Describe gene regulatory networks

What is a GRN?

HOW does one gene regulate another? (first principles)

WHAT is the physical mechanism?

Gene $X$ is transcribed → mRNA → translated → protein $X$ (often a transcription factor).
Protein $X$ diffuses to the DNA and binds a specific sequence (the promoter / operator) near gene $Y$ .
If $X$ recruits RNA polymerase → activation. If $X$ blocks polymerase → repression.

HOW do we turn this into math? We ask: how fast does protein $Y$ accumulate?

Rate of change of $Y$ = (production) − (removal):

$\frac{dY}{dt} = \underbrace{f(X)}_{\text{production, set by regulator }X} - \underbrace{\gamma\,Y}_{\text{degradation + dilution}}$

Why this form? Every protein is made at some rate depending on its regulators, and destroyed/diluted at a rate proportional to how much is present (first-order decay). $\gamma$ = degradation rate constant.

Deriving the input function $f(X)$ (Hill function)

Why not just $f(X)=kX$ ? Because binding is saturable — a promoter has a finite number of binding sites. Once TF is abundant, adding more can't increase binding. So we model occupancy.

Let $X$ bind the promoter with dissociation constant $K$ . Probability the site is bound:

$P_{\text{bound}} = \frac{X/K}{1 + X/K} = \frac{X}{K + X}$

Why this fraction? From equilibrium $[\text{bound}]/[\text{free}] = X/K$ ; normalising over (bound + unbound) gives the fraction above — a classic saturation curve.

Real promoters bind cooperatively ( $n$ TF molecules together), giving the Hill function:

$\boxed{\;f(X)=\beta\,\frac{X^{n}}{K^{n}+X^{n}}\;}\quad(\text{activator})$ $f(X)=\beta\,\frac{K^{n}}{K^{n}+X^{n}}\quad(\text{repressor})$

$\beta$ = maximal production rate.
$K$ = activation threshold (value of $X$ giving half-max).
$n$ = Hill coefficient (cooperativity → steepness → switch-like behaviour).

Why $n$ makes it switch-like: large $n$ makes the curve nearly a step at $X=K$ — the gene is essentially OFF then ON. That is how digital-like decisions emerge from analog chemistry.

Steady state

Set $\frac{dY}{dt}=0$ : $\gamma Y^* = f(X) \;\Rightarrow\; Y^* = \frac{f(X)}{\gamma}$

Why useful? Tells you the final protein level for a fixed input — the "output" of the circuit.

Figure — Describe gene regulatory networks

Network motifs (the 80/20 core)

Real GRNs are built from a few recurring wiring patterns = network motifs (Uri Alon). Learn these 3 and you understand most circuits:

Worked example — will gene Y switch ON?

Worked example Activator with a threshold

Gene $Y$ is activated by TF $X$ with $\beta = 10$ , $K = 5$ , $n = 4$ , $\gamma = 1$ . Input $X = 8$ . Find $Y^*$ .

Step 1 — compute Hill fraction: $\dfrac{X^n}{K^n+X^n} = \dfrac{8^4}{5^4+8^4}=\dfrac{4096}{625+4096}=\dfrac{4096}{4721}\approx0.868$ . Why this step? This is the fraction of time the promoter is bound & active.

Step 2 — production rate: $f(X)=10\times0.868=8.68$ . Why? Max rate $\beta$ scaled by occupancy.

Step 3 — steady state: $Y^*=f(X)/\gamma = 8.68/1 = 8.68$ . Why? At steady state production = degradation.

Interpretation: $X=8 > K=5$ , so the switch is ON (near-max output). With $n=4$ the transition around $K$ is sharp.

Common mistakes

Flashcards

What are the nodes and edges of a GRN?

Nodes = genes/TFs; edges = directed regulatory interactions (→ activation, ⊣ repression).

Why is a GRN a directed graph?

Regulation is causal and one-way per edge: A's product controls B's transcription, not vice versa.

Write the general ODE for a regulated protein Y.

dY/dt = f(X) - \gamma Y

(production minus degradation/dilution).

What is the Hill activation function?

f(X)=\beta X^n/(K^n+X^n)

Physical meaning of K in the Hill function?

The activation threshold — the TF concentration giving half-maximal expression.

What does the Hill coefficient n control?

Cooperativity → steepness of the response; large n gives a sharp switch-like ON/OFF.

Steady-state output of a regulated gene?

Y^* = f(X)/\gamma

(production = degradation).

What does negative autoregulation (X ⊣ X) achieve?

Faster response and reduced noise / homeostatic output level.

What behaviour does positive feedback create?

Bistability — a toggle switch with cellular memory.

What does a coherent feed-forward loop do?

Sign-sensitive delay: filters brief input pulses, responds only to persistent signals.

What wiring makes a repressilator oscillate?

A ring of an odd number of repressors (e.g. A⊣B⊣C⊣A).

Why doesn't more TF always increase output?

Promoter binding saturates; the Hill function plateaus at β.

Recall Feynman: explain to a 12-year-old

Imagine each gene is a light switch, and each light switch has a tiny robot arm that can flip other switches. Some arms turn switches ON, some turn them OFF. Draw all the arms and switches and you get a map of who flips whom — that's a gene regulatory network. Because the arms feed back on each other, the cell can do clever things: a room that stays lit even after you let go of the switch (memory), or Christmas lights that blink on their own (an oscillator). Same switches, wired differently → totally different light show. That's how one set of genes builds a brain cell or a skin cell.

Concept Map

Hinglish (regional understanding)

Intuition Hinglish mein samjho

Socho har gene ek switch hai, aur uske protein product ka kaam hai doosre genes ko ON ya OFF karna. Jab aap saare "kaun kisko control karta hai" ke arrows draw karte ho, to jo wiring diagram banta hai use hum gene regulatory network (GRN) kehte hain. Ye ek directed graph hai — arrow (→) ka matlab activation, aur blunt arrow (⊣) ka matlab repression. Yaad rakho: arrow ka matlab "A se B banta hai" nahi hai, balki "A, B ke transcription ko control karta hai" — ye information ka flow hai, material ka nahi.

Har gene ka level kaise decide hota hai? Simple: $dY/dt = f(X) - \gamma Y$ , yaani production minus degradation. Production function $f(X)$ ek Hill function hai: $\beta X^n/(K^n+X^n)$ . Yahan $K$ threshold hai (jitne TF pe aadha response), aur $n$ (Hill coefficient) batata hai response kitna sharp hai. Bada $n$ matlab ekdum switch jaisa ON/OFF. Isiliye zyada TF hamesha zyada output nahi deta — binding saturate ho jaati hai, curve flat ho jaata hai. Ye ek badi galti hai jo students karte hain.

Asli power aati hai motifs se. Negative autoregulation (X apne aap ko repress kare) response fast aur noise kam karta hai. Positive feedback bistability deta hai — cell ko memory milti hai, isiliye ek cell neuron banta hai aur doosra skin cell, same genes hone ke bawajood. Feed-forward loop chhote noise pulses ko filter karta hai. Aur repressilator (teen repressors ka ring) ekdum ghadi ki tarah oscillate karta hai. 80/20 rule: agar ye 3-4 motifs samajh liye, to zyaadatar GRN behaviour tumhe clear ho jaayega.