One significant advantage of OpenFold -- beyond its active open-source
development, its pytorch implementation (thus being more interoperable with
common external ML libraries) -- is its release of training code. So if
Odyssey wants to fine-tune predictions on a protein family of particular
interest to Odyssey, OpenFold's your answer. (You all can actually join the
OpenFold consortium -- which organizes not just OpenFold the existing
project but diverse additional open source ML development directions -- for
a small annual donation.)
In some respects single-sequence models like OmegaFold have advantages --
you don't need a heterogenous compute environment, with a good GPU for
inference but a lot of CPU cores for MSA precomputation. It's not a
total panacea though; embedding in a large language model is a significant
computation.
On Thu, Sep 22, 2022 at 11:09 AM Chip Lesburg
Hi PHENIX users,
At this past summer’s Diffraction Methods GRC meeting https://www.grc.org/diffraction-methods-in-structural-biology-conference/202..., Tom Terwilliger presented the utility of incorporating AlphaFold2 predictions during structure fitting and rebuilding. I wonder if anyone has experience using that or an alternative such as OmegaFold https://www.biorxiv.org/content/10.1101/2022.07.21.500999v1 or OpenFold https://github.com/aqlaboratory/openfold? Are any of these easier or harder to get set up?
Chip
Charles Lesburg
*Vice President, Structural Biology* *M* 908-265-2064 301 Binney Street, Cambridge, MA 02142 *odysseytx.com http://odysseytx.com*
This message is for the named person's use only. It may contain confidential, proprietary or legally privileged information. If you receive this message in error, please immediately delete it and all copies of it from your system, destroy any hard copies of it and notify the sender. ODYSSEY THERAPEUTICS, INC. ("ODYSSEY THERAPEUTICS") reserves the right to monitor all e-mail communications through its networks. ODYSSEY THERAPEUTICS cannot be committed by this e-mail or warrant the accuracy or validity of any information presented. Any views expressed in this message are those of the individual sender. _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]
