Merge branch 'main' into pinnacle_networks

README.md

@@ -116,6 +116,7 @@ We provide  tutorials to get started with TDC:
 | [202](https://colab.research.google.com/drive/1kYH8nt3nW7tXYBPNcfYuDbWxGTqOEnWg?usp=sharing) | TDC-2 Resource and PrimeKG |
 | [203](https://colab.research.google.com/drive/13MYlg5tWpywWbKYsJQXafKAlVF2hz-sP?usp=sharing) | TDC-2 Resource and External APIs |
 | [204](https://colab.research.google.com/drive/17Pd328W27mn-iBCRkHIa78L3pukKcfW1?usp=sharing) | TDC-2 Model Hub |
+| [205](https://colab.research.google.com/drive/1kHdFG4gUic5nmiul7b1hUh0HLCxLQnw_?usp=sharing) | TDC-2 Molecular Property Cliff Prediction Task |
 
 
 ## Design of TDC

tdc/single_pred/mpc.py

@@ -22,8 +22,30 @@ def __init__(self, name, path="./data"):
         self.name = name
         self.data = None
 
-    def get_data(self):
-        from MoleculeACE import Data, Descriptors  #TODO: support non-MoleculeACE
+    def get_from_gh(self, link):
+        import pandas as pd
+        import requests
+        import io
+
+        data = requests.get(link)
+        try:
+            data.raise_for_status()
+        except:
+            raise Exception(
+                "invalid link provided. choose a link for datasets in https://github.com/bidd-group/MPCD"
+            )
+        self.data = pd.read_csv(io.StringIO(data.text))
+        return self.data
+
+    def get_data(self, link=None, get_from_gh=True):
+        if (not get_from_gh) and link is None:
+            raise Exception(
+                "provide dataset github link from https://github.com/bidd-group/MPCD"
+            )
+        elif get_from_gh:
+            return self.get_from_gh(link)
+        # support direct interfface with MoleculeACE API as well
+        from MoleculeACE import Data, Descriptors
         try:
             self.data = Data(self.name)
             self.data(Descriptors.SMILES)