CAPRI: Critical Assessment of PRediction of Interactions

We are happy to announce CAPRI Round 57: the 6th joint CASP-CAPRI Assembly Prediction challenge, run as part of the CASP16 prediction season. As previously, this Round will accept submissions by servers and predictors and will offer a scoring challenge whenever possible.

The CASP16 season will open on May 1 and run through August 2024. Members of both the CASP and CAPRI communities will be invited to model various types of protein complexes and multi-component assemblies. A description of the CASP16 experiment can be found at: https://predictioncenter.org/casp16/index.cgi

CAPRI participants wishing to take part in Round 57 are invited to register for the entire round in advance.

Registration to CAPRI Round 57 is now open. As in previous joint experiments we require CAPRI predictors to register also with the CASP16 season. Groups only participating as Scorers only need to register with CAPRI.

Registration to the CASP16 season is now also open. Only groups registered with both experiments will be part of the joint CASP-CAPRI challenge. As before (i.e. CASP15-CAPRI challenge) models submitted through the CAPRI system will be seamlessly forwarded to the CASP prediction center. Please submit through the CAPRI system, as otherwise they will not contribute to the Scoring Experiment.

Important: Please register with CASP before registering with CAPRI, as the CAPRI registration will require you to enter your CASP ID.

Round 57 targets will be announced as soon as they become available from CASP. As in previous CASP-CAPRI Rounds, the number or type of targets offered in this Round is currently unknown, as targets are being submitted piecemeal during the entire duration of the CASP16 session. Decision on targets designated for CAPRI docking and scoring predictions will be made by CAPRI MC members Alexandre Bonvin and Marc Lensink, and communicated to CASP. These targets will be earmarked as such in the CASP target list.

CASP might run some additional experiments in which the stochiometry of a complex is not given. As CAPRI we decided to only include as CAPRI targets those for which the stochiometry is provided. Interested CAPRI participants can of course submit those “unknown stochiometry” targets directly to CASP.

Target information: Information on the different targets and CAPRI submission deadlines (for servers, predictors and scorer) will be provided by CAPRI as usual. Target information can also be obtained from CASP: https://predictioncenter.org/casp16/targetlist.cgi

Model submission: To ensure that submitted models strictly adhere to the standard format and residue numbering, a template with the required submission format will be provided for each target by the CAPRI system. Predictor groups are strongly encouraged to use this template. Predictors and servers are expected to submit (up to) 100 models for each target and rank their 5 best models in order of decreasing model quality.  All submitted models will be checked for consistency with the formats of both CASP and CAPRI by the CAPRI online system, and non-compliant submissions will be refused.  The top 5 models of the Predictor submission will be forwarded to CASP.  For further details and tips, please visit the CASP-CAPRI page. Only models submitted through the CAPRI system will be considered for the Scoring Experiment.

Scoring submission:Scorer groups and server are expected to submit (up to) 10 models for each target and ranked in order of decreasing model quality.  All submitted models will be checked for consistency with the formats of both CASP and CAPRI by the CAPRI online system, and non-compliant submissions will be refused. Only models submitted through the CAPRI system will participate in the Scoring Experiment.

Model evaluation: All the submitted models (to CAPRI or CASP) of homo- and hetero-complexes on CAPRI-selected targets will be evaluated by the CAPRI evaluation team as usual. Participant ranking will be done based on the 5 top ranked models as in previous CASP-CAPRI Rounds. The top 5 models submitted by CAPRI participants will also be evaluated by the CASP independent evaluation team. The CASP and CAPRI assessor teams will meet to discuss how the evaluation results will be published in the CASP15 Special Issue.

For scoring, the top 10 models submitted by CAPRI participants will be evaluated by the CAPRI evaluation team.

We are happy to announce Round 56, with one target.

Due note that the schedule for this target is a bit tighter than usual; registration starts *today*, prediction March 11th.

We are happy to announce Round 55. The Round contains four targets, three of them with antibodies.

Registration will open Monday Nov 13th, with prediction starting two weeks later.

We are happy to announce CAPRI Round 54: the 5th joint CASP-CAPRI Assembly Prediction challenge, run as part of the CASP15 prediction season. This Round will accept submission by servers and predictors and will offer a scoring challenge whenever possible.

The CASP15 season will open May 02 and run through August 2022.Members of both the CASP and CAPRI communities will be invited to model various types of protein complexes and multi-component assemblies. Description of the CASP15 experiment can be found at: http://predictioncenter.org/casp15/index.cgi

CAPRI participants wishing to take part in Round 54 are invited to register for the entire round in advance.  Registration to CAPRI Round 54 is now open. As in previous joint experiments we require CAPRI participants to register also with the CASP15 season. Registration to the CASP15 season is now open. Only groups registered with both experiments will be part of the joint CASP-CAPRI challenge. As before (i.e. CASP14-CAPRI challenge) models submitted through the CAPRI system will be seamlessly forwarded to the CASP prediction center. Please register with CASP before registering with CAPRI, as the CAPRI registration will require you to enter your CASP ID.

