IBIS news & updates, March 21, 2024

[Ivan Kulakovskiy]

Dear IBIS participants and supporters,

Here is the first digest of IBIS news & updates
with several important announcements.

(1) We will host an introductory Telecon on March 26 describing the
IBIS setup, underlying data, and key design decisions. The
presentation will be followed by an interactive Q&A session.

Date: March 26, 2024; Venue: Online, Zoom
Time: 13-00 EDT (Toronto/New York) - 18-00 CET (Berlin/Geneva)
- 20-00 EEST (Moscow/Istambul) - 02-00 (AM) JST (Tokyo)

* Zoom link will be posted at https://ibis.autosome.org/news
and in https://t.me/ibischallenge one hour before the meeting.

(2) To increase the stability of the evaluation metrics for the
genomic data, we decided to extend the ChIP-Seq and GHT-SELEX test
data with regions from both chromosome arms, which were previously
exclusive for each of the two genomic data types. For AAA submissions,
please re-download the test data package and the leaderboard AAA
submission template. Train data remains unchanged.

(3) Based on the feedback from the community, we have decided to allow
the participants to use protein-level metadata on IBIS transcription
factors (including but not limited to protein sequence and domain
information). To keep it concise and manageable, we stick to the data
available directly in UniProt (any human proteins, not only those used
in IBIS). The challenge rules posted online will be updated
accordingly. We do not expect these data to be highly useful, but we
would be glad to be surprised.

(4) The validation scripts are available online, and the benchmarking
and data preparation code is publicly open, see
https://github.com/autosome-ru/ibis-challenge/. While the benchmarking
code won't work until we publish the complete test data, you are
welcome to inspect the implementation of the benchmarking protocols
and data preparation.

(5) Last but not least, more than three dozen teams have registered
for the challenge, and we expect more to come. Please help us to
spread the word so more people can learn about the challenge. There
are many data types to process and disciplines to compete in!

On behalf of the IBIS Organizing Committee,
Vanja (Ivan V. Kulakovskiy)

[IBIS Challenge]

Dear IBIS subscribers and participants,

📢 The introductory Teleconference of the IBIS Challenge is going live in a few hours.

13-00 EDT (Toronto/New York) | 18-00 CET (Berlin/Geneva) | 20-00 EEST (Moscow/Istanbul) | 02-00 (AM) JST (Tokyo)

We will cover the available experimental data and challenge design.

Prepare your questions and join the meeting:
Zoom link: https://zoom.us/j/99921795014?pwd=eHJ0enFqVGtwUEhEWkU0M3JiNjg0QT09
Meeting ID: 999 2179 5014
Passcode: 802565

See you soon,

On behalf of the IBIS Organizing Committee,

Ivan V. Kulakovskiy

[IBIS Challenge]

Dear IBIS participants and spectators,

Our Challenge on TFBS modeling and prediction has reached its zenith
and we would like to share a few insights & updates.

(1) More than 100 teams registered for the challenge yet only two
dozen are using the opportunity to explore their models on the
Leaderboard. This is not mandatory to enter the Final stage, but it is
a nice sandbox to test your models. Don't be shy, test your mettle!

(2) Remember, the training data for Leaderboard and Final are
available online, and you are free to mix-and-match it in any
combination.

(3) For Leaderboard, we have received 650 total submissions of AAA
(advanced) and 296 PWM models (a whopping set of 6682 PWMs in total).

(4) For PWMs, the baseline is left far behind, but some TFs remain
tough nuts to be cracked.

(5) For AAAs, there is a solid improvement over the best-submitted
PWMs in A2G, but the G2A discipline remains unpopular.
Free hint: this is your chance!

(6) Another free hint: If your model tends to overfit given a limited
set of data for some TFs don't forget to use reverse-complement and
shift augmentations. Also, don't hesitate to use multitarget models
i.e. predicting the binding of multiple TFs at the same time.

(7) Last but not least, try to combine knowledge from all accessible
experiment types, especially for G2A discipline (ChIP-Seq & genomic
HT-SELEX) in a single model!

(8) Remember, that you optionally can reveal your GitHub account link
for the Leaderboard. It is fun to see each other in the crowd.

(9) Finally and importantly, following the requests from the
community, we decided to EXTEND the Leaderboard until the final
submission deadline.

(10) The final submission deadline is also EXTENDED until Aug 15. The
final submission form and details will be posted on the IBIS website
in the first half of July, follow our Telegram group and mailing list
(see the links at https://ibis.autosome.org).

Looking forward to seeing your fresh submissions on the Leaderboard!