CASP ROLL: the Community Wide Experiment on the
Critical Assessment of Techniques for Protein Structure Prediction on rolling basis
CASP ROLL Experiment

Registration Targets Predictions Results
Register for CASP ROLL
Target List
Target Submission
Submit a prediction
Model Viewer
Server Queries
Server Prediction Status
Server Tarballs
RESULTS (before CASP10)
Parsable data (before CASP10)

CASP ROLL in numbers

Description of the experiment

Background Goals Scope Timetable Participation Targets Format Assessment Publication Organizers


Currently, CASP assesses the ability of predictors to model protein structures in two difficulty categories: 1) template-based modeling (TBM), where easier to predict targets with detectable evolutionary similarity to experimental structures can often be modeled relatively well based on the related structures as templates, and 2) free modeling (FM), where difficult to predict targets displaying no detectable (or useful for modeling) similarity to available structures can be modeled with varying degrees of success using the so-called ab-initio methods. In recent CASPs, there has been a shortage of FM targets during the approximately 3-month long regular prediction season. This is especially true for true new fold targets and membrane proteins, which are becoming very rare in CASP (just 4 in CASP9), as fewer and fewer such structures are now being solved experimentally. The lack of difficult targets makes it problematic for the assessors to reliably quantify the state of the art in this area of prediction.


The rolling CASP experiment aims at more rigorous evaluation of template free prediction methods through assessment of a larger number of targets. We expect to achieve this by extending our current crystallographer /NMR spectroscopist /PDB target network, and by introducing a rolling mechanism to catch more difficult targets outside of the regular CASP prediction season. Unlike LiveBench and EVA, this experiment will be conducted in the blind-prediction spirit of CASP, i.e. all the predictions will be made on unknown yet structures.


  • Tertiary structure prediction (TS)
    • Regular CASP-style structure modeling
    • Contact guided structure modeling (for selected targets)
    • Chemical shifts guided modeling of NMR structures (for selected targets)
    • Structure modeling based on molecular replacement with ab initio models and crystallographic diffraction data (for selected targets)
  • Residue-residue contact prediction (RR)


Registration for the experiment starts in late November, 2011. New servers/ regular groups can join the experiment at any time after the experiment starts. Testing of server connectivity ("dry run" for server predictors) will be conducted shortly after the server registration. The first prediction targets will be released not earlier than December 1 (subject to availability). Thereafter, we will be posting new targets for prediction as they come in. The rolling experiment will run all year round. During the regular CASP10 prediction season (May-August, 2012) the targets that are estimated to be in the appropriate difficulty category will be posted to both experiments.


Participation is open to all. If you already have an account with the Prediction Center, you will be able to go directly to the CASP ROLL registration page. Please check, though, that your basic registration information is current. If it has changed - please update it through the My Personal Data link from the main Menu. If you are new to CASP and don't have an account with us, you will have to register with the Prediction Center first, and only then - for CASP ROLL. Separate registration forms for different types of registration are available through this website. Predictors with servers are requested to register as soon as the registration page is available to have enough time for testing the server connectivity.


For the experiment to succeed, it is essential that we obtain the help of the experimental community. We invite protein crystallographers and NMR spectroscopists to provide us information about sequences likely hard for structure modeling that they are working on and can keep on hold for at least 8 weeks after submitting them to CASP. Please suggest your protein as a CASP target by following this link .

Once selected for the experiment, targets are posted at the Rolling CASP Target List page and automatically pushed to the registered prediction servers. The registered participants will be periodically updated on the status of posted targets through email.

Targets will be released on business days only, around 9am PDT. All targets are assigned two expiration dates: one - for server predictors (3-day deadline) and another - for regular predictors (at least 3-week deadline). If possible, we will set more distant deadlines for regular group predictions so that physics-based groups get more time for running their methods. Please be aware, though, that in the unlikely situation when a target is publicly exposed before the regular deadline (and after the 3-week soft deadline), only the models received within the initial 3-week prediction window will be considered. Therefore we recommend submitting the best models you obtain within the 3-week prediction period first, and then resubmitting the final models before the final deadline. Predictions must be received and accepted before noon, 12pm PDT on the corresponding expiration date.


Predictions must be submitted to this web site through the prediction submission form or by email. Please comply with the instructions on CASP submission procedures and format. Server predictions will be made publicly available shortly after closing prediction window for a specific target.

Assessment of the results

Submitted models will be automatically evaluated at the Prediction Center using the standard CASP measures , additionally assessed by the free modeling CASP10 assessor and discussed at the regular CASP meeting in December 2012.


The rolling CASP results will be published in a scientific journal along with the results of the regular CASP10 experiment (see publications of previous experiments).

Organizing Committee

       John Moult, CASP president; CARB, University of Maryland, USA
       Krzysztof Fidelis, University of California, Davis, USA
       Andriy Kryshtafovych, University of California, Davis, USA
       Torsten Schwede, University of Basel, Switzerland
       Anna Tramontano, University of Rome, Italy
You can send us your methods abstracts now
Dear predictors, Yesterday three events were marked on the CASP11 calendar. 1. We received the last CASP11 prediction. Thanks for all your hard work over the period of three and a half months! ...
Last targets in contact-assisted category
Dear CASPers, 1. This week we are releasing the last 6 targets in contact-assisted categories Ts and Tc. Three of these targets (Tc826, Tc827 and Ts812) will be released today after closing the cor ...
End of CASP11 regular prediction season
Dear CASP predictors, Yesterday we closed the last three regular CASP11 targets. All in all, we released 100 targets including 55 all-group targets, and collected over 55,000 predictions. Thanks a ...
Protein Structure Prediction Center
Sponsored by the US National Institute of General Medical Sciences (NIH/NIGMS)
Please address any questions or queries to:
© 2007-2014, University of California, Davis