Round 54 targets will be announced as soon as they become available from CASP.  As in previous CASP-CAPRI Rounds, the number or type of targets offered in this Round is currently unknown, as targets are being submitted piecemeal during the entire duration of the CASP15 session. Decision on targets designated for CAPRI docking and scoring predictions will be made by CAPRI MC members (Shoshana Wodak and Marc Lensink) and communicated to CASP. These targets will be earmarked as such in the CASP target list. More of the CASP assembly targets will likely be earmarked for CAPRI than in previous Rounds, since we expect the availability of DL methods to help with modeling targets for which few or no template are available.

Target information: Information on the different targets and CAPRI submission deadlines (for servers, predictors and scorer) will be provided by CAPRI as usual. Target information can also be obtained from CASP: https://predictioncenter.org/casp15/targetlist.cgi

Model submission: To ensure that submitted models strictly adhere to the standard format and residue numbering, a template with the required submission format will be provided for each target by the CAPRI system. Predictor, uploader and scorer groups are strongly encouraged to use this template.  Predictors and servers are expected to submit 100 models for each target and rank their 10 best models in order of decreasing model quality.  All submitted models will be checked for consistency with the formats of both CASP and CAPRI by the CAPRI online system, and non-compliant submissions will be refused.  The top 5 models of the Predictor submission will be forwarded to CASP. For further details and tips, visit the CASP-CAPRI page on this website.  Only models submitted through the CAPRI system will be considered for the scoring experiment.

New: Model accuracy estimates:Predictors and scorer are encouraged to submit ‘self-reported’ accuracy estimates for protein complexes and inter-subunit interfaces now also considered by CASP. These estimates would be analogous to the predicted local-distance difference test (p-LDDT) provided by AlphaFold. These estimates are to be provided at the atomic level (in priority for backbone atoms) in the B-factor column, and span values of 0-100.

Model evaluation: All the submitted models (to CAPRI or CASP) of homo- and hetero-complexes on CAPRI-selected targets, will be evaluated by the CAPRI evaluation team as usual.  Participant ranking will be done based on the 5 top ranked models as in previous CASP-CAPRI Rounds. The top 5 models submitted by CAPRI participants will also be evaluated by the CASP independent evaluation team. The CASP and CAPRI assessor teams will meet to discuss how the evaluation results will be published in the CASP15 Special Issue.

Expecting many of you to exploit AlphaFold2 and other Deep Learning-based methods, we look forward to exciting results from this Round.

We are happy to announce CAPRI Round 53, with Targets 187 and 188. Registration of the Round has started! Predictors and scorers are requested to register for the entire Round.

The system of both target is the same - important - bacterial protein-DNA complex, whose high-resolution structure is solved for the first time by single-particle cryo-EM. T187 is the more challenging problem, where predictors are provided only with sequence information for the protein and the DNA construct used for the cryo-EM study. T188 represents an easier modeling exercise, where predictors are provided with the three-dimensional structure of the bound DNA moiety from the complex. Scoring challenges are offered for both targets independently.

This system is a challenging prediction problem that the community should be able to tackle using available tools.

We are announcing CAPRI Round 52, offering target T186, representing a large protein complex determined at high resolution by cryo-EM, by Prof. Rouslan Efremov, at the VIB-VUB center for structural biology, Brussels Belgium

This Round which was initially planned as to start May 14, 2021, was not correctly announced on the CAPRI website. We have therefore shifted the timing so that the Round is now starting on June 1, 2021.

The target is part of a stable sub-complex (peripheral arm) of the E.coli respiratory complex:  NADH:Ubiquinone oxidoreductase, which comprised 6 polypeptides. One of the subunits has a ~100 residue loop, bound to a large pocket of the complex. The bound loop adopts a well-defined conformation that contains little secondary structure. The main challenge of this target will be to correctly model the bound conformation of the ~100 residue loop in the E.coli complex.

Registration to Round 51 opened Sep 15, 2020 and will end Dec 29, 2020.A total of 4 targets are being offered in this Round. Three are complexes of SARS-COV-2 proteins with human host proteins (prioritized from the interaction proteomics study of Gordon et al. (2020) [1]) and the 4^th target is a multi-component complex of SARS-COV-2 proteins and RNA. Description of the Round can be found at: https://www.ebi.ac.uk/pdbe/complex-pred/capri/covid-capri/

Round 51 will be conducted in the spirit of open science. Models submitted by all groups/teams will be made immediately available online after the submission deadline in full transparency. Since for many of the considered targets, experimental structures may not be available for some time to evaluate predicted models, participating groups will work together to evaluate the quality of submitted models, compare and rank them on the basis of agreed-upon criteria and derive consensus predictions whenever appropriate. The protocol for this evaluation will be worked out after the Round is completed.  We remind everyone that the plan is to report the results of this collective effort in a common publication coauthored by members of all participating groups/teams.

1. A SARS-CoV-2 protein interaction map reveals targets for drug repurposing,

Gordon DE et al., Nature. 2020 Jul;583(7816):459-468. doi: 10.1038/s41586-020-2286-9. Epub 2020 Apr 30.

