That's a very good question! With respect to the ProteinInterfaceDesign simulations, on Rosetta @ Home our goal is to scan the vast conformational-sequence space in order to get good leads. Sampling for those is then intensified on our own machines where we can run high-memory trajectories for longer. So, in a sense we are implementing a primitive version of your suggestion. During energy minimization, by the way, we have exactly this delta energy criterion that you mention as a stopping condition.

What you are suggesting is actually being attempted now in a more sophisticated form than what I mentioned above by some of the people on the Rosetta team for structure prediction of very large proteins, where very many trajectories are run, clustered and the best few are then intensified. From what I've seen this is an extremely promising direction of research so let's keep our fingers crossed for it!

Thank you. Now it is clear from this acceleration comes. Due to the possibility of modeling is not all the protein completely, but only the most important (for the current study), part of it.

2 Sarel
And yet another question. I am not a scientist, so this may seem silly, but as a result of observing the process of calculations I had the idea.
It concerns the limit of 500 steps in modeling protein-protein including the last MDMX. It is clear that the shorter limit is introduced to speed up processing. And in most cases it seems to be enough - a graph of energy usually manages to "find minimum" for these 500 steps and more just jumping around him.

But periodically I see a model in which the energy almost continuously go down, but the simulation of the same breaks at the limit of 500 steps. (Obviously not a straight line, but with variations here and there, but the overall trend is absolutely clear - down) Though if given the additional steps would be found a configuration with a much lower energy. And since this model does not differ from the total mass, because its calculation was stopped before she could reach its minimum. Simply increasing the number of steps are not very effective way, because it increase the use of computing resources in times (or reduce the number of models with fixed resources), for 5% of models (about as much on my observations do not manage to reach the "saturation") is not effective. (Though perhaps this is the most valuable model, so that theoretically could be more valuable than the other 95%)

But what if instead of a fixed hard limit on the number of steps to use a dynamic limit is based on "delta energy" criteria? Ie compare the minimum energy found suppose the last 100 steps (for example, concrete value must be chosen experimentally), with previous values. If there is some improvement, then give the model additional steps, and so until the reduction of energy does not stop. The fixed limits (in particular number of steps) are also needed, but only as a acceptable framework - minimum (say cover those 500) and maximum (it should be big enough, but within reasonable limits. To finish on time "bad model", without waiting for the activation of watchdog).

Perhaps this idea has already been tried before? Then it would be interesting to know the results and why abandoned.

---

That's a very good question! With respect to the ProteinInterfaceDesign simulations, on Rosetta @ Home our goal is to scan the vast conformational-sequence space in order to get good leads. Sampling for those is then intensified on our own machines where we can run high-memory trajectories for longer.

Out of curiosity, how much memory do the more intense studies that are in-house require and is that a limiting factor to the lab's throughput? I assume all the computational limitations (in-house and R@H) are limiting in some way...

---

It varies, but I think that 750Mb would be a good guesstimate. Though that's not too bad, another thing that makes these trajectories a poor fit for Rosetta @ Home is that they might take several hours to produce output. Since we only do these runs for a few tens of the best of what we found on Rosetta @ Home, we can do it efficiently on our own machines.

That's a very good question! With respect to the ProteinInterfaceDesign simulations, on Rosetta @ Home our goal is to scan the vast conformational-sequence space in order to get good leads. Sampling for those is then intensified on our own machines where we can run high-memory trajectories for longer.

Out of curiosity, how much memory do the more intense studies that are in-house require and is that a limiting factor to the lab's throughput? I assume all the computational limitations (in-house and R@H) are limiting in some way...

---

I see lot of new p-p tasks with names started like "simIF_" and dated 18Jul2010
Its new protein target?

hi, thanks for asking. The simIF jobs are for the design of protein-based neutralizers for intimin, the bacterial Velcro system that allows Ecoli to stay in your intestines (see post https://boinc.bakerlab.org/rosetta/forum_thread.php?id=4477&nowrap=true#66323).

Hello,

My name is James Moody and I am a new graduate student in the Baker lab. I am working on protein-protein interface design and am excited to have the help of all of the participants of Rosetta@home!

