Big Update: Experimental Client (Drug Design)

Hello Everyone!

There is a new update to the ligand design client (experimental update group). In order to get the experimental client to work, follow the steps outlined here. There is a new puzzle, two new filters, and two new tools released with this update. Also, several bugs have been addressed and others noted:

Bug fixes that are complete
• Fix atom tree crashes when wiggling complex molecules/molecules clashing with backbone
• Fix errors associated with MMFF minimization
• Fix several loading issues
• Move MMFF to wiggle/shake/design submenu in selection tool
• Add new icon for MMFF

Known issues on the schedule for fixing
• When switching from design panel to undo, it is possible to have spheres left behind from the design panel.
• Original interface needs a position for MMFF button
• Switching tracks in undo menu works – occasionally
• Graphics for drawn 2D small molecules in Ligand Queue Tool needs to be greatly improved
• Scaling of bonus for Rule of Five and Similarity filter

New puzzle released
When a patient is diagnosed with HIV, they are almost immediately given anti-viral drugs that target HIV Protease 1. This protease is essential for the life cycle of HIV and inhibiting it helps prevent the spread of HIV within the body. To inhibit the protease, scientists have developed small molecules that trap the protease in a conformation that stops the enzyme from working normally.

To aid the scientists in their designs, they often look at specific ligand centric metrics. Ligand centric methods focus on attributes of the small molecule and ignore the protein. Two metrics, ligand similarity and Lipenski’s Rule of Five are often used during the design process. For this puzzle, we have provided two new filters that help track ligand similarity and Lipenski’s Rule of Five along with two new tools.

The new Rule of Five filter
Lipenski’s Rule of Five is a set of rules that are used to evaluate designed small molecules. The rules are based off of observations for known chemical entities that have made it to market. This is a ligand centric filter, meaning it only takes into account the small molecule being designed, not the protein – small molecule complex. You will be penalized if you violate the rules during design. The rules are:
• No more than 5 hydrogen bond donors
• No more than 10 hydrogen bond acceptors
• Molecular weight less than 500 daltons
• A logP of 5 or less

Molecular weight is determined by summing all the elements weights in a small molecule. For example, carbon has a weight of 12 and hydrogen has a weight of 1. If you have 6 carbons and 6 hydrogens, the weight would be 78. LogP is a measure of how soluble a small molecule is. The more soluble a molecule is, the more likely it is to be taken up into your body. LogP is determined based on the type of atoms present in a small molecule.

All of Lipenski’s rules are reported by the Rule of Five Filter; however, you can view these properties by clicking the ligand view tool.

The new Ligand Similarity filter
Ligand Similarity is used to describe the similarity between small molecules and known drugs. In the new HIV Protease puzzle, the Ligand Similarity filter is used specifically to identify how similar the designed small molecules are to known “tight” binders. This means that the higher the bonus received, the closer that you are to known chemical entities. This tops out at 75% similarity, to encourage looking for new designs. If you show the filter while using it, atoms that are similar to known binders will be highlighted. Changing those highlighted atoms will result in divergence away from known binders.

The new Ligand Queue tool
This tool will be changed from its current state to work more like the remix tool. The purpose of the Ligand Queue tool is to provide a set of small molecules to players; the set comes from actual experiments -– high-throughput screens, virtual high-throughput screens, and automated designs. For the HIV Protease puzzle, the small molecules that are shown are very low binders to HIV. Modifications of these ligands can result in making a small molecule that tightly binds HIV Protease. The ligands provide a good starting spot for design. This tool is still in a rough developmental state, but it is currently useful for providing context for a new concept.

High throughput screening and virtual screening are tools that scientists use to quickly identify small molecules that might bind a protein target. This is usually done by a robot that will screen millions of compounds rapidly. After the compounds have been screened, a process of identifying the molecule and modifying it occurs. This whole process is one of the first things done for each new protein target. We will discuss this more in a later blog post!

The new Ligand View tool
The ligand view tool provides ligand centric values that help you decide if the small molecule being designed is “drug-like”. The current values reported by the ligand view tool will be later incorporated into a filter to further guide you in your designs. In addition to the ligand centric metrics, there is also an iso surface tool that draws an isosurface around the ligand binding pocket to help better show packing against the ligand.

Enjoy! Post your questions and comments here and we'll do our best to get them addressed in future updates.

( Posted by  free_radical 83 1754  |  Wed, 05/18/2016 - 19:52  |  3 comments )
Joined: 12/27/2012
Groups: Beta Folders
some issues, where to report?

I've noticed the new experimental release is very slow to stop a shake, taking at least a minute.

Clicking on a ligand in the ligand queue interface displayed the ligand as a guide (?) but the client crashed quickly after that.

I could post the log.txt here, but that makes for a not-very-readable thread. Would it make more sense to open feedbacks for the new issues, allowing the use of attachments instead?

The client's last words may have been:

core.scoring.hbonds.hbonds_geom: NAN occurred in H-bonding calculations!
core.scoring.hbonds.hbonds_geom: Dxyz -NAN(IND) -NAN(IND) -NAN(IND) Hxyz -NAN(IND) -NAN(IND) -NAN(IND)

Joined: 01/12/2015
Groups: None
Hi LociOiling, You can post a

Hi LociOiling,

You can post a feedback for the new issues. Could you include if you are using the original interface or the selection interface. Also include if it is Windows or OSX.

A word about shake. During shake, rotamers for the small molecule have to be built, which can take a couple seconds to more than a minute, depending on the ligand. I plan on changing the way this is performed in a later release.

Joined: 01/12/2015
Groups: None
IRC is blocked at my office,

IRC is blocked at my office, so if you want to contact me, you can use discord. My handle is @free_radical.

Get Started: Download
  Windows    OSX    Linux  
(10.7 or later)

Are you new to Foldit? Click here.

Are you a student? Click here.

Are you an educator? Click here.
Only search
Recommend Foldit
User login
Top New Users

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, Microsoft, Adobe, RosettaCommons