Update DB5 README

Author: amorehead

project/datasets/DB5/README

@@ -1,24 +1,27 @@
 Cleaned up version of Docking Benchmark 5 (https://zlab.umassmed.edu/benchmark/). 
 
-Released with "End-to-End Learning on 3D Protein Structure for Interface Prediction."
+Originally released with "End-to-End Learning on 3D Protein Structure for Interface Prediction."
 by Raphael J.L. Townshend, Rishi Bedi, Patricia Suriana, Ron O. Dror
 https://arxiv.org/abs/1807.01297
 
 Specifically, bound chains and residue indexes were aligned across unbound and bound complexes.
 
 A total of 230 binary protein complexes are included.
 
-Processing code to regenerate and use the provided tfrecords is located at
+Processing code to regenerate and use the original provided tfrecords is located at
 https://github.com/drorlab/DIPS
 
 MANIFEST
 
-raw/            - All pre-aligned and cleaned DB5 structures, organized into directories 
-                  with individual files for ligand-unbound, ligand-bound, receptor-unbound, 
-                  receptor-bound.
+final/
+    raw/            - List of all postprocessed (i.e. feature-enriched) pairs.
+
 interim/
-    parsed/     - All DB5 structures processed to pickled dataframes.
-    complexes/  - List of all possible pairs in parsed.
-    pairs/      - Dill files of individual pairs listed in complexes.
-processed/
-    tfrecords/  - pairs converted to tfrecords.
+    complexes/      - List of all possible pairs in parsed.
+    external_feats/ - List of all residue-level features generated using external software packages.
+    pairs/          - Dill files of individual pairs listed in complexes.
+    parsed/         - All DB5 structures processed to pickled dataframes.
+
+raw/                - All pre-aligned and cleaned DB5 structures, organized into directories
+                    with individual files for ligand-unbound, ligand-bound, receptor-unbound,
+                    receptor-bound.