(z_tgt, fake_z_tgt)
    Loss_adv = -torch.mean(d)

    # store generator loss for printing training information and backward
    self.M_loss = self.weight['w_rec']*Loss_rec + self.weight['w_adv']*Loss_adv
    self.M_loss.backward()
    self.opt_M.step()

fit 

fit(generator: GeneratorAD, raw: Dict[str, Any])

Find Kin Pairs

Source code in src\stands\align.py 
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
def fit(self, generator: GeneratorAD,  raw: Dict[str, Any]):
    '''Find Kin Pairs'''
    tqdm.write('Begin to find Kin Pairs between datasets...')

    raw_g = raw['graph']
    ref_g, tgt_g = self.split(raw_g)
    self.G = generator.to(self.device)

    # freeze self.G weight
    for param in self.G.parameters():
        param.requires_grad = False

    self.init_model(raw, ref_g.num_nodes(), tgt_g.num_nodes())

    self.M.train()
    self.D.train()
    with tqdm(total=self.n_epochs) as t:
        for _ in range(self.n_epochs):
            t.set_description(f'Train Epochs')

            # generate embeddings
            z_ref = self.G.extract.GeneEncoder(ref_g, ref_g.ndata['gene'])
            z_tgt = self.G.extract.GeneEncoder(tgt_g, tgt_g.ndata['gene'])

            self.UpdateD(z_ref, z_tgt)
            self.UpdateM(z_ref, z_tgt)

            # update learning rate for G and D
            self.D_sch.step()
            self.M_sch.step()
            t.set_postfix(G_Loss = self.M_loss.item(),
                          D_Loss = self.D_loss.item())
            t.update(1)

    self.M.eval()
    with torch.no_grad():
        z_ref = self.G.extract.GeneEncoder(ref_g, ref_g.ndata['gene'])
        z_tgt = self.G.extract.GeneEncoder(tgt_g, tgt_g.ndata['gene'])
        _, _, m = self.M(z_ref, z_tgt)
        pair_id = list(ref_g.nodes().cpu().numpy()[m.argmax(axis=1)])
        ref_g = dgl.node_subgraph(ref_g, pair_id)
        tgt_g.ndata['ref_gene'] = ref_g.ndata['gene']

    tqdm.write('Kin Pairs have been found.\n')
    return ref_g, tgt_g

split 

split(graph: DGLGraph)

Split the integrated graph to reference and target graph

Source code in src\stands\align.py 
80
81
82
83
84
85
86
87
def split(self, graph: dgl.DGLGraph):
    '''Split the integrated graph to reference and target graph'''
    idx = torch.argmax(graph.ndata['batch'], -1).numpy()
    ref_id = list(np.where(idx == 0)[0])
    tgt_id = list(np.where(idx != 0)[0])
    ref_g = dgl.node_subgraph(graph, ref_id).to(self.device)
    tgt_g = dgl.node_subgraph(graph, tgt_id).to(self.device)
    return ref_g, tgt_g