Folding Forum

Sometimes I see some strange molecules ... Two complete separate stings like this one.



Is it
A) absolute correct (then how they stick together?)
B) some glitch in data provided from Core

Project 7505 (Run 0, Clone 26, Gen 297)
Atoms 443, Bonds 271

The same structure is shown in the official FAHViewer, too.

Same WU, different zoom: I can see some single atoms near the main structure. I get used to the single atom outside; those here seems inside the main proteine. Single atoms ? I understand additional H atoms in some cases. But based on color code those are different.

C) known issue with how the viewer renders the data.

For the sake of simpler rendering, the viewing box has virtual sides. As simple example, a chain of proteins sticking out the right side of the box enters again from the left. Like the old arcade game Asteroids.

In the past Bruce and I have discussed how to fix this. Without access to the dimensions and location of the periodic protein box, there is simply not enough information to piece it back together. I have some algorithms that can be used to very quickly identify the split pieces, but without more information I can't put them back together. One of the problems is that FAHClient filters out impossibly-long bonds before FAHViewer renders the protein. This filter needs to happen in FAHViewer so that the raw data can be accessed upstream by third parties. With the bond information, we could see how those sections fit together, allowing us to piece them together.

https://github.com/Jesse-V/Folding-Atomata/issues/26
Early rendering, but I was seeing it too: https://f.butt.github.com/assets/231441 ... 6fb1f1.png

Ok, understand that with the box. That would help with the question for the "satellite" protein part. It's just at it wrong place due to slicing the protein.
But not yet understand the single atoms. And based on their coordinates and orientation in the cube they seems to be part of the game; just not bonded correctly. But u need to re-read Jesse's git blog once again, after a coffee break.

Mathematically, it's absolutely correct but visually there is a problem, and as Jesse said, he can't fix it in his viewer without more information from FAHClient.

Suppose you have a protein that is almost as big as the invisible box. Now suppose somebody moves the box half it's width in the +X direction. Visually, half of the protein will be attached to the left wall and half will be attached to the right wall. A lot of bonds will cross the walls and will appear to be the wrong length because they appear to stretch (almost) all the way across the box. That would not only look ugly, it would be even more incorrect, visually, so those bonds are hidden by FAHClient. They need to be logically unhidden to recreate a single connected protein inside a single box.