13th Community Wide Experiment on the
Critical Assessment of Techniques for Protein Structure Prediction

Prediction Center
Questions and Answers

I have determined a structure of a protein and want to suggest it as a prediction target for CASP13. How can I do that?
There are three easy ways to share your protein information with CASP. You can either directly submit the sequence of your protein and the related information through the web form; mark your PDB deposition as 'CASP target' (check box) within the PDB deposition system; or send us an email (casp AT predictioncenter.org).

I am not in structure determination business, but want to model a protein of my interest. Can I submit the sequence of this protein as a CASP target, so that the CASP community will model it for me?
No, CASP is not a structure modeling facility and we will not be able to do this for you. However, we have good news: many of the methods participating in CASP are available as public servers, and you can try them on your target protein.

Where can I find the list of CASP13 targets?
Targets suggested for prediction in CASP13 can be found at the Target List page . During the prediction season, targets are posted there daily and, additionally, automatically pushed to the registered prediction servers.

Are all targets in CASP13 intended for prediction by all groups?
Targets in the CASP13 experiment are divided into two tracks: those intended for prediction by all groups (i.e., both - human-expert and server groups), and those intended for fully automatic servers only. Human-expert groups predicting tertiary structure and contacts should concentrate on modeling all-group targets, while server groups should predict all targets. Human-expert groups predicting quaternary structure and complexes are requested to submit predictions on all released targets.

How targets are assigned to different tracks?
Assignment of a target to a particular track is made by the organizers and communicated to the predictors through the Target List page. Priority for inclusion in the all-group track is given to targets with low sequence identity to known proteins.

Are prediction time windows for all-group targets and server targets the same?
All-group targets are usually released with the 3-week prediction deadline, while server only targets usually have 72 hour prediction deadline. All targets are assigned two expiration dates (one - for server predictors and another - for human-expert groups).

When do you stop accepting submissions?
All predictions must be received and accepted before noon, 12pm PDT on the corresponding expiration date.

When targets are released?
Targets are released on business days only, around 9am PDT. They are released in batches of not more than 3 targets per day for servers and, usually, one target per day for expert groups. Data for model accuracy estimates category are released at noon, PDT. Sequence and other relevant information about the targets is posted at the Target List web page. Requests to the participating servers are sent shortly after the target release.

How many targets do you expect in CASP13?
We are planning to release 40-50 targets (depending on the availability) for evaluation in the all-group track and as many targets as we can get in the server-only track.

I am a predictor of tertiary structure. Should I predict all targets?
No, unless you participate as a server group. If you are a human-expert group, you should concentrate on all-group targets.

I am a predictor of quaternary structure and complexes (Assembly category). How can I know if the target is a monomer or a multimer? Should I predict all targets?
CASP organizers will do their best to provide predictors with the information on the oligomerization state of the target (column Oligo in the Target List). Predictors in the Assembly category are encouraged to send their models for all submitted targets (including server-only ones) before human expiration deadline.
Example.Targets T0861 and T0862 are two subunits of a complex. Target T0861 is released as a server-only target, while target T0862 - as a human-expert target. Human-expert groups that predict only tertiary structure should predict T0862 and ignore T0861, which will be assessed only in the server-only track of tertiary structure prediction. Predictors of complexes should predict both targets - T0861 and T0862. Predictions from human-expert groups for a server-only target T0861 should be submitted by the Human Expiration dealine.

I am a predictor of contacts. Should I predict all targets?
No, unless you participate as a server group. If you are a human-expert group, you should concentrate on all-group targets.

I am an estimator of model accuracy. Should I predict all targets?
Yes, both human-expert and server groups should predict accuracy of provided 3D server models for all targets. Deadlines for predictions in EMA (a.k.a. QA) category are the same for expert and server groups (2 days).

I participate in CASP as an expert group. Can I use server predictions submitted to CASP as starting points for my method(s)?
Yes, you can (obviously, with the proper acknowledgement of the original prediction). We will be releasing server predictions in 3 stages: (1) up to 20 selected predictions spanning the whole range of model accuracy will be released 2 days after the server TS deadline; (2) best 150 server predictions (according to the ranking from the DAVIS-EMAconsensus method) - 4 days after the TS deadline; (3) all server predictions - 6 days after the server TS deadline (see the QA format description at the format page).

How many predictions can I submit?
A team may submit up to five models in each of the tertiary structure prediction categories (high accuracy modeling, topology, data-assisted and refinement) and one model in the contact prediction category. In the estimation of model accuracy category, predictors may submit two different models: first model - after releasing the limited number of server regular TS models (prediction window will be open for two days after the tarball release), and second model - after releasing a set of tentatively best 150 server TS models.

Protein Structure Prediction Center
Sponsored by the US National Institute of General Medical Sciences (NIH/NIGMS)
Please address any questions or queries to:
© 2007-2018, University of California, Davis