8. 【Lab】 The polymorphism is in the groove and nowhere else · Prof. Kwon's TCR-HLA and CD4/CD8, First Steps

HLA is the most variable locus in the human genome. Everybody knows that.

What this course asks is where that variability is. Spread evenly over the molecule, or bunched somewhere?

Measuring it is easy. Take every full-length protein sequence in IMGT/HLA and count how many different amino acids appear at each position of the mature chain. Summarise that as Shannon entropy — 0 means everybody is identical, larger means more split.

α3 runs between a third and a tenth of α1 and α2.

Python Runs in your browser

# How different is each domain
import thla

for gene in ["A", "B", "C", "E"]:
    n = thla.facts()["hla_n"][gene]
    e = thla.domain_entropy(gene)
    line = "  ".join(f"{d} {e.loc[d,'mean']:.3f}" for d in ["α1", "α2", "α3"])
    print(f"HLA-{gene:<4} {n:>6,} alleles   {line}")
print()
print("α1 and α2 = the two helices that make the groove · α3 = the platform below")

The red band is 223-229, where CD8 binds. It is flat against the floor.

Python Runs in your browser

# Draw it position by position
import matplotlib.pyplot as plt
import thla

d = thla.hla("B")
fig, ax = plt.subplots(figsize=(9, 3.4))
ax.bar(d["pos"], d["entropy"], width=1.0, color="#4C7DE0")
for x0, x1, name, col in [(1, 90, "α1", "#f2f6ff"), (91, 182, "α2", "#eef3ff"),
                          (183, 274, "α3", "#fff4ec")]:
    ax.axvspan(x0 - .5, x1 + .5, color=col, zorder=0)
    ax.text((x0 + x1) / 2, ax.get_ylim()[1] * .92, name, ha="center", fontsize=11)
lo, hi = thla.CD8_LOOP
ax.axvspan(lo - .5, hi + .5, color="#e8483f", alpha=.35, zorder=1)
ax.set_xlabel("HLA-B mature position")
ax.set_ylabel("entropy (bits)")
ax.set_xlim(0, 275)
plt.show()

These numbers coincide with the positions structural papers named 'peptide-binding pockets'.

Python Runs in your browser

# Where are the most variable positions
import thla

for gene in ["A", "B", "DRB1"]:
    h = thla.hotspots(gene, 10).sort_values("pos")
    print(f"HLA-{gene}")
    print("   " + "  ".join(f"{int(p)}({d})" for p, d in zip(h["pos"], h["domain"])))
print()
print("For HLA-A and HLA-B, 9 / 97 / 114 / 116 / 156 are the floors of the groove pockets.")
print("For DRB1, 11 / 13 / 57 / 71 / 74 are the same thing.")

Switch genes and watch variability by position. The groove and the platform separate visually.

Loading…

The most extreme chain of all — DRα

In class II, two chains contribute half the groove each. But in HLA-DR the α chain barely changes at all.

Half of the DR groove is identical in everyone. The other half does all the deciding.

Python Runs in your browser

# How much does DRα differ between people
import thla

for gene in ["DRA", "DRB1", "DQA1", "DQB1"]:
    e = thla.domain_entropy(gene)
    n = thla.facts()["hla_n"][gene]
    line = "  ".join(f"{d} {e.loc[d,'mean']:.4f}" for d in e.index)
    print(f"HLA-{gene:<6} {n:>5,} alleles   {line}")
print()
d = thla.hla("DRA")
var = d[(d["domain"] == "α1") & (d["n_aa"] > 1)]
print(f"of DRA's 84 α1 positions, the number that vary at all: {len(var)}")

🤔 Check yourself

Where is class I polymorphism concentrated?

🤔 Check yourself

What does HLA-DRA being nearly invariant mean?

🤔 Check yourself

What is the evolutionary explanation for HLA polymorphism concentrating in the groove?