Dear Modellers, I am facing some problem with the recently obtained results. Is there any way to improve the models through salign-iterative.py? The script calculates the percent topology of optimal superimposition. But, when implemented over a set of considered near-native decoys, it fails to further improve the topology (not even a bit of TM-Score undoubtedly and consistently) any single time. Rather, sometimes, the model accuracy was further decreased in terms of GDT-HA. However, this is expected as it would have taken the average overall model topology (as clear from the mutually slided model sequences in the alignment file when employed in model-multiple.py). Is there any way to precisely delimit the boundaries, where there is a huge difference (more than a specified threshold) in the backbone distance map, probably with salign-iterative so that the local model topology could be relaxed? Calpha distance 3.5A' is employed. Even with its alterations, expectations were not met.

If this is incorrect, what is the mistake. Or, Is there any other way to cluster some good models through MODELLER?

Ashish Runthala, Lecturer, Structural Biology Cell, Biological Sciences Group, BITS, Pilani Rajasthan, INDIA