Any news about the seven papers you were writing?

---

While we are all most interested in developing cures for diseases, aging and other problems

Thank you for the update, Dr. Baker.

It's a small thing, but I'm very happy that you mentioned "aging" specifically here. It too often isn't considered a disease, while I think it totally is (accumulation of damage over time, and finding ways to fix that damage could have a tremendous positive impact).

So thanks :)

---

Thank you for the reply!
I'm really happy to see that publications comes out thanks to crunchers like me.
Keep us informed, in particular for the publication on Nature! :)

---

Do you plan to mention the users who found the best predictions in your blog again and maybe in the papers and articles as you did some time ago?

May I also ask for a short WU description by someone of the Baker team? It would be interesting to know if WU xyz is crunching a flu target you posted about or if it is an application benchmark run, ...

Thanks :)

---

Do you plan to mention the users who found the best predictions in your blog again and maybe in the papers and articles as you did some time ago?

May I also ask for a short WU description by someone of the Baker team? It would be interesting to know if WU xyz is crunching a flu target you posted about or if it is an application benchmark run, ...

Thanks :)

an update of the "top preditctions" page would be nice too :)

---

Thank you for the suggestions! We will work on updating and reporting this week. Everybody here is so focused on research we do need reminders occasionally when we have fallen behind on reporting back to you all the exciting things that are happening!

---

Thank you for the suggestions! We will work on updating and reporting this week. Everybody here is so focused on research we do need reminders occasionally when we have fallen behind on reporting back to you all the exciting things that are happening!

Hi David,

Please see the middle of this thread where many of us have asked for over a year for better communication from RAH...

https://boinc.bakerlab.org/rosetta/forum_thread.php?id=4061&nowrap=true#54016

I would like to hear a reply from you by Tuesday, the 28th or I am pulling all my machines off this project.

Yes, I am that frustrated with the lack of respsonse and professionalism from this project.

-Eric

---

Thank you for the pointer to this thread. I just posted an (apologetic) response, and now I will give an update on the work we are doing in my journal page.

---

Thank you for the pointer to this thread. I just posted an (apologetic) response, and now I will give an update on the work we are doing in my journal page.

The abstract of David Kim's paper that you posted raises the following question for me. He mentions that smaller proteins have been accurately modeled by Rosetta, but it appears there is a size limit, above which we don't yet have the necessary computing power.

Possibly taking a quote out of context, but approximately how large are the "Larger and more complex proteins" that are spoken of.

As an extension to this, and assuming there is a size limit, does that same size limit come into play for the reverse problem, trying to create a protein chain for a custom shape.

Keeping this in mind, for the proposed flu vaccine, and possibly for your proposed HIV vaccine, do you have any estimates of how many residues these will have, and where that count (again assuming it exists) is in relationship to the above limit.

All of which is a long and convoluted way of asking, "How close are we to having the necessary accuracy in Rosetta, and compute power from all our systems to be able to create the flu vaccine that you spoke of recently?"

---

Thank you for the pointer to this thread. I just posted an (apologetic) response, and now I will give an update on the work we are doing in my journal page.

The abstract of David Kim's paper that you posted raises the following question for me. He mentions that smaller proteins have been accurately modeled by Rosetta, but it appears there is a size limit, above which we don't yet have the necessary computing power.

Possibly taking a quote out of context, but approximately how large are the "Larger and more complex proteins" that are spoken of.

As an extension to this, and assuming there is a size limit, does that same size limit come into play for the reverse problem, trying to create a protein chain for a custom shape.

Keeping this in mind, for the proposed flu vaccine, and possibly for your proposed HIV vaccine, do you have any estimates of how many residues these will have, and where that count (again assuming it exists) is in relationship to the above limit.

All of which is a long and convoluted way of asking, "How close are we to having the necessary accuracy in Rosetta, and compute power from all our systems to be able to create the flu vaccine that you spoke of recently?"

Additional comment on the above - if there is not sufficient computing power available with the current group of crunchers, how much more is needed to bring the project up to the level to tackle the larger and more complex proteins?

One more question to add to those above:

Is the prediction of the shape of larger proteins limited computationally by the total aggregate FLOPS of the project as a whole, or is it also limited by the power of individual computers right now (for example, maybe the work unit of a larger protein would take 4 gigs of RAM and it would take much longer just to run one decoy)?

---

Good questions!

First, our ability to compute structures of larger proteins is limited by the total aggregate FLOPS of the project, rather than the power of individual machines.

Second, Rosetta@home's upper limit for predicting protein structures accurately using only the amino acid sequence of the protein is about 150 amino acids. But, as I've discussed previously in my journal, this limit can be overcome if there is even very limited experimental data available on the structure (for example, suppose you are searching for the lowest elevation point on Earth, and you are told that it is not in North America, or that it is in the Middle East; this isn't a huge amount of information but it still reduces the amount of searching a lot). Some of the work we are doing with Rosetta@home now is testing how high we can go in size if we have limited experimental data available, or if there is information from the structures of related proteins we can use.

Third, this limit for prediction does not carry over to the design problem, for example the design of swine flu blockers. we have successfully designed active enzymes with sizes over 300 amino acids, for example. this is because when we design a new function or activity, we can start with one of the many structures that have already been determined experimentally, and build the new function onto this structure.

---

Very interesting, thanks!

---

Thank you for the pointer to this thread. I just posted an (apologetic) response, and now I will give an update on the work we are doing in my journal page.

