+# At high identities, around 96% and above, compressed indexes are often more sensitive, faster
+# and use less RAM. Compressed indexes are disabled by default, so I generally recommend that
+# you specify the --slots and --w options when clustering at high identities.
+