Right now I am working on designing a protein to bind to a regulatory molecule called EED. EED works with other proteins in our cells to control which parts of our DNA will be used to control the cell and which parts will be silenced (turned off). EED is thought to work by bringing other proteins together and is part of a larger protein machine called the Polycomb Repressive Complex 2 (PRC2). PRC2 helps to ensure, for example, that we have the right number of arms and legs and that they are in the right place on our bodies (by controlling our Hox genes).

Scientists are working to better understand how EED works to control the activity of our DNA and its link to things like stem cells, development, and cancer. To do this, we need a way to interrupt the normal function of EED. This is especially difficult since there are currently no drugs that block EED function. An engineered protein would be able to prevent EED from sticking to one of its binding partners (histone tails) and could be turned on and off at will by the researcher. It is hoped that such a protein would allow researchers to carry out new experiments on EED that were impossible before.

Engineering this novel interaction with EED into another protein requires computationally screening through as many as possible of a billion possibilities, evaluating each protein for how tightly it sticks to EED, and then redesigning the new protein to stick even better. Such a task would be impossible without the help of Rosetta@home participants!

Thank you so much for lending yourselves to participate in this project. Things that I send to rosetta@home are tested first on our machines and then on a subset of Rosetta@home participants to ensure that they don't cause any problems for you. Please don't hesitate to ask if you have questions, feedback, or see errors.

If you would like to learn more about EED or related topics, the following webpages might be helpful!

http://en.wikipedia.org/wiki/Polycomb-group_proteins
http://en.wikipedia.org/wiki/EED
http://en.wikipedia.org/wiki/Histones
http://en.wikipedia.org/wiki/Nucleosome
http://en.wikipedia.org/wiki/Hox_genes

--James

Hi everyone,
I am working on another cool project that aims to generate a new molecular tool to dissect bacterial cell division. In order to propagate, cells need to divide. Separation of two daughter cells from a single cell is quite a complicated event that involves a fine-tuned and concerted molecular machinery. It requires figuring out where to actually draw the line, the assemble of several different proteins into one “divisosome” that will eventually perform the physical separation, and of course on top of all this, the cell needs to keep track of the DNA so that each cell gets its own copy. So it is imaginable that this is a fairly complicated happening.

Bacterial and mammalian cells are very similar in some parts of this event, but also very divergent in others. Of course, it is important to understand both. In mammals, uninhibited cell division results in cancer, but inhibited cell division can effect your body’s ability to regenerate. By knowing the difference between the mammalian and bacterial systems, one might be able to target bacterial cells specifically for new therapeutics without hurting mammalian cell division. The goal of this project is to produce molecular tools that help to analyze the complexity of the bacterial divisosome.

To picture cell division, you could imagine, one would take a string around a cell, pull and thereby separate the cell. Now you just have to imagine that the cells are actually making the string inside of themselves. They make a huge polymer (sort of like PVC). To get this molecular string, they polymerize several units of a protein that is similar between bacterial and mammalian cells, but one of the differences is the machinery it uses to coordinate this polymerization. Hence, one of my first targets of the cell division apparatus is a protein that is important for the localization of the polymer (aka string).

Thanks for your support! This wouldn’t be possible without your help.

/Eva

One of the most important ways in which molecular biologists learn about the functions of protein systems in the cell is by introducing mutations to the protein and seeing what the effects on cellular functions are. This method is extremely powerful but carries the risks that damage will be indiscriminate and the functional readout would be convoluted by many different effects. A more refined way of perturbing cellular functions would be to inhibit a specific interaction within the cell, keeping all other interactions, including of the target protein, intact. Such specific inhibition is more subtle and reduces unwanted effects, and indeed specific inhibitors of cellular pathways accelerate advances in our understanding of cell physiology.

The design of protein inhibitors promises to take this approach and make it widely available to systems that have been recalcitrant to the development of specific inhibitors to date. Since protein interactions have very large surface areas we can design inhibitors that would be exquisitely specific to the target protein as we've done for influenza hemagglutinin. In a posting above, I mentioned that we're designing proteins that would inhibit MDMX but not MDM2, despite very close homology between the two proteins.

