Looking for more journal entries?
We've archived the older entries to help you have a faster download of the current entries.
2006 entries here.
2007 entries here.
2008 entries here.
2009 entries here.
Questions about Dr. Baker's journal?
We have a series of threads over time where you can participate with your questions and comments. Look for a thread in the Science forum with a sticky entitled "DISCUSSION of Rosetta@home Journal". At present, discussion 4 is the active discussion thread.
Rosetta Informational Moderator: Mod.Zilla
We got some good news today. A manuscript that many of you contributed to through Rosetta@home was just accepted for publication in Science magazine, perhaps the most widely read scientific journal. The paper shows that accurate structures can be calculated using Rosetta for proteins up to 200 amino acids long if even a small amount of experimental data (from NMR experiments) is available to guide the search. This is an exciting advance because it could make it very much faster and easier to experimentally determine protein structures. Thanks everybody for your contributions to this work, and to our ongoing research efforts!
Sarel has collected many very promising potential flu virus inhibitors from your rosetta@home calculations over the last ten days, and will be selecting a number of them for experimental testing--see his postings in the "design of protein-protein interactions" thread.
We are entering a very busy and science packed time for Rosetta@home. As described in the "design of protein-protein interfaces" thread, we are now designing proteins to bind to and block several different targets, including the flu virus. At the same time, we are gearing up for CASP9 which will start in May by testing out both our new structure prediction methodology and the improved energy function which underlies it. The new methodology is quite CPU intensive, and we are hoping for as much user participating as possible once CASP starts; whatever you can spare now as well would be great so we can go the last 9 yards on structure prediction methods development before CASP and at the same time proceed as rapidly as possible on the protein-protein interaction designs. thanks! David
This survey request is from David Anderson, the creator of BOINC, and Oded Nov from NYU
Dear Rosetta@home volunteer:
We are conducting a survey of Rosetta@home volunteers in order to
better understand why people participate in volunteer computing and
contribute computer resources.
We would be extremely grateful if you could help us by filling out a
questionnaire. If you are not interested, ignore the rest of this
The survey is at http://boinc.berkeley.edu/survey/ It should take no
more than 10-15 minutes. Your responses will be used for research
purposes and to improve BOINC.
We will be happy to share our findings with you, and they will be made
available once we complete the data collection and analysis.
With many thanks -
Dr. David P. Anderson
University of California, Berkeley
email: davea at ssl.berkeley.edu
Prof. Oded Nov
Polytechnic Institute of New York University
email: onov at poly.edu
Our paper on solving structures of proteins of up to 200 amino acids using very limited experimental data is in the Feb 19 issue of Science magazine (pg 1014) which is on some news stands now. this wouldn't have been possible without Rosetta@home--thanks again everybody!
While the results are still preliminary, it appears that Rosetta@home has produced an extremely exciting result! As I described a few posts ago, many of you through rosetta@home contributed to the design of proteins predicted to bind very tightly to the influenza flu virus. We have now completed the first round of testing of the designed proteins, and one of them in the experiments conducted thus far clearly binds very tightly to the virus. Our data also indicate that the binding is at a site critical to the virus invasion of our cells, and so the protein may be able to neutralize the virus. I will keep you posted over the next couple of months as the picture becomes clearer--but for now--thank you all for making this possible!!
I was asked on the discussion thread about the timescale for learning more about the influenza binding protein I described in my previous post. I'm reposting my answer here:
We are doing a series of tests and control experiments in my lab in the next 2-3 weeks to rule out various possible artifacts. If, as we expect, the design passes with flying colors, we will send it to Scripps research institute where the ability of the design to neutralize the virus in cell based tests and the extent to which the design neutralizes different strains of virus will be measured. I would expect we would know the results of this in several months. We will also work to solve the crystal structure of the design bound to the virus to confirm the design binding mode. This hopefully will not take more than a few months as well.
I will keep all of you posted here about the results from these experiments. I am very optimistic, but one should be cautious about getting to excited too early about results like these--there are very many places where things can go wrong just with the biochemistry, and after this there are very many steps to actually make a protein into a drug--this is why there are so few new drugs for curing diseases being discovered.
For those of you who would like to try your hand at improving designed binders to the influenza virus, we are now posting virus inhibitor design challenges on foldit.
