• Home

Welcome to Cryo-EM Modeling Challenge 2010

(and PSB 2011 Workshop)

July 1 - December 1, 2010

Challenge results are now publicly accessible

No account is required to look at the results of the challenge, but if you wish to make comments on records, you will need to create one. The results can be accessed in a variety of ways:

The results of the challenge will be presented in a special issue of Biopolymers. The deadline for submissions is November 1, 2011. Submissions by those who did not participate in the original challenge WILL be permitted. For details, please contact sludtke@bcm.edu.

Note : This is called a 'challenge' rather than a competition for a reason. The goal is to open communication channels between the modeling community and the cryo-EM community, each of which has something to offer to the other. If the challenge is a success, it will provide a basis for more controlled 'competitions' in the future. The results of the challenge will be permanently posted as a resource, so when people in the cryo-EM field are considering what techniques they can use to interpret one of their shiny new maps, they will have a broad range of examples to consider. Similarly, modelers will know what methods have already been tried, which should help spur development of new techniques.

Modeling as a tool for interpretation of cryo-EM reconstructions

Cryo-EM single particle analysis is a method for determining structures of large molecules and macromolecular assemblies at resolutions ranging from 3.5 - 30 A. Interpreting the density maps produced by this technique represents an ongoing challenge, for which molecular modeling techniques offer some unique solutions.

Over the last five years, cryo-EM single particle analysis has begun producing structures at resolutions better than 5 A, with subnanometer resolutions becoming common. At resolutions between 5 and 9 A it becomes possible to move beyond simple rigid-body docking and alter atomistic models to reposition helices and sheets, to better fit the cryo-EM based density maps. At 3-5 A resolution de-novo C-alpha traces and in some cases full atomistic models can be constructed directly from the cyro-EM density without invoking x-ray crystallography.

We call this project a challenge rather than a competition because, unlike CASP, there is no hidden answer to be revealed. In this project, we will provide cryo-EM densities for a selected set of structures at different resolutions, and challenge those in the modeling community to apply their tools to extract as much information as they can from each. At the end, the results will be evaluated by comparing the results of different groups, and validating against any other existing knowledge about each target. We hope this will yield new insights into these published structures, and at the very least, it will establish the capabilities of current modeling methods, and give the cryo-EM community some guidance as to how to proceed with maps in various resolution ranges. For modelers it provides a new area in which to apply/develop their techniques, and demonstrating tools' capabilities may lead to new opportunities for collaboration.

The idea for the challenge came as a result of a round-table discussion at the 2010 Modeling of Cryo-EM Maps workshop hosted at the NCMI in January

Challenge Details:

In addition to the challenge itself, Wah Chiu, Helen Berman, Steven Ludtke, Gerard Kleywegt are co-organized a workshop at the Pacific Symposium on Biocomputing in Hawaii on Jan 3-7, 2011. This workshop was devoted to discussing the results of the challenge.

Rules:

  1. We will provide a set of structures and other information through this server. You must set up an account before you will be able to download the relevant data or submit your results.

    • For each target, we provide the density map resulting from the cryo-EM reconstruction. At present there is no procedure for 'improving' these maps through use of modeling techniques, meaning, from a modeling perspective, they are unbiased interpretations of the raw data. These maps represent the actual targets of the challenge.

    • In most cases, PDB models are also provided. These are the models that the author submitted as interpretations of their density maps, and have been produced using a wide range of different techniques. While you certainly may make use of the author's PDB models or other information we provide, you are also completely free to make use of ANY other resources you wish.

    • If you wish to communicate ideas or make comments visible to other participants on a specific target or folder, notice the 'comments' box at the bottom of each page. Anyone is free to make a comment, and all comments will be visible to all participants.

  2. There will be six categories of entries to accommodate the various types of analysis suitable for the covered resolution range:

    • Volume Analysis - techniques for annotating density maps without producing a full atomistic model
      • Protein Segmentation - Segment a density map into useful or meaningful subregions
      • Secondary Structure Annotation - Annotate specific secondary structures within a density map, preferably in conjunction with a quality score
      • Backbone Tracing - For cases where the resolution is sufficient to trace the protein backbone, but not to place sidechains
    • Structural Modeling - techniques for producing or altering atomistic models based on density maps
      • Rigid Body Modeling - Position one or more existing atomistic models into a density map
      • Flexible Modeling - Beginning with a crystallographic or homology model, alter the structure to best fit the density map
      • ab-initio Modeling - Building a full atomistic model for a density map using methods similar to crystallography

  3. Solutions should be submitted to this server as soon as they are completed, but will initially be visible only to you and the challenge moderators.

  4. After the closing date of the challenge (December 1) permissions will be relaxed so those who submitted valid entries will be able to see other submitted solutions for the same model...

  5. After the PSB meeting, all submitted results will be made publicly available

  6. Invitations to present at the Workshop will be issued by late August, so it is to your advantage to submit preliminary results as soon as possible. Until the closing date, you will be free to revise or retract your submission whenever you like. If you are interested in presenting, but don't have significant preliminary results by mid August, email Steve Ludtke (sludtke@bcm.edu) with a brief description of what you are trying to do.

  7. The challenge moderators will aggregate the results and present a summary, and any overall conclusions that can be drawn from the results of this challenge.

Folders: