Back to Recipes Homepage
recipe picture
Recipe: iwdn HBNW 1.3
4.90909
Your rating: None Average: 4.9 (11 votes)

Profile

Name: iwdn HBNW 1.3
ID: 105175
Created on: Mon, 09/06/2021 - 14:18
Updated on: Mon, 09/06/2021 - 21:18
Description:

Finds H-Bond networks



Best For


Comments

Joined: 03/28/2020
Groups: Go Science
Release notes iwdn hbnw version 1.3

This recipe is meant for finding semi-automatically H-Bond networks. (HBNW: H-Bond-Networker). It builds on the excellent work of spvincent's H-Network Probe (HNP).
The general idea of HBNW is quite similar to HNP but HBNW does not depend on the H-Bond network filter to identify and store nets. Instead all bonds are evaluated and especially in the Auto-Grow (AG) mode a found net is tried to be extended in further iterations. Like this, a net will in many cases grow all over the structure and very huge nets are possible. This recipe was very helpful for me in order to understand how H-Bond networks should be found and formed and I used this recipe for example to find the network described in Lab report #20.
The basic procedure is described in the header of this recipe and you will know most of it already from HNP. It works on unfrozen sidechains and these are randomly mutated in order to find H-Bond-connections. Shake and wiggle are used to sort out clashes and thereby it is tried to find stable bonds. This is done in multiple iterations in AG-mode and for puzzles with H-Bond bonus this often leads to quickly finding (in a matter of minutes to hours) a multitude of bonus-nets. There is no limit on the number of unfrozen sidechains to work on but of course things get slower the more sidechains you leave unfrozen.
Be aware though that the found nets are often "raw" nets which need manual optimization. Especially BUNs in the net should be removed or further bonded in order to get really good bonus. Further manual shakes and wiggles need to be done to stabilize a net. And sometimes whole branches of a net should be removed or "cleaned up" in order to get a good-scoring and nice-looking net. So it is no "wonder-tool" but it can show you how networks could look like and where on the structure they are likely to form.
There are two main modes (single- and Auto-Grow) and multiple options in this recipe. I recommend using only AG-mode which leads to larger nets in shorter time. The default settings should work well for most cases (AG-mode is active by default). If it is desired, I can release a short description of each option. For the start just leave the default options active and then try to play around with the options if you are interested.
This recipe can be used on every kind of puzzle. Especially on binder-puzzles it can be used to find interface-bonds or even whole networks.
For ligands there is the option to accept only bonds to a single segment ("AG connect only to fixed center") and with this all kinds of bonds to the ligand can be found in relatively short time.
This recipe has taught me how and where networks should be formed and it enabled me to create them manually and quickly now. I hope that this recipe can be useful to whomever is interested in finding and understanding H-Bond networks.

Thanks again to spvincent for his excellent pioneer's work with H Network Probe! Also many thanks go to Jeff101 for the good discussions we had on functions and features for this recipe. This has been very helpful even though I did not find the time to implement it all.

This recipe is quite a complicated "beast" and it gave me quite a headache while implementing some routines. But after all it was still fun to figure it out. Bugs are likely to still be in there even though I tried to test it as thoroughly as possible.

If you have questions, feel free to contact me. And as always: Happy folding and greetings to all, iwdn

Want to try?
Add to Cookbook!
To download recipes to your cookbook, you need to have the game client running.
Parent
Children

none

Authors
Sitemap

Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, Boehringer Ingelheim, RosettaCommons