More good news!
Our paper titled "Protein structure determination using metagenome sequence data" was released today in the journal Science. We would like to thank all Rosetta@Home participants who provided the computing required for this work. In the paper, we describe using predicted co-evolving contacts from metagenomics sequence data and Rosetta to accurately predict the structures for 622 protein families that are not represented in the PDB. Among these structures, over 100 were new folds. Since experimental protein structure determination is costly and often difficult, this study highlights the ability to use computational methods with metagenomics data for reliably structure determination. With the rapidly growing size of genomics data, the future in mapping the structure space of protein families looks bright! Thank you Rosetta@Home participants!
Here is an interesting perspective written by Johannes Söding about the paper and it's significance, "Big-data approaches to protein structure prediction".
and related news articles:
Big data (and volunteers) help scientists solve hundreds of protein puzzles
Seeking Structure With Metagenome Sequences
Decoding the Origami That Drives All Life
Some good news!
We recently published an article in Nature titled "Accurate de novo design of hyperstable constrained peptides".
We would like to thank all Rosetta@Home participants for their help with this work.
In the paper, we present computational methods for designing small stapled peptides with exceptional stabilities.
These methods and designed peptides provide a platform for rational design of new peptide-based therapeutics.
Constrained (stapled) peptides combine the stability of conventional small-molecule drugs with the selectivity and potency of antibody therapeutics.
The ability to precisely design these peptides in custom shapes and sizes opens up possibilities for "on-demand" design of peptide-based therapeutics.
Other developments described in the paper:
We can now accurately design 18-47 amino acid peptides that incorporate multiple cross-links.
We can now design peptides that incorporate unnatural amino acids. Specifically, we designed peptides with a mix of natural L-amino amino acids and D-amino acids (mirror images of L-amino acids). D-amino acids tend to provide better protease resistance and lower immunogenicity; both of which are desired properties in a therapeutic peptide. Unnatural amino acids also let us sample much more diverse shapes and functions.
We can now design peptides that are cyclized via a peptide bond between their N- and C-terminus. Cyclic peptides provide increased resistance against exopeptidases as they have no free ends, and thus are ideal candidates for engineering peptide therapeutics.
We are now working to use these computational methods for designing peptides that target therapeutically relevant targets, such as, enzymes that impart antibiotic resistance in pathogenic bacteria.
Structure prediction runs on Rosetta@Home for these designed peptide models played a key role in selection of good designs that were experimentally synthesized and characterized.
Thank you all for your help in making this work possible! -- Gaurav B.
For more information:
Nature paper "Accurate de novo design of hyperstable constrained peptides".
Nature review "The coming of age of de novo protein design". In the last few weeks our project has experienced significant issues resulting in slower than usual work unit distribution, result processing, and credit granting. The cause of this was due to the increasingly large number of new hosts causing our database server to become very sluggish and eventually run out of disk space. Our short term solution was to reconfigure and optimize the project configuration and existing database server, purge old data quicker than usual, and temporarily stop resource intense database queries. This recovery mode will continue until the project stabilizes which may take a few days to a week.
An interim solution will be to temporarily upgrade our database server and thanks to our sys admins, we already have a machine ready to go that has plenty of disk space and double the memory. The upgrade will require a day of downtime which is planned to happen early this week.
The long term solution will be a complete system hardware upgrade to all our servers. The BOINC server software will also be upgraded. We are in the process of ordering these machines and hope to have them running within the next few months.
The project is somewhat stable now and clients should be getting work as usual. However, result processing and credit granting may still be slow and our status information/page may not be up to date as we are in recovery mode and our servers continue to catch up on things. Work history will also be shortened to temporarily save space. We expect things to be back to normal in a few days to a week.
Sorry for any inconvenience and thank you for your continued contributions!