A new protein target that we're interested in is called ERGIC-53. A complex between ERGIC-53 and another protein called MCFD2 has been shown to be an important cargo receptor that shuttles proteins between various compartments in mammalian cells. Even though the complex has been extensively studied major questions about its function remain unanswered, among which are how the complex interacts with its cargo proteins. A specific inhibitor of the interaction between ERGIC-53 and MCFD2 will provide a way to control the formation and dissociation of the cargo receptor in a time-controlled manner, decreasing unwanted effects, and hopefully increasing our understanding of this important system.

I should mention that mutations in either ERGIC-53 and MCFD2 have been identified in human populations and cause an inherited bleeding disorder. For more information on this link between the cargo receptor and disease, please see
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1895703/

Over the next few weeks I will submit workunits with different strategies for designing anti-ERGIC proteins. Thank you very much for your contributions to this project!

---

One of the most thrilling directions of research for protein-interface design is generating binding proteins for use as diagnostics in developing countries. The aim here is to develop materials for an early-detection kit, where the materials are cheap and robust enough to last for months or years without refrigeration. Diagnostics kits are typically comprised of antibodies, which are often expensive to mass produce and sometimes deteriorate over long periods of time. We believe that redesigning small stable proteins as binders would provide proteins of the necessary qualities to significantly improve the usefulness of diagnostics.

Our first target with a possible diagnostic application is the envelope protein of dengue virus. Dengue is a mosquito-borne virus that is widespread in the tropics and causes debilitating fever and death. Early-detection kits are extremely important because dengue-infected individuals should be protected from mosquito bites to cut off the spread of disease in their community.

Over the next few weeks I will submit design trajectories to Rosetta @ Home which will be labeled "Dengue binder design". I'm looking forward to seeing the binders that your computers will crunch this time!

To read more about dengue please see:
http://en.wikipedia.org/wiki/Dengue_fever

---

Hello,

As you may have read on David's blog we've had a lot of exciting developments on the influenza hemagglutinin front lately, including a crystal structure confirming one of the designs and biochemical data with respect to another that show that the protein inhibits the action of hemagglutinin. Hopefully this means that the proteins can serve as platforms for the development of therapeutics and diagnostics for many different flu types. More broadly, the thought that protein design could produce proteins that are as useful as antibodies is a sign of the field's maturity. I'm very pleased to mention also that both of these designs, and in fact 80% of the designs we generated in the hemagglutinin project came from your computers on Rosetta @ Home. This project has been so complex that without your amazing resources it would have been impossible to get working binders.

On a related issue, we're designing more putative binders of RhoA, so you will see more design jobs labeled "design of inhibitors of RhoA". I previously introduced RhoA in entry:
https://boinc.bakerlab.org/forum_thread.php?id=4477&nowrap=true#66178

Many thanks! Sarel.

---

I remember seeing an article saying that a certain important virus used a cluster of three of a certain protein for entering human cells, and normally had the rest of its envelope arranged to hide this cluster. Also, single instances of this protein appeared elsewhere on its envelope, perhaps to allow the body's natural defenses to bind where they wouldn't harm the virus.

Unfortunately, I've lost the link to where I found it, or I'd insert that link here. I'll keep trying to find it again, so you can consider whether it's practical to design a protein that binds to clusters of three of the virus's protein, but not to single instances.

Could be one of these, but the others look related as well:

http://www.rkm.com.au/VIRUS/Influenza/Swine-Flu.html

http://www.biomedcentral.com/1472-6807/9/62/figure/F5

http://vir.sgmjournals.org/cgi/content/full/88/12/3209

http://www.biomedcentral.com/1472-6807/9/62

The hemagglutinin protein, which you've done some work on already.

The influenza virus.

If the hemagglutinin protein is safe enough without the rest of the virus, you could also consider binding clusters of three of them at the right distance, in order to use the clusters as a vaccine.

In the last few days I see a lot of new WUs from the *ProteinInterfaceDesigh* series starting from 8, 10 and 16 December.
For example ApBp_eed2_eed2_* or rhoA8Dec2010_1lb1_2bf0_* or AEty_1_eed2_eed2_* etc
Some brief updates on the new search targets?