Experiments this past week have made us even more confident that the designed influenza binder is working as in the design model. we used "directed evolution" methods to identify amino acid changes that make the rosetta@home designed protein bind even more tightly to the virus. we found mutations at two positions: first, at an alanine residue in the design, the evolution process found a valine, and inspection of the design model showed some extra space around the alanine that would be filled by the slightly larger valine. the second amino acid change involved a charged aspartate residue in the design that in retrospect was too close to the virus protein--it was changed to a non charged residue which is less energetically costly to bury upon binding.
we are now combining these two substitutions, and expect that the combination should bind still more tightly to the virus than any protein we have tested so far. we should know later this week--I'll keep you posted!
We got some good news today in an announcement by Vice President Biden:
We were funded to work with three other research groups to develop a completely new pathway for using solar energy to transform CO2 into the large molecules that the world has grown to depend on (fuels, etc)--this if successful could greatly reduce dependence on fossil fuels and contribute to removing CO2 from the atmosphere.
While the large majority of rosetta@home calculations will remain focused on biomedical problems, expect to see from time to time work units relating to design of enzymes for CO2 capture and conversion.
We got some more good news today: a manuscript we submitted to Science magazine on rosetta based de novo design of a new enzyme which catalyzes the formation of two carbon-carbon bonds between two small molecules was accepted for publication. The work described in the manuscript is a real step forward in designing enzyme catalysts for reactions not catalyzed by naturally occurring enzymes, and could provide new routes to drug molecules which can be hard to synthesize using traditional methods.
CASP9 is now in full swing and we need your help! We are being overwhelmed with targets and need as much CPU power as possible!
I just got this from the organizers:
Subject: CASP update - May 7
First week of CASP9 prediction season is over. We have released 14 targets. The vast majority of them were easy TBM targets. Next week you will find some harder targets in the human prediction category.
As of today, we have 125 groups (predominantly servers at this early stage) contributing models to the Prediction Center. You can always find the latest CASP statistics at http://predictioncenter.org/casp9/numbers.cgi .
If you are interested in following the prediction season as it happens, the above web site is a good source of information.
We are absolutely delighted by the recent increase in the total throughput of rosetta@home, which could not come at a more critical time! we are having to make very difficult choices between CASP9 structure prediction calculations and the next generation of pathogen inhibiting proteins building on our success with the flu virus inhibitor, and the new contributions of computer power many of you are making are helping immensely. Thank you very much!
We have now confirmed the tight binding of our designed Spanish Flu inhibitor to the flu virus using multiple different methods (it is always good to be totally certain with exciting results like these!). For those of you with some chemistry background, the binding constant is about 20nM.
With collaborators at Scripps research institute we are now trying to determine the structure of the designed complex between the inhibitor and the virus by x-ray crystallography (to see whether binding is as in the design model). With the tight binding confirmed, we are now starting to investigate whether the designed protein prevents the virus from infecting cells.
A manuscript describing the results on FoldIt, which many of you contributed to, was just accepted for publication in Nature. The idea for FoldIt came from rosetta@home participants who posted on the message boards about wanting to be able to guide the course of the folding trajectory. Please keep letting us know your thoughts and suggestions!
Rosetta@home has now been directly responsible or closely associated with two papers in Science (one on enzyme design, one on new approaches for structure determination) and two papers in Nature (this one on Foldit, and one last year on endonuclease design for gene therapy) in the last 9 months. This kind of impact at the forefront of scientific research is I think a first for volunteeer computing, and perhaps the strongest indication to date of the power and value of volunteer computing for pushing forward the boundaries of scientific understanding.
Thank you all for your invaluable contributions to our collective efforts!
The most recent issue of Science magazine has our paper on the use of Rosetta to design a new carbon-carbon bond forming enzyme, along with a commentary. This paper has attracted a lot of attention in the press. Thank you all for your contributions to this work and to our ability to move forward with designing enzymes and other proteins that will hopefully be of use to society in not too long.
There have been exciting developments in our work to develop general methods for designing proteins that can bind to and block the activity of any desired target protein. There are now three targets for which we have designed and experimentally validated binders: a widely used "model" protein called lysozyme, a protein involved in biosynthesis in the bacteria that causes tuberculosis, and a key protein on the surface of the H1N1 flu virus. In the flu case, our collaborators have just solved the structure of our designed protein bound to the virus protein and it is amazingly close to our computational design model.
