Closed andreas-wilm closed 11 months ago
Hi Andreas, thanks for checking. We indeed have thought about it, and performed some preliminary tests of that, but haven't had the opportunity to perform full-scale testing and development for that scenario. Our preliminary tests indicated that interface pLDDT score is possibly more predictive of true interactions, but we have much more to do before coming to any clear conclusions. If we do end up moving ahead with that, we can let you and others know via web server and/or Github update (and a preprint/writeup, if warranted).
Best, Brian
On Wed, Oct 11, 2023 at 10:35 PM Andreas Wilm @.***> wrote:
Thank you for creating and sharing TcrModel2!
I was wondering if I could use TcrModel2 to rank a list of peptides (fixed MHC) against a TCR, or likewise, could I rank a list of TCRs against pMHCs based on their structural compatibility? For this, TCRDock has the concept of a binding score, which is computed by taking the AlphaFold PAE for the TCR:pMHC interface and subtracting TCR-intrinsic and pMHC-intrinsic factors (see e.g. Figure 3 in the TCRDock paper https://elifesciences.org/articles/82813). Can something similar by achieved by using TcrModel2'sTCR-pMHC ipTM?
Many thanks, Andreas
— Reply to this email directly, view it on GitHub https://github.com/piercelab/tcrmodel2/issues/9, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSP6VO6VIQ23KKEJFRPNIDX65JQRANCNFSM6AAAAAA5444QUU . You are receiving this because you are subscribed to this thread.Message ID: @.***>
Excellent. Thanks!
Thank you for creating and sharing TcrModel2!
I was wondering if I could use TcrModel2 to rank a list of peptides (fixed MHC) against a TCR, or likewise, could I rank a list of TCRs against pMHCs based on their structural compatibility? For this, TCRDock has the concept of a binding score, which is computed by taking the AlphaFold PAE for the TCR:pMHC interface and subtracting TCR-intrinsic and pMHC-intrinsic factors (see e.g. Figure 3 in the TCRDock paper). Can something similar by achieved by using TcrModel2'sTCR-pMHC ipTM or is this score on its own already a good proxy?
Many thanks, Andreas