fix

openfold/data/data_modules.py

@@ -57,6 +57,9 @@ class OpenFoldSingleDataset(torch.utils.data.Dataset):
                     Path to a directory containing template mmCIF files.
                 config:
                     A dataset config object. See openfold.config
+                chain_data_cache_path:
+                    Path to cache of data_dir generated by
+                    scripts/generate_chain_data_cache.py
                 kalign_binary_path:
                     Path to kalign binary.
                 max_template_hits:
@@ -121,26 +124,28 @@ class OpenFoldSingleDataset(torch.utils.data.Dataset):
                 c for c in self._chain_ids if c in chains_to_include
             ]
 
-        # Filter to include only chains where we have structure data
-        # (i.e. entries in chain_data_cache)
-        original_chain_ids = self._chain_ids
-        self._chain_ids = [
-            c for c in self._chain_ids if c in self.chain_data_cache
-        ]
-        if len(self._chain_ids) < len(original_chain_ids):
-            missing = [
-                c for c in original_chain_ids if c not in self.chain_data_cache
+        if self.chain_data_cache is not None:
+            # Filter to include only chains where we have structure data
+            # (entries in chain_data_cache)
+            original_chain_ids = self._chain_ids
+            self._chain_ids = [
+                c for c in self._chain_ids if c in self.chain_data_cache
             ]
-            max_to_print = 10
-            missing_examples = ", ".join(missing[:max_to_print])
-            if len(missing) > max_to_print:
-                missing_examples += ", ..."
-            logging.warning(
-                "Ignoring %d alignment entries (%s) that have no corresponding "
-                "entries in chain_data_cache (%s).",
-                len(missing),
-                missing_examples,
-                chain_data_cache_path)
+            if len(self._chain_ids) < len(original_chain_ids):
+                missing = [
+                    c for c in original_chain_ids
+                    if c not in self.chain_data_cache
+                ]
+                max_to_print = 10
+                missing_examples = ", ".join(missing[:max_to_print])
+                if len(missing) > max_to_print:
+                    missing_examples += ", ..."
+                logging.warning(
+                    "Removing %d alignment entries (%s) with no corresponding "
+                    "entries in chain_data_cache (%s).",
+                    len(missing),
+                    missing_examples,
+                    chain_data_cache_path)
        
         self._chain_id_to_idx_dict = {
             chain: i for i, chain in enumerate(self._chain_ids)