<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Week 5 HW: Protein Design Part II :: 2026a-daniel-tseng</title><link>https://pages.htgaa.org/2026a/daniel-tseng/homework/week-05-hw-protein-design-part-ii/index.html</link><description>Part A: SOD1 Binder Peptide Design (From Pranam) Part 1: Generate Binders with PepMLM Begin by retrieving the human SOD1 sequence from UniProt (P00441) and introducing the A4V mutation. 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. To your generated list, add the known SOD1-binding peptide FLYRWLPSRRGG for comparison. Record the perplexity scores that indicate PepMLM’s confidence in the binders. index Type Binder Pseudo Perplexity 0 Known Binder FLRYWLPSRRGG 21.73592494089691 1 Generated AWWPVYVGVKAWRKX 12.725233370840693 2 Generated AWWGVYTVRYAWAAX 12.277514856931905 3 Generated AWYPVLVAVYELKAA 20.11580123195301 4 Generated WWWGPYAAVKELRKK 16.584271595654002 The known SOD1-binding peptide FLYRWLPSRRGG yielded a pseudo-perplexity score of ~21.74, suggesting moderate model confidence. This value provides a baseline for comparing PepMLM-generated peptides, where lower scores would indicate potentially stronger or more compatible binders.</description><generator>Hugo</generator><language>en</language><atom:link href="https://pages.htgaa.org/2026a/daniel-tseng/homework/week-05-hw-protein-design-part-ii/index.xml" rel="self" type="application/rss+xml"/></channel></rss>