From e7cd4f32a3b6508e2d950cd3e911abd6255d53fe Mon Sep 17 00:00:00 2001
From: Augustin Zidek <augustinzidek@google.com>
Date: Mon, 4 Jan 2021 15:23:20 +0000
Subject: [PATCH] Clarify how to open the dataset pickles under Python 3.

PiperOrigin-RevId: 350015548
---
 alphafold_casp13/README.md | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/alphafold_casp13/README.md b/alphafold_casp13/README.md
index b1318c2..bd49f59 100644
--- a/alphafold_casp13/README.md
+++ b/alphafold_casp13/README.md
@@ -81,7 +81,9 @@ file. Each target directory contains the following files:
     [TFRecord](https://www.tensorflow.org/tutorials/load_data/tfrecord) file
     with serialized tf.train.Example protocol buffers that contain the features
     needed to run the model.
-1.  `contacts/TARGET.pickle` file(s) with the predicted distogram.
+1.  `contacts/TARGET.pickle` file(s) with the predicted distogram. These pickles
+    were pickled using Python 2, so to unpickle them in Python 3 you will need
+    to set the `encoding='latin1'` optional argument for `pickle.load()`.
 1.  `contacts/TARGET.rr` file(s) with the contact map derived from the predicted
     distogram. The RR format is described on the
     [CASP website](http://predictioncenter.org/casp13/index.cgi?page=format#RR).