Computing peptide-protein binding affinity

[rjrich]

Does boltz-1 include any capability for computing peptide-protein binding affinity or might it have such capability in the future? If not, do you have recommendations for methods, software, or web servers for computing peptide-protein binding affinities? I would already have the peptide-protein complex (either from the PDB or predicted by boltz-1); therefore, I would not need the method, software, or server to do docking or other predictions of the peptide-protein complex. Thank you.

[gcorso]

Hi @rjrich, Boltz-1 doesn't currently support binding affinity predictions although this is definitely something that we are interested in adding for the future. In the meantime, users have reported good correlations between Boltz-1's confidence_score or iptm and binding affinity, while others have integrated the poses predicted by Boltz-1 with pipeline of other tools like Rosetta or OpenMM

[navvye]

@rjrich If you're looking for protein-protein binding affinity, we just launched APPT yesterday. We'll incorporate protein-peptide binding affinity soon

[rjrich]

Many thanks for this news! Could APPT be used for peptide-protein affinity as well as for protein-protein affinity? What is the cutoff in terms of, say, the sequence length between a protein and a peptide? Best wishes!

…

On Fri, Jan 17, 2025 at 12:26 AM Navvye Anand ***@***.***> wrote: @rjrich <https://github.com/rjrich> If you're looking for protein-protein binding affinity, we just launched APPT <https://github.com/Bindwell/appt> yesterday. We'll incorporate protein-peptide binding affinity soon — Reply to this email directly, view it on GitHub <#151 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAP6T7NNZU2CRHG3ZDLYKO32LCH7HAVCNFSM6AAAAABVDLD3DGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKOJXGQ3DCNJUGE> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[navvye]

Hey @rjrich,

APPT can also utilized for protein-peptide binding affinity, as it requires only FASTA sequences. We're adding more data points for protein-peptide binding affinity by the end of this week. We haven't defined a cutoff for peptides vs proteins yet — but it probably will be ~50 amino acids.

[rjrich]

Thanks for the update. Did you intend to say "by the end of next week" instead of this week? Thanks for the amazing work you and your colleagues are doing!

…

On Fri, Jan 17, 2025 at 12:43 PM Navvye Anand ***@***.***> wrote: Hey @rjrich <https://github.com/rjrich>, APPT can also utilized for protein-peptide binding affinity, as it requires only FASTA sequences. We're adding more datapoints for protein-peptide binding affinity by the end of this week and will have a better model by then. — Reply to this email directly, view it on GitHub <#151 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAP6T7NP3AL4RYWQWYXSTO32LE6NLAVCNFSM6AAAAABVDLD3DGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKOJYHA3TKOBWHE> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[navvye]

Nope, I meant by Monday :)

Thank you for the compliments! A star on APPT and PLAPT would really help.

[rjrich]

Wow! Not only amazing, but also amazingly fast! I was not familiar with stars, but I found out how to add them -- gladly done! Regarding PLAPT, it would be great if you could add ways to add 3D ligand formats such as SDF, MOL2, or PDB. Lately, I've been working with ferrocene derivatives, and I have been unable to get SMILES strings to work for these organometallic complexes. Thanks again.

…

On Fri, Jan 17, 2025 at 12:59 PM Navvye Anand ***@***.***> wrote: Nope, I meant by Monday :) Thank you for the compliments! A star on APPT <https://github.com/Bindwell/appt> and PLAPT <https://github.com/Bindwell/plapt> would really help. — Reply to this email directly, view it on GitHub <#151 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAP6T7K3FCDRUOYUHHVTATD2LFAHNAVCNFSM6AAAAABVDLD3DGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKOJYHEYDCOJUGA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[navvye]

@rjrich Could you email me at [email protected]? I'd be more than happy to go through PLAPT for your specific use case — I believe there's a way to get SMILES from these complexes that's a bit involved, but do-able. We can set up a pipeline for that.

[rjrich]

Thanks for your kind offer. I have sent a message to the address you provided, and to that message I have attached a v3000 SDF of one of my somewhat exotic structures. This version of SDF can incorporate information about "dative" bonds. Ferrocene compounds have 10 dative bond between the central Fe(II) and the two parallel cyclopentadienyl anion rings.

…

On Fri, Jan 17, 2025 at 1:42 PM Navvye Anand ***@***.***> wrote: @rjrich <https://github.com/rjrich> Could you email me at ***@***.***? I'd be more than happy to go through PLAPT for your specific use case — I believe there's a way to get SMILES from these complexes that's a bit involved, but do-able. We can set up a pipeline for that. — Reply to this email directly, view it on GitHub <#151 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAP6T7MVWQWBRLUJ23TUYQ32LFFH3AVCNFSM6AAAAABVDLD3DGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKOJYHE4DGNZXHA> . You are receiving this because you were mentioned.Message ID: ***@***.***>