The abstract of David Kim's paper that you posted raises the following question for me. He mentions that smaller proteins have been accurately modeled by Rosetta, but it appears there is a size limit, above which we don't yet have the necessary computing power.

Would producing a version of minirosetta that runs in 64-bit node, and can therefore use more memory on machines that have enough, make it easier to handle the larger proteins? A separate pool of workunits might be needed for these, so the results for these usually don't need to be compared with running the same workunits with versions of minirosetta in 32-bit mode.

Also, producing a GPU version would give more computing power, but usually not more memory for each processor core.

Hi David...can you please give us an update? See this post:


https://boinc.bakerlab.org/rosetta/forum_thread.php?id=4061&nowrap=true#62854



-Eric

---

Hello Dr. Baker, Hello Team,

my sincere congratulations for your upcoming publication in Nature - one of the most prestigious scientific magazines! It is good to participate in such a productive project.

I hope you will find time to deliver the all-around description of recent work, achievements and challenges you promised 2.5 weeks ago.

Good luck with all your efforts,
a.m.Poland

Congrats on the Nature publication! Glad to be helping (a tiny bit) with gene therapy by crunching for Rosetta@home.

---

For those that have not heard the term before, I believe that the "phase problem" Dr. Baker referred to in his last post is where specific atoms within the protein are in a sort of unstable or orbiting state where they aren't in a fixed position. This throws off the existing methods of prediction because they get a larger then expected fuzzy area and aren't certain what to make of it.

Perhaps Dr. Baker or others can expound on the basics of the problem being solved when they've dug through the influx of EMails their results have generated.

But, in laymen's terms, what he is saying is that the existing prediction methods of crystallizing and x-raying (or NMR) a protein failed to accurately predict the shape. I presume the Rosetta fold and dock protocol was using the scan data as one of it's inputs. And so the program was able to use the scan data and produce a more complete and accurate structure than the existing methods were able to do with the same data.

For some very interesting background, reading in layperson's language, here is a link to an article in Wried from 2001, on the existing x-ray methods of structure prediction. Understand that there is still several weeks of manual work required after the X-rays are taken to interpret the fuzzy blobs that result and reveal the structure. Also, the article was written with just a single protein in mind, but Dr. Baker is discussion a case where there are two of them bound together.

The ultimate goal of Rosetta is to predict the structure using only the chemical sequence of the protein as input. The chemical sequence is much more easily obtained then the X-ray crystallography. But using Rosetta to accurately interpret the X-ray data, and reliably eliminate the weeks of manual work done after the X-ray data is gathered will be a great stride forward.

Understanding how proteins bind together is how treatments will be made for disease. How a protein behaves in the body or it's environment has everything to do with it's shape. And if you bind another protein to it, you have changed it's shape and it will no longer be able to function in the way it did before. So if you bind a strand on to a flu virus, it will no longer behave like the flu virus. You still must study and understand what behavior it does have after binding with your agent, confirming that it doesn't become something somehow worse. But generally, if you can bind to it, you've disabled it and the person no longer has the flu, or HIV, or other disease.

This is why you don't often see statements from the Project Team about "we're working on cancer (or HIV, or the flu) today". Because most of what they are trying to do is master the fundamentals science required to combat all of these things. If you study just one protein to death (pardon the pun!) you will likely devise a program that does not accurately solve other proteins. This is why you study proteins from E. Coli, or snake venom and a suite of others that sound unrelated to HIV and cancer. And you do it each time you make revisions to the program. Others have already solved these protein structures, but they still don't know how to solve it using just the chemical sequence and a computer model. And they still don't have a complete and reliable understanding of how to use that knowledge to make a cure for these diseases.

There are probably some small inaccuracies, or better ways to make my statements here. I am not a protein scientist, just a very enthusiastic message board volunteer. I hope I've shed some light on why Rosetta@home is important. And perhaps part of why it is hard to explain sometimes. It's just not as simple as "we're working on cancer this week". I hope you find this information useful and others with more technical knowledge will use this as a basis for further clarification.

---

Rosetta Moderator: Mod.Sense

Dr. Baker,

I read with some interest your post from Aug 15 that talks about molecules binding to proteins. Am I correct in thinking this is exactly the task that Keith Davis was doing with the Find-A-Drug project?

The reason i ask is that I was working on F.A.D. prior to coming here to Rosetta, and I'm still interested in finding out whatever I can about the science of that project.

I also remember that at the end of the distributed computing portion of F.A.D. Keith Davis explained that we'd pretty much exhausted all combinations of drug / protein interaction possible. He was then planning to take our results and do further research in an effort to identify viable drugs. Presumably this is required because of what you mention in your post where a theoretical bind from a computer simulation doesn't always lead to a good bind in practice.

---

Yes, it is the same problem. Computer based drug screening, which we and F.A.D have carried out, currently can increase the odds of finding a good drug candidate, but cannot unambiguously identify tight binders. We are currently working to increase the reliability of the basic methodology.

Dr. Baker,

I read with some interest your post from Aug 15 that talks about molecules binding to proteins. Am I correct in thinking this is exactly the task that Keith Davis was doing with the Find-A-Drug project?

The reason i ask is that I was working on F.A.D. prior to coming here to Rosetta, and I'm still interested in finding out whatever I can about the science of that project.

I also remember that at the end of the distributed computing portion of F.A.D. Keith Davis explained that we'd pretty much exhausted all combinations of drug / protein interaction possible. He was then planning to take our results and do further research in an effort to identify viable drugs. Presumably this is required because of what you mention in your post where a theoretical bind from a computer simulation doesn't always lead to a good bind in practice.

---