Huh. The extra 2N7002 pins aren't for thermal simulation, are they?
Anyway, assuming mid frequency conditions, the capacitors go away, and we see that the bottom half of RS is shorted into Vin. We don't care about that current flow, so we can ignore it entirely. That reduces the circuit to a variable resistor between M1-S and Vin.
We calculate the DC bias condition with the 82k + 68k gate divider, Vgs(th) and K_n, and RS (the total, since C3 is open at DC). From this we get the small-signal equivalent, which will have an effective source input resistance approx. 1/Gm and drain output resistance something something lambda.
We calculate the AC equivalent circuit, where the capacitors are effectively shorted and the resistances and gain are as found in the previous step.
Tim