CAGI1 Challenge

The first round of CAGI was organized in the fall 2010. We want to express our appreciation to everyone who participated in CAGI and RiskSNPs 2010. This initial experiment was a great success with over 100 prediction submissions from 8 countries. The workshop on 10 December 2010 gathered an enthusiastic international group of 40 people at Berkeley. A summary of the CAGI 2010 Conference proceedings has been published in Nature News.

CAGI 2010 Results

The organizers and assessors are now finalizing the assessments of CAGI 2010. We are planning to write a manuscript describing the results.

Assessor Pauline Ng's results on the CBS dataset are available for download:

Presentations on CAGI

Presentation by Assessor Sean Tavtigian at the pre-pro-CAGI workshop, 10 December 2010.

Presentation by Assessor Pauline Ng at the ISMB 2011 conference. The presentation includes the assessments of the CBS and CHEK2 datasets of CAGI 2010. Presentation available for registered users.

Video presentation by Predictor Rachel Karchin and her group is available. The presentation describes methods that Prof. Karchin developed to participate in the PGP challenge.

CAGI 2010 Videos

pre-pro-CAGI Workshop, At University of California, Berkeley 10 December, 2010

CAGI Dissemination and Embargo

CAGI aims to advance phenotypic interpretation of genomic variation. However, the success of CAGI hinges critically on the generous contribution of pre-publication datasets and the participation of predictors and assessors. Our hope is that CAGI will enhance their scientific activities. It has the potential to elevate careers of participants, allowing recognition for achievements. However, it also has the possibility of harming some of those who participated if their contributions are presented inappropriately. Especially as pre-pro-CAGI encouraged imperfect rapid submissions and preliminary assessments to help develop the CAGI experiment, it would be grossly inappropriate for anyone's participation to redound negatively.

One key aspect of this is dissemination of CAGI results to the community. However, to prevent harm to any individual or group, we are imposing an initial partial embargo on the CAGI results. We will have discussion at the CAGI meeting regarding publication plans and public dissemination of the results. Until the embargo is modified or lifted, the policies are:

  • All experimental dataset phenotypes are confidential until publication and may not be shared with anyone outside the immediate group of predictors or assessors without explicit written permission from the CAGI organizers.
  • Predictors may not present the assessors' results for any other identified predictors, other than their own, by name. All presentation of assessors' results must be labeled as preliminary.
  • Predictors may not present any assessment made themselves as being a CAGI assessment.
  • Discussions about specific methods at the meeting are to be shared only the immediate group of attendees.

Videos of the meeting presentations are available below. To view the videos, you need to have Quicktime installed on your computer. If you are using Windows, please go here to download the software: http://www.apple.com/quicktime/download/. On Macs, the software is already installed. CAGI 2010 Conference Videos are available here for registered users.