Does CAMEO build models for users?

No. CAMEO is not a meta-server and will not answer user requests. Its sole purpose is to continuously evaluate the accuracy of prediction servers based on known experimental structures released by the PDB.

What is the schedule of submission and evaluation in CAMEO?

CAMEO is hinged on the PDB pre-release, which currently is scheduled to be published for Saturday morning (CET). The pre-release is then processed to exclude too easy structure prediction targets and ~20 selected sequences are sent to the CAMEO 3D participants. For the QE category protein structure models are required and thus there is a delay of ~24h. All the structure predictions (ie atomic coordinates) of the public CAMEO 3D servers that have been received by CAMEO until then, are submitted to QE participants.
After the following Wednesday 2am (CET) predictions are no longer accepted for any CAMEO category. The evaluations are subsequently started on Wednesday morning (CET) after the PDB released the experimental structures.

How does the submission work?

Upon submission, you will receive an HTTP POST or GET request with the target information including target sequence, structure file for QE, project title and email address. CAMEO is flexible and allows you to specify which parameters to bind the data to. The registration form lets you tell us precisely how to submit the targets.

For certain categories like QE and 3D Structures and Complexes, validation systems are also available to test your server beforehand.

How can I change submission information?

Please contact us directly by email ( with the new information. If you have multiple server registered, please make sure to include the server ID.

How do I return my predictions?

You must return your predictions by email to the email address that was provided during the submission. The models should be included as attachment to the email.

Can I submit several models for a target protein in CAMEO?

Yes, you can include several models in an email as attachments. CAMEO will evaluate up to 5 models, in attachment order. However only the first model will be considered in summary tables. If you server can be tuned with different parameters (e.g. for highest coverage or for highest accuracy), you will need to register two servers with different parameters. Please use the following on-line registration form [REGISTER], or contact us directly by email ( if you have any questions.

The models I sent were wrong, can I submit new ones?

Yes, you can send us a new email. CAMEO will only look at predictions in the last email received (within the time window) per server for each target sequence.

Is all CAMEO data public?

Authors of a prediction server can decide if their evaluation data is public or only accessible privately. Services initially registered as "development" (e.g. during the development phase) can later be changed to "public" (e.g. for publication purposes).

Can I suggest a new score / evaluation method?

Yes, of course. CAMEO is a community service, and we don’t expect that our initial set of scores will be perfect. In fact, we know it is not. If you would like to see a different evaluation score applied, please let us know!

Which servers will be evaluated?

CAMEO will only evaluate servers with the consent of the author. We invite all server developers to register their services to CAMEO. Please use the following on-line registration form, or contact us directly by email if your service does not fit within these categories.

See our complete list of registered servers.

Which prediction methods are covered by CAMEO?

In the current release, CAMEO addresses the evaluation of services for Protein Structure prediction and 3D Protein Structure Quality assessment. CAMEO also featured a category for Ligand Binding Site Residue prediction and one for Residue-Residue Contact prediction which have been deprecated. In addition we have a beta category for the prediction of macromolecular complexes.

How does CAMEO relate to CASP?

In CASP, expert groups aim to predict structures of 50-100 target proteins for which the coordinates will not be published during the prediction window (3 weeks). The results are assessed blindly by independent assessors. In CAMEO, automated prediction servers are continuously benchmarked using thousands of newly released structures from the PDB on a weekly basis. The assessment is fully automated.

How do I register my modeling server with CAMEO?

In the modeling category each server will submit a single model, so if your server normally delivers one model for highest coverage and one for best quality you would need to register two servers each submitting one model.
Similar, for the quality estimation (QE) and residue-residue contact prediction (CP) categories only one prediction is accepted as well.

Please use the on-line registration form, or contact us directly by email if your service does not fit within these categories.