Week 5 HW: Protein Design II

Part A: Generate Binders with PepMLM

Begin by retrieving the human SOD1 sequence from UniProt (P00441) and introducing the A4V mutation.

Sequence with mutation: MATKVVCVLKGDGPVQGIINFEQKESNGPVKVWGSIKGLTEGLHGFHVHEFGDNTAGCTS AGPHFNPLSRKHGGPKDEERHVGDLGNVTADKDGVADVSIEDSVISLSGDHCIIGRTLVV HEKADDLGKGGNEESTKTGNAGSRLACGVIGIAQ

Using the PepMLM Colab linked from the HuggingFace PepMLM-650M model card: Generate four peptides of length 12 amino acids conditioned on the mutant SOD1 sequence.

WRYYAAALRHKG WRYYAVAARHKK WRSYVVVLELGG HHYPAVAVALKG FLYRWLPSRRGG

To your generated list, add the known SOD1-binding peptide FLYRWLPSRRGG for comparison.

Part B

Peptide #1: WRYYAAALRHKG AlphaFold: ipTM = 0.38 pTM = 0.8 Record the ipTM score and briefly describe where the peptide appears to bind. Does it localize near the N-terminus where A4V sits? Does it engage the β-barrel region or approach the dimer interface? Does it appear surface-bound or partially buried?

ipTM = 0.38 is low confidence of binding. It would be surface-bound and far from the beta-barrel region. Number #2 WRYYAVAARHKK Alphafold: ipTM = 0.31 pTM = 0.75

Record the ipTM score and briefly describe where the peptide appears to bind. Does it localize near the N-terminus where A4V sits? Does it engage the β-barrel region or approach the dimer interface? Does it appear surface-bound or partially buried? Surface-bound, relatively close to the mutation site. It does not engage the beta barrel site. iPTM = 0.31 which shows low confidence.

Number #3 WRSYVVVLELGG Alphafold: ipTM = 0.75 pTM = 0.86

Record the ipTM score and briefly describe where the peptide appears to bind. Does it localize near the N-terminus where A4V sits? Does it engage the β-barrel region or approach the dimer interface? Does it appear surface-bound or partially buried? This one is bound on the surface. ipTM = 0.75 which is somewhat confident (we treat a score between 70 and 90 as “confident”). It is somewhat close to the beta-barrel area. It isn’t too far from the mutation area, but also not super close.

Number 4: HHYPAVAVALKG Alphafold: ipTM = 0.29 pTM = 0.86

Record the ipTM score and briefly describe where the peptide appears to bind. Does it localize near the N-terminus where A4V sits? Does it engage the β-barrel region or approach the dimer interface? Does it appear surface-bound or partially buried? ipTM = 0.29 which is very low confidence of binding. No, it’s not close to the mutation site. It’s on the opposite side of where the beta-barrel sits. It is surface-bound and closest to the positions 86-87.

Number 5: FLYRWLPSRRGG Alphafold: ipTM = 0.35 pTM = 0.8

Part C

Record the ipTM score and briefly describe where the peptide appears to bind. Does it localize near the N-terminus where A4V sits? Does it engage the β-barrel region or approach the dimer interface? Does it appear surface-bound or partially buried?

ipTM = 0.35, hence confidence of binding is low. It’s far from the N-terminus that has the A4V. It is surface-bound. It is close to the beta-barrel region (the arrows).

In a short paragraph, describe the ipTM values you observe and whether any PepMLM-generated peptide matches or exceeds the known binder.

Yes, WRSYVVVLELGG has a higher iPTM (0.75 vs 0.35) plus a lower perplexity score (18 vs 20), even though it has the highest perplexity score out of the other generated ones. In general, the iPTM values for all of them except this one are low (~0.30), meaning that AlphaFold is not very confident about their vinding affinity.

In a short paragraph, describe what you see. Do peptides with higher ipTM also show stronger predicted affinity? Are any strong binders predicted to be hemolytic or poorly soluble? Which peptide best balances predicted binding and therapeutic properties?

Yes, WRSYVVVLELGG performs better overall, with a higher iPTM score (0.75 vs 0.35) and a lower perplexity score (18 vs 20), even though it still has the highest perplexity among the generated candidates. In general, though, the iPTM values for most of the peptides are fairly low (~0.30), suggesting that AlphaFold is not especially confident in their binding interactions.

WRSYVVVLELGG also scores the highest iPTM in AlphaFold and the strongest predicted binding affinity in PeptiVerse. The trends are pretty consistent between both platforms, with stronger binders generally showing higher scores across methods. Solubility is close to 1.0 for all peptides, so none seem to have major solubility concerns.

Most of the peptides also show low hemolytic probabilities (<0.1), with one exception: the strongest predicted binder. Interestingly, this peptide is also one of only two with a positive GRAVY score, meaning it is more hydrophobic. Since hydrophobic peptides are more likely to disrupt membranes, this raises some concerns from a toxicity standpoint. In other words, the peptide with the strongest predicted binding also appears to carry the highest risk for toxicity and aggregation.

Peptide #1 (WRYYAAALRHKG) has the lowest predicted hemolysis score, giving it a strong safety profile, but its binding affinity is relatively modest, making it a weaker candidate overall. Peptide #2 shows weak binding affinity and a high positive charge, so it is likely less favorable. Highly positively charged peptides can increase nonspecific interactions with negatively charged membranes and biomolecules, which may increase toxicity, hemolysis, and off-target effects.

Choose one peptide you would advance and justify your decision briefly.

I would choose peptide #1 because it has the second highest binding affinity (despite still being moderately low) but has a good safety profile.