Simulating protein dynamics to find binding-competent states

Mutation to the tumor-suppressor protein p53 is a common feature in most cancers. MDM2–a protein whose job it is to downregulate p53 via their direct binding interaction–has therefore become a prime target for cancer therapeutics. Normally, a small helical region of p53 binds the MDM2 receptor site, but if a molecule with a similar shape can bind the receptor site of MDM2 strongly enough, it can prevent p53 from binding, thus making more p53 available to perform its tumor suppression abilities.

Many different kinds of molecular mimics of the p53 helix have been designed to disrupt the p53-MDM2 interaction, including stapled peptides, cyclic hairpin peptides, beta-peptides, peptoids (N-substituted oligoglycines), oligoarylamides, and spiroligomers, just to name a few. These molecules are much larger than typical small-molecule drugs, and have interesting folding properties that must be overcome to achieve tight binding. Stapled peptides, for instance, feature a hydrocarbon “staple” that helps rigidify the helical conformation in solution, which in turn enhances the binding affinity.

Using molecular simulations on Folding@home, we have been studying the coupled folding and binding of the p53 helix to MDM2 to address several key questions. One goal is to understand the roles the conformational dynamics in shaping the binding mechanism – such information can ultimately help to design better-binding molecular mimics.

Another question is whether or not molecular dynamics simulations can be used to discover binding-competent receptor conformations of MDM2 in the absence of a bound crystal structure. In new work from our lab (Pantelopulos et al. Proteins 2015), we show that ligand-free simulations of MDM2 starting from conformations with a closed binding cleft can sample open-cleft conformations capable of binding. We also tested the performance of several recent force field models in predicting experimental NMR measurements. We found that that all of the force fields perform similarly well, but that longer simulations (out to a microsecond) result in better agreement with experiment.

You can read about our work in the latest issue of Proteins:

Microsecond simulations of MDM2 and its complex with p53 yield insight into force field accuracy and conformational dynamics George A. Pantelopulos and Vincent A. Voelz. Proteins: Structure, Function and Bioinformatics, Accepted (2015)

First full version of our Folding@Home client for Android Mobile phones

We’re proud to announce the first full version of our Folding@Home client for Android Mobile phones.  This version is available to all Android Mobile phones with version 4.4 (Kitkat) and above.  Thanks to suggestions by donors we have redesigned the user interface and added new features such as:

  • contribute processing time continuously: just connect to a WiFi network and a charger.
  • one can login to Google Game Services, earn collaboration achievements, compete against his / her friends in processing time.
  • collaborate processing time from multiple devices under the same Google Game Services account.
  • settings screen has been removed. No need to configure anything!
  • details about the currently selected research type can be queried by touching the Research Type title on main screen, or by choosing “Active Research” on menu.

Scientifically, as in our previous beta run, we continue to focus on breast cancer with our mobile app. In this project, we’re investigating the nature of drug resistance mutations in key proteins (kinases) that are targets for breast cancer drugs. By studying the nature of how these mutations change these key drug targets, we will be able to both advance our basic biophysical understanding of these key proteins as well as build a tool to be used for patient specific breast cancer treatment— by sequencing the tumor and seeing what mutations are present, our tool seeks to recommend the best drug for a specific patient.

The beta has been a success with positive feedback from our community. We plan to add additional features in the comming months to further enhance the user interface and experience. The beta version has been downloaded more than 170,000 times worldwide, with more than 62,000 mobile phones contributing at the same time. We want to thank our community for your feedback and continued support.

We’ve also added a new movie to advertise and highlight the app.  It’s on YouTube here:

Problem with NaCl Client

We see that there is an issue with our NaCl Client (at http://folding.stanford.edu/nacl), with donors seeing this error:

Warning: Unexpected response to AS assignment request: error,DB ERROR: IO error: log.leveldb/016519.ldb: Too many open files

The server is being overloaded and our sysadmin team will take a look at it when they get back on Monday.  Long term, it looks like it’s time for us to upgrade the server this is running on since it’s getting overloaded.  We have been moving to get new servers ready for this and so getting to that should happen in about a week (servers are here and the sysadmin team has been working to getting them ready for their new roles in FAH).

New Core tech update: OpenMM (GPU) and Gromacs (CPU)

We’ve been pushing hard to improve the performance of OpenMM, especially in OpenCL as it’s now used in Folding@home.  We’ve got some great news hot off of the presses.  These are the benchmarks described at http://wiki.simtk.org/openmm/BenchmarkOpenMMDHFR.  They’re using the very latest OpenMM code, what will be in OpenMM 6.3.  They’re using CUDA 6.5 and running on Titan X.  All numbers are in ns/day.

Benchmark Calculation    CUDA   OpenCL
Implicit, 2 fs 471 366
Implicit, 5 fs 684 589
Explicit-RF, 2 fs 305 265
Explicit-RF, 5 fs 508 460
Explicit-PME, 2 fs 161 164
Explicit-PME, 5 fs 318 354

 

We’re especially pleased with those OpenCL PME numbers.  OpenMM Lead Developer Peter Eastman has put a lot of work into that for this release, and it now is actually faster than CUDA (For the Titan X).  Curiously, that is not the case on GTX 980.  It’s still slower than CUDA there, although it comes a lot closer than it used to.

This will be spun into an updated Folding@home core.  The upshot for GPU donors is that PPD for that new core should increase, due to the expanded capabilities of the new code.