Right! We introduced the RhoA project in this entry:

https://boinc.bakerlab.org/forum_thread.php?id=4477&nowrap=true#66178

Hopefully more on eed soon...

Thanks for your interest, Sarel.

In the last few days I see a lot of new WUs from the *ProteinInterfaceDesigh* series starting from 8, 10 and 16 December.
For example ApBp_eed2_eed2_* or rhoA8Dec2010_1lb1_2bf0_* or AEty_1_eed2_eed2_* etc
Some brief updates on the new search targets?

---

For those you you asking about EED:

See for a description of this project. If you have other questions don't hesitate to ask!

Thanks again for all of your help!

--James

Hello,

My name is Shawn Yu, and I'm a new graduate student in the Baker lab. I have several really exciting projects in protein-protein interface design that I want to share with everyone involved with Rosetta@home. I'm designing proteins to bind to targets on the measles and Ebola viruses.

De novo protein-protein interface design has been an elusive problem in the field for quite some time. It is an extremely challenging endeavor that will require us to continually refine our algorithms and seek more computional power. Relatively recently--due in large part to the generosity and support of Rosetta@home users--we able to tackle these problems with reasonable expectation of success, as demonstrated with the influenza project, but we would like to continue to expand on this success.

Each viral target represents a different challenge that tests our knowledge of how protein interacts in different ways, so they are really interesting for their scientific merits alone. Furthermore, these methods are mutually reinforcing, so with more successful designs we'll have a better idea of what works and what doesn't work. In the long term, I hope the results from these projects can form the basis of future protein-based therapies, and I know many Rosetta@home users like myself contribute their time because they hope to help find cures to these diseases.

Over the next week, I will be discussing more details about my projects. In the meantime, thank you so much for volunteering your time and your computers to help to our projects. You are making a major contribution to our research!

-- Shawn

As I had alluded to earlier, I am working on a protein to bind to the hemagglutinin protein (MV-H) of measles virus. Measles is a significant cause of childhood morbidity and mortality worldwide. Measles virus enters our cells in a two-step process: an attachment protein allows it to bind to a host cell receptor, which allows a second protein to mediate fusion of the virus to the host cell. MV-H is the attachment protein responsible for the first step, and SLAM (or CD150) is the predominant host cell receptor to which MV-H binds.

Recently, researchers were able to solve crystal structures of MV-H in complex with SLAM. As you might imagine, a high-resolution structure of the target (MV-H) is crucial to our design efforts, so we were quite excited about this development!

In several recent design efforts in the Baker lab, such as with influenza, we identified several "hotspot" residues on the target surface; although these hotspots made up only a small portion of the surface, they contributed a very large proportion to the binding interaction. We designed disembodied amino acid residues ("stubs") to interact specifically in these regions and then fit the rest of the designed protein accordingly around these stubs.

Likewise, in our measles project, we have identified hotspots on MV-H where we would like our protein to bind. One interesting challenge unique to measles is that our hotspot approach focuses on charged amino acids such as lysines to bind to oppositely charged regions of MV-H. Most of the previous success we have had involved binding to hydrophobic regions, but we'd like to push the envelope on what we can do!



I think the final designed protein binder will include some subset of the amino acids depicted above. It would bind MV-H where it would have bound to SLAM, thereby preventing its entry into our cells. However, in order to find a possibly successful design, we have to search through an astronomically large number of configurations, and the assistance of Rosetta@home users is of critical importance to this effort. So once again, thank you very much for your contributions to our research! Please let me know if you have any questions or comments!

What will your tasks be called so we know when they come to our systems?

Hi!
My computer is crunching something like this at the moment:

MVH_2s_K_2q6m_ProteinInterfaceDesign_20110505_26665_45_0

And I think this must be what we're talking about ;)
Am I right?

Hi!
My computer is crunching something like this at the moment:

MVH_2s_K_2q6m_ProteinInterfaceDesign_20110505_26665_45_0

And I think this must be what we're talking about ;)
Am I right?

Sounds right to me!

---