Posts by dadriano

1) Message boards : News : Another publication in Nature describing the first de novo designed proteins with anti-cancer activity (Message 93102) Posted 2 Apr 2020 by dadriano Post: Dear steffen_moller: What a massive oversight from me! This is Daniel-Adriano Silva, author in the paper. You are entirely right. I spend a long time in disbelief looking for the Rosetta@Home acknowledgment in the article (it should have been there), and it is missing. I am sorry, please accept my apology and let me elaborate: In figure #1 panel "e", the bottom panel shows the use of "Forward Folding" and ("Fast Forward Folding", also known as biased Forward Folding, see Marcos et. al., https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6219906/) to filter/evaluate the computational designs. We executed many of these simulations for thousands of computational designs, and many of these run on BIONC's Rosetta@Home. The legend of Fig.1 should (instead) read as: ... e) Top: Rosetta flexible backbone sequence design; Bottom: Forward Folding simulations for filtering computational-models with smooth folding funnels (see Methods)... The methods and/or acknowledgments do mention the use of multiple computational resources. However, they should have also included R@H, but it seems that we/I accidentally miss it. I apologize. I hope this clarifies that R@H did contribute to this development. Please receive my acknowledgments, and know that we value your important contribution that helps us bring de novo protein to solve real-world problems. Best, Daniel
2) Message boards : Rosetta@home Science : New publication of Baker's Lab (Message 87429) Posted 30 Sep 2017 by dadriano Post: We used rosetta@home at various stages of the mini-proteins design process in order to select/filter those designs that could computationally fold best (Forward Fold) into the designed structures. In the article, we acknowledged the computing time kindly donated by Rosetta@Home participants. Thanks again!
3) Message boards : Rosetta@home Science : HetTest? (Message 77850) Posted 23 Jan 2015 by dadriano Post: Hi Timo, First of all, thanks all for your contribution and sorry for the delayed answer. Here is a description of the project: One of the paradigms in protein design is the design of novel protein-protein interactions, this is the basis of molecular recognition in living organisms and an important milestones yet to be attained by human-designed proteins. Different kinds of protein-protein interactions happen all the time on living organisms. A common well-known example that highlights the importance of protein-protein recognition is the specific interaction that happens during human immune response where neutralizing antibodies recognize proteins from pathogens, leading to the inactivation and eventual clearing of the pathogen. The design of protein-protein interactions is not a trivial task and recently we have started a project that combines the design of novel proteins and the task of mastering the design protein-protein interactions, with the aim to create novel protein heterodimers (i.e. two different proteins interacting specifically,for a previous example of this kind of design but using as starting point natural-existing proteins see http://www.bakerlab.org/node/353/), by doing this we aim to gain the ability of constructing molecular switches that eventually permit the creation of more complex functions, for example enzymatic reactions that take place only when two specific-proteins are present, namely a molecular switch. In order to do this we start by using a database of novel protein structures which have been designed and experimentally corroborated by our group (mainly form: http://www.bakerlab.org/node/465/ and http://www.bakerlab.org/node/82/), then we use an exhaustive sampling algorithm to generate thousands of plausible combinations of pairs of these proteins that hold the potential of attain geometric complementarity (think about a key and a lock), however these proteins weren't originally designed to interact between them (and they don't), henceforth we use computational methods to design molecular complementarity between them, this is to create a novel protein-protein interface. By using this general method, we design thousands of potential pairs of unique proteins that have the potential to form a stable heterodimer. After multiple rounds of design and selection, we still have hundreds of combinations that are potential candidates for experimental testing. It is at this stage that we finally use our most powerful computational resource (Rosetta@Home) to validate and select the best designs among the pool. Here is where the contribution of donor’s computational time (like you, thanks!) becomes a crucial part of our project. It is at this last computational stage of the design process that we combine two computational-intensive strategies to verify and select our best designs. First is the verification of the protein's sequences to attain/fold-into the desired three-dimensional structure (namely, forward folding), and the second is to test the ability of a given pair of proteins (that succeed in the previous stage) to form in an specific heterodimer-interaction (namely, ab-initio docking), the former is the type of computation that you are enquiring in this post. A great explanation about the aim and inner-workings of the docking calculation has been previously provided by Chu Wang in this forum (see https://boinc.bakerlab.org/rosetta/forum_thread.php?id=2395#29146). Currently we are running 333 ab-initio docking simulations for our candidate heterodimers, all of them will be finished soon. Using the results of these calculations we will select for experimental validation only the best candidates that succeed in the aforementioned computational tests. As this project progress I'll keep posting regular updates in this forum.. Once more, thanks to all the donors for their invaluable contribution to R@H. Daniel-Adriano Silva