Now that the methods seem to be working pretty well, we are thinking more about applications. One of these is to make cheaper and more robust diagnostics kits. We are now collaborating with groups interested in developing low cost diagnostics for the flu virus (and other pathogens). our designed proteins are very easy to make in large quantities, and our collaborators are going to test how well they work in place of more expensive and less stable antibody molecules in diagnostic kits.
In the past I've described a brand new approach using Rosetta to design vaccines for which there are not effective current vaccination treatments. HIV, for example, has turned out to be fiendishly effective at evading the immune system, and as you probably know, despite much work there is no really good vaccine. In collaboration with other groups, Rosetta has been used to design small proteins that present "Achilles heel" regions of the virus to the immune system to stimulate the production of antibodies which recognize these regions. The first papers on this have now been published and, while the designed proteins have not yet elicited strongly neutralizing responses, there is considerable excitement over this new approach. You can read about it at
The Gates foundation has just awarded $1,000,000 to a collaborative project to develop specific enzymes that cut within HIV DNA in cells into which the virus has integrated:
One of the many nasty things about HIV is that it can reside for a long time in a latent state where it can't be detected by the immune system. If we can generate enzymes that cut up the virus when it is hidden inside a genome, its hiding place will be destroyed. Keith Jerome's lab at the UW is developing methods for delivering enzymes to possibly HIV infected cells, and a graduate student working between our groups is using Rosetta to design endonucleases which cut within the HIV DNA sequence. I'll keep you posted on the progress of this exciting project!
First, I would like to thank everybody for bearing with us while we recovered from a critical server hardware failure. Over the next month or two we will be installing more powerful and more robust hardware so hopefully this will not happen again.
Second, I'd like to tell you briefly about another exciting success with Rosetta. When structural biologists work to solve protein structures by putting protein crystals into x-ray beams and recording the diffraction pattern, they only have half of the necessary information. The other half (the "phase" information) can be quite difficult to obtain. In the past six months, we've collected about 15 cases where protein crystallographers were stuck and could not solve the structure. Using Rosetta, we built models for these proteins of sufficient quality to allow the inference of the missing information and subsequently the solution of these structures. This opens the door to a much easier way of solving challenging protein structures, and there are lots of scientists excited about using the new method. The new method is described in a manuscript which will likely appear soon in Nature magazine.
Again, thank you for sticking with rosetta@home during our recent server problems -- there is a lot of exciting scientific research that is only possible because of your contributions!
The new hardware is now being installed, and we are very happy to put the server failures behind us-hopefully we will have no outages this severe again for a long time!
The manuscript I described in my last post on solving crystal structures using Rosetta has now been formally accepted for publication in Nature; the paper will be in an issue on newstands in a month or two.
In a previous post I described the design of small proteins which bind to and block the function of the key surface protein on the influenza virus, called the haemagluttinin (I can never spell that right!). We are very excited about the possibility of making more proteins that bind to the various strains of the virus that could serve as anti flu drugs (this would only be for very acute infections as you probably wouldn't want to take a dose of these too many times) and are actively working on this. Meanwhile, a manuscript describing the design of the first proteins and how they block the haemaglutinin from the Spanish Flu influenza virus has just been accepted as a full research article in Science magazine. We are excited (and nervous) because we've never been this close to making an actual drug before (as I've explained before, most of what we do is directed more at basic understanding than actual drug development). Still, of course, it is a long road (clinical trials, etc if we get that far) to get something to the point it can be used as a drug. I'll keep you posted as we move along with this.
The paper on the Rosetta method which allows determination of the structures of a large class of proteins using limited crystallography data has now been published in Nature magazine. Thanks to all of you for making this work possible!
Graduate student Shawn Yu is now posting on current Rosetta@home efforts to design inhibitors for viruses that cause disease in the "Design of Protein Interactions" thread on the Science message boards. Take a look if you are interested; he is happy to answer questions in the thread as well.
