diff --git a/extras/data_generation/structure_db/README b/extras/data_generation/structure_db/README
index 227313066aa288d29d215b5309f38ceab4689116..ed5115ddf021c6095061af6b05ae971248816da0 100644
--- a/extras/data_generation/structure_db/README
+++ b/extras/data_generation/structure_db/README
@@ -28,3 +28,13 @@ BUT BE AWARE, THE ACCORDING FREQUENCIES IN THE DATABASE WILL BE SET TO 0.0!
    initial database generated in step 2: 
    use assign_structure_profiles.py to perform this task
 
+
+To qualitatively reproduce the default StructuralDB in ProMod3, you first 
+perform step 1 and 2 with a non redundant set of protein structures as 
+defined by PISCES with around 25000 chains (e.g. seq id threshold: 90, 
+resolution threshold: 2.2). 
+Repeat step 1 and 2 with a smaller PISCES list (5000-6000 entries, 
+e.g. seq id threshold: 25 , resolution threshold: 1.8). 
+The first database serves as default StructureDB and the second db as 
+the source db for the structural profiles generated in steps 3-5.
+