Phyre Searching TOpology by Rapid Matching


What is PhyreStorm for?
PhyreStorm is used to rapidly and comprehensively compare a protein structure to the entire PDB. The structure used to search could be an experimental structure, or a predicted model such as one produced by Phyre2.

Protein structures can be remarkably similar even in the absence of any detectable sequence similarity. These similarities can reflect either remote evolutionary connections (remote homology) or convergent evolution (analogy). Such similarities in structure can illuminate potential functions for proteins and expose otherwise undetectable relationships of biological interest.

Why did you build PhyreStorm?
The motivation for creating PhyreStorm was a dissatisfaction with existing tools/websites designed for this purpose. Many such sites are either a) too slow to allow speculative exploration, b) limited in their coverage so that only highly similar structures are found, c) not continuously up-to-date with the ever-growing PDB or d) difficult to navigate and understand.

With PhyreStorm we hope to address all of these shortcomings.

How does PhyreStorm work?
PhyreStorm uses the de facto standard structural alignment program TMalign to align your structure to all chains in the current Protein Data Bank. This would ordinarily be computationally unfeasible, taking approximately 1 CPU day.

We overcome this problem through a combination of hierarchical search and intelligent use of spare processing time on our compute farm. The PDB provide an all versus all structural comparison of all the chains/domains in their database. We use this information as input to a hierarchical clustering program (available here). This generates approximately 16,000 clusters of structures. Each cluster is representated by the member closest to the centre of the cluster (the medoid).

PhyreStorm initially scans these medoids and only scans members of a cluster if the match to the medoid is above a user-defined threshold.

These thousands of jobs are handled by the freely available ActiveMQ system. More details of the implementation are available in the paper.

How do I use PhyreStorm?
Please see the Help pages for how to submit, navigate and interpret results.

Can I get my own local copy of PhyreStorm?
Yes. Installation instructions and data can be found here

Who made PhyreStorm?
PhyreStorm was primarily built by Stefans Mezulis with input from Lawrence Kelley in the Structural Bioinformatics Group at Imperial College. The project was made possible thanks to ActiveMQ and TM-align

PhyreStorm: A web server for fast structural searches against the PDB.
Mezulis SM, Sternberg MJE & Kelley LA
Journal of Molecular Biology,
Volume 428, Issue 4, 22 February 2016, Pages 702–708