Dear ENCODE-DREAM Challenge Participants,
This post is intended to provide information on the next phase of the ENCODE-DREAM Challenge and the DREAM Conference
**Conference Round**
* Teams autosome.ru and J-TEAM will present their method at the DREAM Conference on November 9, 2016. We will be recording the presentations and making them available after the conference. We are also looking into live streaming the DREAM presentations. We will send out information when we know more.
* For the Conference Round, we have collected code and write-ups from most teams. Those teams that have supplied this information will be included in the ?ENCDOE-DREAM Community? that will be included in the challenge overview paper.
* The top 2 teams from the Conference Round (autosome.ru and J-TEAM) will be coauthors on the overview paper submission and their methods will be highlighted in the paper.
* Since the next phase of the Challenge will be starting soon, teams can keep their code and write-ups private, but we do encourage teams to share at least their write-ups with the group. We will be contacting you separately to ask if you are willing to share your write-ups and we will link to them via the results tables posted in section ?5 ? Conference Round?.
* Please respect the data terms of use that are in place throughout the competition.
**Phase 2 of the ENCODE-DREAM Challenge**
* The second phase leaderboard queues will be open early next week. The queues are
* ENCODE-DREAM Transcription Factor Binding Site Prediction Challenge (LEADERBOARD ROUND 2) (7366344)
* ENCODE-DREAM Transcription Factor Binding Site Prediction Challenge (FINAL ROUND 2) (7373880)
* During the first phase of the challenge, a few datasets were flagged as potentially problematic. Specifically, the SK-N-SH DNase dataset was found to be from a stimulated condition whereas theChIP-seq datasets were in unstimulated conditions. Hence, we have removed all SK-N-SH datasets. These were all restricted to training datasets. Second, some of the newer, unreleased ATF2 ChIP-seq datasets showed irregular overlap fractions with DNase peaks in the matched cell type. To be on the safe side, we have eliminated this TF from the competition. ATF2 was also not used in the final round 1 evaluation and ranking. We have eliminated these datasets in Phase 2 of the challenge.
* This phase will be structured the same as the first phase, where teams are given 10 submissions per TF/Cell type pair in the leaderboards and then allowed one final submission, where we will score the last submission made by teams.
* We had problems with a flood of last minute submissions in the Conference Round. We believe that this has been addressed, but we request you not to wait until the last minute to make your submissions as it can potentially massively overload the system and not give us sufficient time to respond to all your queries.
Please follow up with any questions here. Suggestions are welcome as well.
Kind Regards,
ENCODE-DREAM Organizers
Created by James Costello james.costello Hi,
in my understanding, the second phase of the challenge addresses the *within-cell-type* prediction
and that the ChIP-seq data for the held-out cell-types would be released after the first phase.
However, I wasn't able to find the ChIP-seq data for the held-out cell-types.
Am I mistaken about the assumption that the held-out ChIP-seq data will be released
or where can I find it?
Thank you!
Wolfgang Hi Opt,
Yes, any team that participates, supplies code and a write-up at the end of the challenge will be included in the Community. If a team performs better than the top two teams in the Conference Round and is in the top two teams at the end of the challenge, then they will be included as co-authors on the overview paper submission.
Cheers,
Jim Will teams that submit to Phase 2 be included in the ENCODE-DREAM Community and will the top performing teams from that phase be co-authors on the overview submission?