It’s important to stress that SMP/CPU donors aren’t left out of new performance (and therefore PPD) updates either: FAH Lead Developer Joseph Coffland has been working hard on a new Gromacs core and that should also see performance benefits, as we roll out AVX support for FAH.

Introducing Shukla Group@Illinois

Shukla group (http://www.shuklagroup.org) at University of Illinois at Urbana-Champaign has just configured new Folding@home servers (ds01[a-d].scs.illionis.edu), which would help us carry out exciting computational experiments in collaboration with the vibrant F@H community.

Before joining Illinois in January 2015, I was a post-doctoral fellow in Pande Lab, working on conformational change mechanism of proteins related to a variety of diseases including cancer, neurodegenerative & cardiovascular disorders. Some of the key results obtained using Folding@home resources on conformational change mechanisms of G-Protein Coupled Receptors and Kinases are highlighted in previous blog posts.

The mission of my group is to combine theory, computation, and experiments to develop quantitative models of biological phenomena relevant for health, energy and environmental challenges. These grand challenges would not only require new scientific methodologies and insights but also development of platforms that enable broader participation of the community of informed citizens in the pursuit of the solutions. Folding@home is one such unique platform that enables engagement with volunteers and donors to help us solve challenging scientific problems. Our group is excited to be a part of the Folding@home team and we look forward to working with all of you on projects related to key challenges in human health. Specific project details will be posted soon on folding forum and F@H blog.

Shukla Group

Multi-core CPU jobs

We’ve been getting reports that FAH is low on CPU jobs.  We’re in the process of adding more multi-core jobs to existing projects.

Also, currently lead developer Joseph Coffland’s main project is to get a new Gromacs CPU core out to enable some new science on CPU cores (that’s currently only easily doable on GPU cores).  We expect a rough ETA for the first testing of that new core to be in a few weeks.

Issue with fah-web.stanford.edu

Likely due to a recent extremely heavy (and unusually rare) electrical storm, we’ve had some server issues last week.  I thought we’d gotten them all but we see now that there’s an issue with fah-web.stanford.edu, which serves up both the Folding@home stats and project descriptions.  While the stats are being accumulated on a separate server, fah-web is the web server that displays them to donors.  I’ve taken a look at it and there’s a more serious issue with a particular server than I can take care of myself. I’ve filed a ticket with the sysadmin team. Best guess ETA on this being fixed is Monday at noon (assuming this is something simple).

The stats accounting is still going on and this appears to be just an issue with the web server (fah-web), so we expect that this resolution should be simple enough once our sysadmin team gets to this on Monday .

UPDATE Monday May 18 at 10am pacific time:  We’ve got the machine back up and everything is looking good.  It appears that the problem was the server was under heavy load post storm, leading to a PERQ reset under load, a longstanding issue, which causes filesystems fo go read-only or offline, with the start of the problem was May 16 19:21:52.  We’re planning on buying new hardware to help here, especially since this hardware is on the older side now.

Issues with 171.67.108.60 and 171.64.65.124

We see the issues with 171.67.108.60 and 171.64.65.124 and are looking into it.

Fixes for recent FAH server outage

We recently ran into some problems with our assignment server (AS).  The AS is responsible for distributing the computational power of Folding@home by sending client’s to different work servers (WS), which in turn assign parts of the protein folding simulations to clients.  In the interest of transparency, here’s what happened.

Two issues compounded to cause some clients to not get work assignments for many hours.  The first problem is an issue we’ve run into before where the AS exceeds the number of open files allowed by the operating system.  When this happens it continues to run but fails to assign.  To address this problem, our lead developer (Joseph Coffland) has added code to the AS which will check the maximum allowed open files at startup and increase the limit to the highest possible value.  If the value is still too low it will print a warning to the log file.  This will help us ensure that our file limit settings are actually being respected.

The second issue was that failover to our second AS (assign2) didn’t work for some clients.  This was related to how we handle clients that cannot connect to port 8080 and WS that cannot receive connections on port 80.  The folding client will first attempt to connect to assign.stanford.edu on port 8080 if this fails it will try assign2.stanford.edu on port 80.  The AS assumes that connections on port 80 are from clients which don’t support connections to 8080 and only assigns them to WS which support port 80.

In a failover situation, this assumption is invalid.  The result is far fewer WS are available during a failover.  To solve this problem the AS was modified to prefer rather than require WS which support port 80 for connections on port 80.  This change can cause client/WS port mismatches but only when no better match was possible.  Yes, it’s a tangled web.

In addition to these changes, we have plans to implement an early warning system which should help to alert us to such situations sooner.  We already get SMS notifications if the AS goes down but we need more thorough reporting for situations where the AS is alive but not assigning. This new notification system will be put in place in the next few months.

Thank you for your patience and for your ongoing contributions to Folding@home!

Two server issues being worked on

We have two server issues being worked on.  There were some issues with the main AS and with the server serving SMP WUs (171.64.65.124).  We’ll post updates as we have them.

Add your computer's power to over 327,000 others that are helping us find cures to Alzheimer's, Huntington's, Parkinson's and many cancers ...

... in just 5 minutes.

Step 1.

Download protein folding simulation software called

Folding@home

.

Step 2.

Run the installation. The software will automatically start up and open a web browser with your control panel.

Step 3.

Follow the instructions to Start Folding.

Stanford University

will send your computer a folding problem to solve. When your first job is completed, your computer will swap the results for a new job.

Download the protein folding simulation software that fits your machine.

()


Installation guide
Or download Folding@home for your Android (4.4+) phone.