Will DeepMind publish on free energy computation before 2025
Basic
6
Ṁ173
Jan 1
57%
chance

Resolves on January 1st 2025 at the latest, or earlier if a suitable publication is exhibited.

'DeepMind' understood as Google DeepMind or any successor entity/division, as major contributor of the work. (Open to interpretation if dissolved fully in Google AI by then, but if so, based on publishing history of the publication authors for loose assignation)

Publish: peer reviewed scientific publication, in a major scientific journal, either generalist (e. g. Nature as they seems to like to do) or specific to this domain (JCTC, JCIM...). Time of apparition of the ASAP version counts, no need to be actually in print. Preprints do not count.

Free energy computation: Computation of relative or absolute free energy of binding, for small ligand to protein. Free energy of solvation explicitly excluded (too easy, already lots of non-NN stuff that works). If the work is on ligand optimization or chemical space exploration, fine, but free energy must be an output of the work (enter into the loss function at least), not only relative ranking of binders. Comparison to existing methods (FEP et al.) allowed if FE is an output of the method.

The method must be generally applicable. I judge fairly unlikely that they would publish a very minor advance (e. g. method applicable not generally but to a small chemical series), but if so, reserve the right to discount it (caveat emptor!)

Get
Ṁ1,000
and
S3.00
Sort by:

@DavidChee Sorry, I'm new here - could you tell me why this market is marked non-predictive? This is not my intention, and if necessary I can tighten the conditions (there is some ambuigity in the corner cases, but I feel it is fairly well defined and non-self referential)

@CamillePerrin it was just the "auto-tagger" going wild.

© Manifold Markets, Inc.Terms + Mana-only TermsPrivacyRules