This week's issue of Science magazine features an article on the use of Rosetta@Home to design novel proteins which bind tightly to the Spanish Flu (H1N1) Influenza Virus. The paper shows that the experimentally determined atomic structure of the complex between one of the designed proteins and the virus is precisely as in the computer model. The designed proteins block the function of the flu surface protein in biochemical tests, and we are guardedly optimistic that the designs will block flu infection. This is an important milestone for computational protein design (and for distributed computing)--the first atomic level accuracy design of a high affinity protein-protein interface, and the designed proteins are exciting leads for new flu therapeutics. In the next few months, we will be using Rosetta@Home to design proteins that bind tightly and hopefully block other pathogens which cause disease. Thanks to all Rosetta@home users for their invaluable contributions to this research!!
(if you want to learn more, the Science web site has a podcast discussing the work:
A recent issue of Nature describes an exciting approach we are taking with collaborators to fight Malaria. The title of the paper is "A synthetic homing endonuclease-based gene drive system in the human malaria mosquito" and the PDF is available at my lab web site. The idea is to use enzymes which cut within critical genes in mosquitos to greatly reduce the number of malaria parasite infected mosquitos. There are still many issues that must be overcome for this strategy to be used against malaria in the real world, but this paper is an important first proof of concept of the strategy.
This week's issue of Nature magazine has an exciting article (http://www.nature.com/nature/journal/vaop/ncurrent/full/nature10154.html) describing work we are doing with collaborators using Rosetta to design a new class of inhibitors of amyloid fibril formation. Amyloid fibrils have been implicated in Alzheimer's and many other diseases. The designed peptides are not suitable for use as actual therapeutics in their present form, but hopefully will help lead the way to effective drugs.
Today's issue of Nature Structural Biology reports the determination of the structure of a protein by FoldIt players. This is exciting because it is perhaps the first example of a long standing scientific problem solved by non-scientists. You might read about this in your newspaper; here is a report that does a good job in explaining how FoldIt came out of Rosetta@home:
A recent issue of Nature describes an exciting result from Rosetta@home in collaboration with the NMR spectroscopy laboratory of Lewis Kay in Toronto. Like almost all machines, proteins in order to carry out their functions have to move (change their conformation somewhat) but it has been extremely difficult to determine what these conformational changes are. Lewis Kay's group has developed new methods for getting experimental information on the higher energy very shortlived conformations proteins visit while carrying out their functions. This data is not sufficient to determine the structure of these "excited state" conformations using conventional methods. However, as the paper shows, we can use these experimental data to guide Rosetta and Rosetta@home structure calculations, and produce models of these states. We went one step further than this in the paper by using Rosetta design calculations to stabilize the excited state, and subsequent experiments confirmed the validity of the model. This combination of experimental NMR data, Rosetta structure calculations, and Rosetta design should be very powerful in understanding how proteins carry out their functions.
Today's issue of Science magazine describes an exciting new approach to HIV vaccine design using Rosetta. In contrast with other viruses such as polio and influenza, inactivated HIV or HIV proteins have not worked as vaccines, and hence as you know there is currently no effective HIV vaccine. Our approach to vaccine design is to take the bits of the HIV surface protein that people make antibodies to, and using Rosetta graft them onto small stable scaffolds that can be made in large quantities and potentially could be useful as vaccines. We've shown earlier that this can be done straightforwardly with Rosetta if the bits of the HIV protein are contiguous along the sequence, but it is much harder if the antibody recognizes multiple bits close in three dimensions but far in sequence. In this paper we show how such "discontinuous" epitipes can be transferred from HIV gp120 to a simple scaffold protein. More work will be required to determine whether this or other vaccine candidates designed using this approach will be effective as HIV vaccines-let us all hope so!!
In response to requests from many of you, we will be posting descriptions of the many scientific problems currently being tackled with Rosetta@Home on the Science message boards in the next couple of weeks--stay tuned! I also want to describe a new research direction we are now embarking on aimed at future cancer therapies. There are a small set of proteins which are frequently found at much higher levels than normal on the surface of cancer cells. We are starting to design small proteins which bind tightly to these tumor cell markers. If we are successful, we have collaborators who will be testing these proteins for their ability to target cancer cell killing agents to the tumors.
