Recording (and most of our past seminars) are released here, www.youtube.com/@CASPRNASIG, enjoy!

And of course thanks to John Moult and Andriy Kryshtafovych for helping recruit all these amazing structural biologists to donate their structures and time to the CASP effort. Thanks for letting me coordinate this paper for the second time, such a cool collection works!

(10) @emillk.bsky.social , Nikolaj H. Zwergius and @esandersen.bsky.social discuss 2’F modified RNA origami and protein-binding aptamer
(11) Heewhan Shin and @phoeberice.bsky.social discuss SPβ LSI-RDF synaptic complex
(12) Reinhard Albrecht, Yimin Hu and Marcus D. Hartmann discuss UDE bound to DNA

(7) Shekhar Jadhav, Michela Nigro, and Marco Marcia discuss group IIC intron
(8) Liu Wang, Jiahao Xie, and Zhaoming Su discuss ROOL RNA nanocage
(9) @rachael-kretsch.bsky.social , Yuan Wu, Rhiju Das, Wah Chiu discuss OLE RNA dimer

(3) Anna Lewicka, Yoshita Srivastava and Joseph A. Piccirilli discuss a 3’ CITE
(4) @cpricejones.bsky.social discusses ZTP riboswitches
(5) Hsuan-Ai Chen and Claudia Höbartner discuss SAMURI
(6) Eric Largy, Philip E. Johnson, and @cdmackereth.bsky.social discuss a DNA aptamer

We each analysed the predictions for the target we expirementally determined, resulting in 12 excellent chapters:
(1) @jgezelle.bsky.social and @lenasteckelberg.bsky.social ‪‪discuss a xrRNA
(2) Manju Ojha and Deepak Koirala discuss HIV RRE SLII

So much fun working with and learning from some of the best NA structural biologists across the world. It is the hard work of experimental structural biologists that will provide the data and, perhaps most importantly, set the goal-posts for what structure prediction should be aiming to achieve.

I am excited to see the innovation that will progress nucleic acid structure prediction accuracy as evaluated in RNA puzzles and CASP17! I am perhaps even more excited to see the new ideas that will push our notion of “structure” (or what objectives structure prediction should have) farther.

CASP16 did challenge predictors with very challenging large RNAs with no previous structures. While no predictors were able to tackle these problems, there are early indications that human experts and automated methods can extract tertiary information from MSAs if sufficiently deep!

Despite, sub-human performance, servers have seen a jump in performance in CASP16; for the first time servers are able to improve on best available structure template.

The dependence on template availability was observed in RNA-RNA, NA-protein, and NA-ligand interaction prediction in our RNA-multimer, hybrid and NA-ligand category assessments. These categories and the prediction of DNA aptamer structure remain key future challenges.

Throughout the history of NA structure prediction, accuracy has been dependent on template availability, a trend that unfortunately continues in CASP16. Protein structure prediction has shown this is a surmountable challenge.

Prediction of other motifs, such as A-minor interactions, that form long range interactions important for stitching together the 3D structure remains challenging.

However, CASP16 participants, including automated methods trRosettaRNA and AF3, are more accurate at predicting secondary structure than popular secondary structure prediction algorithms.

Although the AF3-server can now predict nucleic acid structure, its performance on nucleic acid monomer structure prediction was average; 12 groups out-performed AF3.

As in CASP15, expert human groups ranked best across targets and categories. These groups are increasingly using deep-learning algorithms to sample structures, which they often then refine or select using physics based algorithms or expert knowledge.

Work by @alissahummer.com, Shujun He, Rongqing Yuan, Jing Zhang, Thomas Karagianes, Qian Cong, Andriy Kryshtafovych, Rhiju Das, and sizable participation of scientists across the world which provided 42 targets and predictions from 65 groups.

Please reach out to me in a message or email if interested!

We will be meeting, Thursdays 3pm UTC = 8am PDT / 11am EDT / 4pm CEST / 11pm CST. For our colleagues in the more difficult timezone, we can accommodate changes in time!