The CASP14 season will open May 18 and run until approximately July 30, 2020. The first assembly target will be offered during the week of May 25.  Members of both the CASP and CAPRI communities will be invited to model the interfaces of protein hetero-complexes, homo-multimers and domain-domain interactions in CASP14 assembly targets. Description of the CASP14 experiment can be found at: http://predictioncenter.org/casp14/index.cgi

CAPRI participants wishing to take part in Round 50 are invited to register for the entire round in advance.  Registration is now open. As in previous joint experiments we require CAPRI participants to register also with the CASP14 season. Only groups registered with both experiments will be part of the joint CASP-CAPRI experiment (for further details please consult the CASP-CAPRI page using the link on the top menu).

 The CAPRI community is mobilizing its resources and expertise to model the 3D structures and interaction interfaces of SARS-COV-2 - human protein complexes and larger assemblies (homo- or hetero-complexes) starting from experimentally determined structures (by x-ray diffraction or cryo-EM), or from predicted structure, of the individual components.

An important focus of our initiative is to prioritize relevant COVID-19 complexes that will be offered as target for community-wide CAPRI prediction Rounds. These prediction Rounds will be run as a collaborative undertaking where data, analyses and results will be closely integrated and widely shared, hopefully moving COVID-19 science fast forward.

Whenever possible our initiative will leverage COVID-19 efforts of CASP (CASPcommons) ) and the SIB/Modeller (SIB/MOdeller) on predicting the 3D structure of individual proteins, by using the high-quality models they produce to predict the association modes of these proteins.

 The CAPRI community remains eager to continue to improve and test their modeling tools by taking on increasingly challenging prediction problems.  In the spirit of open science we hence call upon structural biologists working on COVID-19 related protein complexes to offer these complexes as targets for CAPRI prediction Rounds prior to publication (for details how this is done see contributing targets.) 

If you are interested in joining our initiative or exploring a collaboration, please contact us at: capri.docking@gmail.com

Target 163 is an interesting, and likely challenging prediction problem. The target is a heterocomplex with A2B2 stoichiometry (confirmed experimentally) of two small helical protein components of the synaptonemal complex.

We are happy to announce CAPRI Round 48, with Targets T161 and T162. The systems of both targets is a toxin-antitoxin complex.

Predictors and scorers are requested to register for the entire Round.

Additional notes:

We are announcing the first round of 2019, Round 47. Please see Round 47 page for registration and more details.

This round features one target, Target 160. The is the assembly domain of SAP the surface layer (S-layer) protein from a highly pathogenic microbe. This assembly domain comprises 6 discrete structural domains, which self assemble into a 2D array whose structure is currently under study by Cryo-EM.

Protein-protein interactions and other interactions between macromolecules are essential to all aspects of biology and medical sciences, and a number of methods have been developed to predict them. CAPRI is a community wide experiment designed to assess those that are based on structure. If we know the 3D structure of two components of a complex and build a model of their assembly, how reliable and accurate is that model likely to be?

CAPRI is a blind prediction experiment managed by Capri management. Its targets are unpublished crystal or NMR structures of complexes, communicated on a confidential basis by their authors to the CAPRI management. Participant predictor groups are given the atomic coordinates of two proteins that make biologically relevant interactions. They model the target complex with the help of the coordinates and other publicly available data (sequence, mutations etc~), and submit sets of ten models for assessment on the CAPRI Web site. In addition, the predictors are invited to upload larger sets that are communicated to scorer groups who evaluate and rank them, and make a separate ten-model submission. After the prediction round is completed, the CAPRI assessors compare the submissions to the experimental structure, et evaluate the models on criteria that depend on the geometry and biological relevance of the predicted interactions.

Since CAPRI began in 2001, the experiment has had two to four prediction rounds each year, with one or a few targets per round. Each round is announced a week in advance, it lasts 3-6 weeks, and the results are known in the following weeks. To be on the mailing list, please send a message to Sameer Velankar.

Structural biologists and scientists actively engaged in the study macromolecular interactions are invited to participate: 

• by offering targets - more information 
• by registering prediction rounds as predictors and/or scorers 

Please read and observe the rules of the experiment and the confidentiality agreement that all CAPRI participants must sign.

Information on targets of past CAPRI rounds is available here.

We are announcing CAPRI Round 52, offering target T186, representing a large protein complex determined at high resolution by cryo-EM, by Prof. Rouslan Efremov, at the VIB-VUB center for structural biology, Brussels Belgium

This Round which was initially planned as to start May 14, 2021, was not correctly announced on the CAPRI website. We have therefore shifted the timing so that the Round is now starting on June 1, 2021.

The target is part of a stable sub-complex (peripheral arm) of the E.coli respiratory complex:  NADH:Ubiquinone oxidoreductase, which comprised 6 polypeptides. One of the subunits has a ~100 residue loop, bound to a large pocket of the complex. The bound loop adopts a well-defined conformation that contains little secondary structure.

An adequate template representing a related complex from T.thermophilus, lacking the 100 residue loop of the E.coli complex is available. The main challenge of this target will be to correctly model the bound conformation of the ~100 residue loop in the E.coli complex.