Last year we described in Science magazine the design of a new enzyme which catalyzes a chemical reaction called the Diels Alder reaction involving the formation of two carbon-carbon bonds. This reaction is interesting because no natural enzymes are known to catalyze the reaction. However, it wasn't a very good enzyme, and we asked FoldIt players to try to improve it. As described in Nature Biotechnology this month, remarkably FoldIt players were able to make the designed enzyme 20 times faster by inserting a completely new loop which helps the enzyme bind the chemicals it links together. The combination of Rosetta@Home and FoldIt is turning out to be powerful indeed for solving challenging problems in biomedicine!
In the last two months we believe we have made quite a breakthrough in structure prediction, and are excited to test the new method in CASP10. We need your help though--we are now testing many aspects of the new approach and are seriously limited by available CPU cycles. There are now so many flu inhibitor design and structure prediction jobs queued up on Rosetta@Home that there is an eight day wait before they are getting sent out to you. This would be a great time to temporarily increase Rosetta@Home's share on your computers and/or recruit new users--we need all the help we can get! thanks! David
I've described in the past our work using Rosetta and Rosetta@Home to create new enzyme catalysts. In Nature Chemical Biology last month we describe the design of an enzyme which destroys organophosphate
nerve agents and pesticides. These compounds kill by blocking key enzymes, and our designed enzyme eliminates this toxicity. This illustrates how Rosetta@Home enzyme design work can help to solve current problems, including man-made problems.
I have just been told the very good news that Rosetta@home will be the first project of the BOINC pentathlon, and would like to thank all of the participating teams. I also just learned from the discussion thread that Rosetta@home will be the project of the month for BOINC synergy-this is more excellent news!!
Your increased contributions to rosetta@home could not come at a better time! We've been testing our improved structure prediction methodology in a recently started challenge called CAMEO. For most of the targets, the Rosetta@home models are extremely good, but for a minority of targets the predictions are not good at all. We've now tracked down the source of these failures and it is what we are calling "workunit starvation"; in the limited amount of time the Rosetta server has to produce models (2-3 days) in these cases very few models were made-this happens because many targets are being run on the server so that only a fraction of your cpu power is focused on any one target. while we are working to fix this internally, by far the best solution is to have more total CPU throughput so each target gets more models.
You can follow how we are doing at http://www.cameo3d.org/. You will see that Robetta is one of the few servers whose name is not kept secret-this is because Rosetta is a public project. Our server receives targets from CAMEO and soon CASP, sends the required calculations out to your computers through Rosetta@home, and then processes the returned results and submits the lowest energy models.
We are excited that the workunit starvation problem may go away through your increased efforts for Rosetta@home. Thanks!!!
A big THANK YOU to all of you who have scaled up your contributions to Rosetta@Home-this is a record level of computing power for us and is super well timed. THANKS!!!
Many common materials such as silk and wool are made out of regular repeating arrays of proteins, and symmetric protein arrays make up the coats of viruses and many other assemblies inside cells. The ability to robustly design self assembling materials made out of proteins would have huge numbers of applications-the naturally occurring assemblies are useful, but imagine if we could make custom materials for 21st century problems. In this weeks Science magazine, we describe the use of Rosetta to design self assembling protein nano structures with very high accuracy. We are now attempting to create many different types of symmetric materials, and you will be seeing more symmetric calculations on your rosetta@home screen server. Thank you for making possible this completely new approach to nanotechnology!
A recent paper in Nature Biotechnology describes how we have combined computational protein design with the high throughput DNA sequencing methods developed for sequencing the human genome to generate potent influenza virus inhibitors. These designed proteins block infection by the flu virus in cell culture experiments, and they are now going through the (quite lengthy) process of being developed as possible anti-flu drugs.
We are now testing the latest batch of novel designed proteins that you have helped us create in our brand new Molecular Engineering laboratory at the UW. You can see pictures of the space where the rosetta@home computed designs are being experimentally tested in a recent newspaper article:
The native structures are slowly being released for CASP10 targets; all of them will be available by the end of November. In the meantime, you can look at the results of a much larger scale test of prediction methods called CAMEO. CAMEO takes newly solved protein structure before they are published, and sends the amino acid sequences out to structure prediction servers. This happens every week, so it is great to assess new methods as they are being developed. You can look at the results, as well as get more information about CAMEO, at
The best number to compare is the "Average accuracy (all targets)" as some servers only model the easy ones. The good thing about CAMEO is that there are many more test cases than CASP, and also that results are released each week so we can see what is working well and what needs to be improved. You will see that ROBETTA, which is now using some of your computing resources, is doing pretty well recently; before this we had problems with some targets not getting enough work units before the server deadline.
