init

Author: Yan98

README.md

@@ -4,9 +4,6 @@ This repository contains the PyTorch code for our paper "Spatial Transcriptomics
 
 > [paper]() | [arxiv](https://arxiv.org/pdf/2401.14772)
 
-**The code will come soon!**
-
-
 ## Introduction
 Spatial transcriptomics (ST) captures gene expression fine-grained distinct regions (i.e., windows) of a tissue slide. Traditional supervised learning frameworks applied to model ST are constrained to predicting expression of gene types seen during training from slide image windows, failing to generalize to unseen gene types. To overcome this limitation, we propose a semantic guided network, a pioneering zero- shot gene expression prediction framework. Considering a gene type can be described by functionality and phenotype, we dynamically embed a gene type to a vector per its functionality and phenotype, and employ this vector to project slide image windows to gene expression in feature space, unleashing zero-shot expression prediction for unseen gene types. The gene type functionality and phenotype are queried with a carefully designed prompt from a pre-trained large language model. On standard benchmark datasets, we demonstrate competitive zero-shot performance compared to past state-of-the-art supervised learning approaches.
 
@@ -20,16 +17,44 @@ Spatial transcriptomics (ST) captures gene expression fine-grained distinct regi
 <img src="asset/model.png", width=500/>
 </div>
 
-## Requirements
+## Dependency
+* python 3.10.13
+* pytorch_lightning 1.6.4
+* tifffile 2024.2.12
+* Pillow 10.2.0
+* scanpy 1.10.2
+* torch 2.2.1+cu118
 
-Please refer to [requirements.txt](./requirements.txt).
+## Dataset
+* Obtain [10xgenomics dataset](https://www.10xgenomics.com/resources/datasets?query=&page=1&configure%5Bfacets%5D%5B0%5D=chemistryVersionAndThroughput&configure%5Bfacets%5D%5B1%5D=pipeline.version&configure%5BhitsPerPage%5D=500&configure%5BmaxValuesPerFacet%5D=1000).
 
-## How to run
+## Train SGN
+* Change system directory
+```bash
+cd v1
+```
 
+*  Extract features and build graph
 ```bash
-python main.py
+
+python3 extract_feature.py --file_path Please fill # Set the file_path property to the location where the downloaded data will be stored. Remember to unzip the spatial.zip file.
+python3 generate_graph.py  --file_path Please fill # Set the file_path property to the location where the downloaded data will be stored. Remember to unzip the spatial.zip file.
+python3 name_to_feature.py
+
 ```
 
+* Gene expression prediction
+```bash
+cd ../
+python3 main.py # Feel free to adjust the arguments as necessary.
+```
+
+## Contact
+If you have any questions,  please drop [me](mailto:[email protected]?subject=[GitHub]SGN) an email.
+
+
+## Acknowledgement
+EVA-CLIP is built using the awesome [timm](https://github.com/huggingface/pytorch-image-models).
 
 ## Citation
 

main.py

@@ -0,0 +1,113 @@
+import argparse
+import os
+from model import GNN
+import pytorch_lightning as pl
+from functools import partial
+import torch
+import collections
+from train import TrainerModel
+from sklearn.model_selection import KFold
+import glob
+import torch_geometric
+
+
+def load_dataset(pts,file_path):
+    all_files = sorted(glob.glob(f"{file_path}/*.pt"))
+    print(all_files)
+    selected_files = []
+    for i in all_files:
+        for j in pts:
+            if i.endswith(str(j) + ".pt"):
+                graph = torch.load(i)
+                print(graph)
+                selected_files.append(graph)
+    return selected_files
+
+def main(args,idx):
+
+
+    XFOLD = glob.glob(f"{args.file_path}/*.pt")
+    skf = KFold(n_splits=3,shuffle= True, random_state = 12345)
+    KFOLD = []
+    for x in skf.split(XFOLD):
+        KFOLD.append(x)    
+
+
+    cwd = os.getcwd()
+    
+    def write(director,name,*string):
+        string = [str(i) for i in string]
+        string = " ".join(string)
+        with open(os.path.join(director,name),"a") as f:
+            f.write(string + "\n")
+            
+    args.folder_name = "log" + "/"  + str(idx)
+    store_dir = args.folder_name + "/" + "checkpoints_" + str(args.fold) + "/" 
+    print = partial(write,cwd, args.folder_name + "/" +"log_f" + str(args.fold))
+      
+    os.makedirs(store_dir, exist_ok= True)
+    
+    print(args)
+    
+
+    train_patient, test_patient = KFOLD[args.fold]
+    
+    train_dataset = load_dataset(train_patient,args.file_path)
+    test_dataset = load_dataset(test_patient,args.file_path)
+    
+    train_loader = torch_geometric.loader.DataLoader(
+        train_dataset,
+        batch_size=1,
+        )
+    
+    test_loader = torch_geometric.loader.DataLoader(
+        test_dataset,
+        batch_size=1,
+        )
+    
+    print(len(train_loader), len(test_loader))
+    
+    model = GNN(args.hidden_channels, args.embed_dim, args.out_channels, args.gnn_layer,args.feature_dim,args.name_dim)
+    CONFIG = collections.namedtuple('CONFIG', ['lr', 'logfun', 'verbose_step', 'weight_decay', 'store_dir'])
+    config = CONFIG(args.lr, print, args.verbose_step, args.weight_decay,store_dir)
+        
+    if args.checkpoints != None:
+        model.load_state_dict(torch.load(args.checkpoints,map_location = torch.device("cpu")))
+    
+    model = TrainerModel(config, model,args.meta, args.name_feature)
+    
+    plt = pl.Trainer(max_epochs = args.epoch,num_nodes=1, gpus=args.gpus, val_check_interval = args.val_interval,checkpoint_callback = False, logger = False)
+    plt.fit(model,train_dataloaders=train_loader,val_dataloaders=test_loader)
+
+if __name__ == "__main__":
+    
+    parser = argparse.ArgumentParser() 
+    parser.add_argument("--epoch", default = 300, type = int)
+    parser.add_argument("--fold", default = 0, type = int)
+    parser.add_argument("--gpus", default = 1, type = int)
+    parser.add_argument("--acce", default = "ddp", type = str)
+    parser.add_argument("--val_interval", default = 0.8, type = float)
+    parser.add_argument("--lr", default = 1e-4*5, type = float)
+    parser.add_argument("--verbose_step", default = 10, type = int)
+    parser.add_argument("--weight_decay", default = 1e-4, type = float)
+    parser.add_argument("--checkpoints", default = None, type = str)
+    parser.add_argument("--output", default = None, type = str)
+    parser.add_argument("--folder_name", default = "log", type = str)
+    parser.add_argument("--runs", default = 1, type = int)
+    parser.add_argument("--file_path", default="extracted_feature/resnet18/graph", type = str)
+    parser.add_argument("--name_feature", default="name_feature/Intel/neural-chat-7b-v3-1", type = str)
+    parser.add_argument("--meta", default="preprocess/", type = str)
+    parser.add_argument("--feature_dim", default=512, type = int)
+    parser.add_argument("--name_dim", default=4096, type = int)
+    parser.add_argument("--hidden_channels", default=512, type = int)
+    parser.add_argument("--embed_dim", default=256, type = int)
+    parser.add_argument("--out_channels", default=256, type = int)
+    parser.add_argument("--gnn_layer", default=4, type = int)
+    
+    
+    args = parser.parse_args()
+    for idx in range(args.runs):
+        for fold in range(3):
+            args.fold = fold
+            main(args,idx)
+