Chapter 35 — Exercises

How to use these. Items are lettered A through AI. Work them in order if you are reading the chapter straight through; otherwise use the part headings to find the ones that match what you need. Items marked † are the harder ones — they require you to combine this chapter with an earlier one, or to defend a position rather than recall a fact. There are no answers here; several items have more than one defensible response, and the reasoning is the point.

Part 1 — The four routes (§35.1)

A. State the four routes to a peptide drug in your own words, in one sentence each. Do not look back at the chapter until you have tried.

B. For each of the following, name the route that produced the first useful molecule: insulin, semaglutide, exenatide, ziconotide, a peptide selected by phage display against a cancer antigen, a binder generated by RFdiffusion.

C. † Rank the four routes by how much prior knowledge they require before you can begin. Then explain why the route requiring the most prior knowledge has the highest historical hit rate, and what that implies about the relationship between knowledge and luck in drug discovery.

D. The chapter claims all four routes answer the same question. State that question in one sentence, and state the question none of them answers.

Part 2 — Venom (§35.2, §35.3, §35.4)

E. Explain, in three sentences, why the selective pressures acting on venom overlap with the design specification for a drug.

F. A friend says "venom is a poison, so of course it's non-selective — it just wrecks everything." Correct them, using the metabolic-economics argument from §35.2.

G. What structural feature makes many venom peptides resistant to proteases, and why does that feature interfere with a protease's normal mode of action?

H. §35.2 says venom peptides often arrive "pre-solved" for the stability problem of Chapter 4. Name two problems from Chapter 4 that venom does not solve, and explain why a stable peptide can still be undeliverable.

I. State precisely what exendin-4 shares with human GLP-1 and what it does not. Then state what each of those two facts contributes to its usefulness as a drug.

J. Explain why a difference at a single position can make exendin-4 invisible to DPP-4 while leaving its ability to activate the GLP-1 receptor intact. Which structural level from Chapter 1 is doing the work here?

K. † Write the strongest one-paragraph version of the claim "the exendin-4 story shows that nature is a better designer than chemists are." Then write the rebuttal the chapter gives. Which paragraph was easier to write, and what does that tell you about why the first version circulates more widely?

L. Exenatide validated the GLP-1 target and was then displaced. Name the specific pharmacokinetic shortcoming that made displacement possible, and name the engineering solution that addressed it.

M. Trace the captopril lineage in four steps, from the pit viper to the pharmacy. Name the molecule at each step and say whether it is a peptide.

N. † The chapter says the captopril arc and the orforglipron arc (Chapter 33) have the same shape. Draw both as a three-step diagram. Then name one important respect in which they differ, and say whether that difference undermines the parallel.

O. Why must ziconotide be given intrathecally? Answer in terms of where its target is and what it cannot cross.

P. † Ziconotide is described as "Chapter 4's delivery problem in its most extreme clinical form." Construct the argument that ziconotide is evidence against casual claims that a given peptide "crosses the blood-brain barrier." Be specific about what ziconotide's existence demonstrates.

Q. What did Michael Zasloff observe that led to the magainins, and what field did their characterization open?

Part 3 — Rational design and display (§35.5, §35.6)

R. List four peptide drug families that came from route 2, naming the parent hormone in each case.

S. Explain the sentence "you inherit the endogenous ligand's selectivity profile, including its off-target activity," using a specific drug class as the example.

T. † A patient asks why the nausea from a GLP-1 receptor agonist cannot simply be engineered away. Write the answer in three sentences, at a reading level a non-scientist can follow, without using the word "on-target."

U. What is the physical link at the heart of phage display, and why is it the thing that makes the whole method work? Answer without using the word "sequence."

V. Why can ribosome and mRNA display achieve larger libraries than phage display? Name the specific bottleneck that they avoid.

W. † The chapter insists that "display technologies do not design anything." Defend that statement against someone who points out that the researcher designed the library, chose the target, and set the wash stringency. Where exactly is the line between designing and selecting?

X. You read a paper reporting a display-derived peptide with picomolar affinity for a receptor. List the four things you would check next, in order, and say what each one would rule out.

Part 4 — Structure, prediction, and design (§35.7, §35.8, §35.9)

Y. Why were class B GPCRs historically difficult to study structurally, and why does that matter specifically for the drugs in this book? What can cryo-EM produce that crystallography largely could not?

Z. State the four things AlphaFold-class structure prediction did not change. Then say which of the four is most specific to peptides, and why.

AA. † The chapter says that for many short peptides, "predict the structure" is close to a malformed question. Explain what makes it malformed, using the Chapter 1 distinction between a property of a molecule and a property of a molecule plus its partner. Then say what a prediction tool will nonetheless return, and why that is a trap.

AB. Structure prediction confidence scores are lowest for disordered regions. Explain why that is correct behavior for the model and simultaneously inconvenient for biology.

AC. What does de novo design do that routes 1 through 3 cannot? Answer in one sentence about sequence space.

AD. † De novo designed binders are rated 🔬 in this chapter. Write the four-line evidence rating you would issue for the same technology in a hypothetical 2036 where several designed peptides have completed Phase 3. What specifically would have to have happened for you to write ✅ rather than ⚠️?

Part 5 — The bottleneck (§35.10) and synthesis

AE. † Here is the chapter's general rule: a technology that improves one stage of a pipeline improves the whole pipeline only if that stage was the constraint. Apply it to three cases outside medicine entirely — one where the improved stage was the constraint, one where it was not, and one you are genuinely unsure about. Then return to drug discovery and say which of your three cases it most resembles.

AF. Name the two dominant causes of clinical failure and explain, for each, why faster candidate generation does not address it.

AG. State the two real benefits of cheaper discovery that the chapter concedes. Then state, in one sentence, the claim those benefits do not support.

AH. † Chapter 22's substance P antagonists are described as the cleanest demonstration in the book. Write a paragraph explaining what would have happened if those compounds had been discovered by AlphaFold-guided design instead of conventional medicinal chemistry — and be honest about which parts of the story would have been different and which would have been identical.

AI. † Take any ⚠️-rated claim from your own dossier and write the six-component specification from §35's Field 12 exercise. Then identify which single component carries the weight — and, separately, find one real advertisement or press release that uses an origin story where evidence should be, writing down the origin claim, the implication it invites, and the evidence that is missing.