I have exciting news. We and the University of Washington are starting up a new Institute for Protein Design to design new proteins to address current challenges in medicine, energy, and other areas. You can learn more about the institute at http://depts.washington.edu/ipd/. Rosetta@home is and will continue to be a critical part of our efforts. For every new potential protein therapeutic we design, we use Rosetta@home to test whether it will actually fold into the desired structure. And we need help! We have quite a backlog of exciting new designed proteins to test on Rosetta@home because we are designing proteins for quite a range of problems-new anti flu proteins, anti-cancer proteins, and new materials--and it takes 3000-5000 work units to test each one. This Rosetta@home testing is becoming the slow step in the whole design process, often taking over 10 days to complete. So please tell your friends and relations to join us!
A generous donor has provide funds which we want to use to invite 5-10 Rosetta@home participants to visit the Institute and see what we are trying to accomplish first hand. More on this in my next post.
With your help, we have made an exciting breakthrough in protein design that is reported in a research article titled "Principles for designing ideal protein structures" in the journal Nature today. You can read about it at
In this paper, we describe general principles for creating new proteins from scratch. The new Institute for Protein Design is using these principles to design new proteins to treat disease.
Rosetta@home was absolutely critical to this work as described in the news article; Figure 3 in the paper shows how all of your contributions were used to test designed sequences to see if they folded up to the right structure. Most of the work units we are sending out on Rosetta@home these days are for exactly these kind of tests on the new proteins we are designing--this is absolutely critical to the research and to the development of new therapeutic and other functions. Thank you again for all of your contributions!
We would like to acknowledge the Rosetta@Home participants who found the lowest energy structures for Nobu and Rie's designed proteins:
Fold-I : Aalelan (United States)
Fold-II : Jef (United States)
Fold-III : georgebg (Bulgaria)
Fold-IV: medvjet009(Czech Republic)
Fold-V : _2e_ Russia.
See Nobu's message board thread on "Principles for designing ideal protein structures" (http://boinc.bakerlab.org/rosetta/forum_thread.php?id=6113) for more information including a figure illustrating the critical role played by Rosetta@home in this work. And check out the latest on slashdot:
This and other publications from the lab are available at depts.washington.edu/bakerpg
Thanks again to everybody!!
The CASP10 meeting just finished and all the results are on line so you can see what all your contributions made possible! Ray has posted a great summary of the results in the CASP10 thread on the science message boards. Overall, Rosetta@home was top or near top in most categories. I hate to embarrass David Kim who created RosettaHome and has kept it going, but his contact guided predictions were the highlight of CASP10 as they were vastly better than those of any other group. You can see this at
David's predictions are the black lines, those of other groups are in orange, better models stay below the rest of the pack.
Thanks to all of you who contributed to CASP10!!
We have learned to design a new class of proteins which could be useful both as drugs and in sensors. As I've explained before, most drugs are small molecules with fewer than 50 atoms. There are many drugs, such as blood thinning agents, which are very dangerous if given in too large doses. We have succeeded in designing proteins which bind to specific small molecules very tightly. These proteins could be used as antidotes in case of overdose with the target small molecule-for example we've designed a protein which could be useful to treat toxicity due to overdose of the drug digoxin used to treat heart disease. With collaborators, we are working to use these designed proteins to detect levels of the target small molecules in the blood or in the environment.
We have discovered how to make several new classes of protein structures! Rosetta@home has been absolutely critical in this work: when we design a sequence to fold into a new structure, the last thing we do before ordering a synthetic gene so we can make the protein in the laboratory is to send it out to you to predict the structure-if it folds to the structure we designed, we go ahead with it, but if you find that the lowest energy state is a different structure we go back to the drawing board. Our success rate in making brand new structures is far higher than I or anybody else ever expected, and the reason the success rate is so high is that your calculations provide a very stringent test of whether the designed sequence will actually fold the way it is supposed to. In the next few weeks I and other scientists here will describe to you the new classes of proteins we are making, and the many applications they will be useful for. Thank you for your absolutely essential contributions to this newly emerging scientific field!