Design of linear and cyclic peptide binders of different lengths from protein sequence information
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (PREreview)
Abstract
Structure prediction technology has revolutionised the field of protein design, but key questions such as how to design new functions remain. Many proteins exert their functions through interactions with other proteins, and a significant challenge is designing these interactions effectively. While most efforts have focused on larger, more stable proteins, shorter peptides offer advantages such as lower manufacturing costs, reduced steric hindrance, and the ability to traverse cell membranes when cyclized. However, less structural data is available for peptides and their flexibility makes them harder to design. Here, we present a method to design both novel linear and cyclic peptide binders of varying lengths based solely on a protein target sequence. Our approach does not specify a binding site or the length of the binder, making the procedure completely blind. We demonstrate that linear and cyclic peptide binders of different lengths can be designed with nM affinity in a single shot, and adversarial designs can be avoided through orthogonal in silico evaluation, tripling the success rate. Our protocol, EvoBind2 is freely available https://github.com/patrickbryant1/EvoBind .
Article activity feed
-
-
This Zenodo record is a permanently preserved version of a PREreview. You can view the complete PREreview at https://prereview.org/reviews/12691534.
In this preprint, the authors describe the second generation of their EvoBind, which iteratively introduces mutations in a peptide to increase its fitness to host protein. In EvoBind2 some updates were introduced and the metrics for measuring fitness are discussed as well.
The experimental validation of the designed peptides using SPR was a very interesting aspect of the manuscript. The authors describe very good success rates (close to 50%), with the stronger binder showing higher affinity than their positive control.
Major issues
I missed the comparison with other de novo design tools in terms of execution times. EvoBind (first generation, I will call it EvoBind1 to avoid confusion with …
This Zenodo record is a permanently preserved version of a PREreview. You can view the complete PREreview at https://prereview.org/reviews/12691534.
In this preprint, the authors describe the second generation of their EvoBind, which iteratively introduces mutations in a peptide to increase its fitness to host protein. In EvoBind2 some updates were introduced and the metrics for measuring fitness are discussed as well.
The experimental validation of the designed peptides using SPR was a very interesting aspect of the manuscript. The authors describe very good success rates (close to 50%), with the stronger binder showing higher affinity than their positive control.
Major issues
I missed the comparison with other de novo design tools in terms of execution times. EvoBind (first generation, I will call it EvoBind1 to avoid confusion with EvoBind2) had quite long execution times due the extensive sampling of the sequence space. Also, the memory requirements were somewhat quite high. I couldn't myself get it to run on Google Cloud machines, for example. If we compare this requirements with RFDifusion, for example, we have quite fast inference times for the later with cheap memory requirements. A few words about how EvoBind2 compares to EvoBind1 and, possibly with RFDiffusion or other comparable tools, would add a lot here.
Minor issues
The discussion about the 'Affinity and in silico metrics' is very interesting. I believe I would not expect a very high correlation between affinity and pLDDT. However, how does the affinity compares to AFM interface PAE? And how does it compare with ipTM? I guess these are the metrics that were introduced specifically to account to the reliability of the prediction for an interaction, so I would expect them to perform better than pLDDT. It would be great if the authors could bring this discussion their manuscript.
Finally, It would be great to think about the future in peptide design. How can we think in a clever way for other types of cyclization beyond the positional encoder offset (disulfide bond, for example) in AF2/EvoBind2? Is it possible to envision a way to introduce non-canonical amino acids?
Competing interests
The author declares that they have no competing interests.
-