Perennial question, and a bit complicated
Ellis Walentine
This is the classic woodworker question, how to plan for wood movement. It actually depends on a lot of variables: the species of wood, the cut of the wood (quartered, flatsawn, rift), the width of the piece, the current moisture content, and the expected maximum and minimum MC down the line, plus whatever safety factor you want to add.
To calculate wood movement within the typical seasonal MC range, you need to know the shrinkage coefficient for that species and cut. Radial and tangential coefficients are different, with radial usually being significantly smaller than tangential. The difference between them is generally proportional to the amount of warp you might expect in that board.
I get the shrinkage coefficients in the Wood Handbook, originally published by the USDA Forest Products Laboratory and now in the public domain. Lee Valley sells it under the name Wood Handbook, Wood as an Engineering Material. According to the Wood Handbook, the shrinkage coefficient for quartersawn walnut (CR) within the 6% to 14% MC range is .00190. Multiply that by the width of your piece and the expected difference in MC to get the amount of movement. Here's the equation:
ΔD = D1 (CR (MF - M1))
If your piece is 26" wide and currently at 10% MC, and you think it could, in the driest heating season, get down to 6%,
D1 = 26
MF = 6
M1 = 10
the equation is
ΔD = 26 (.00190 ( 6 - 10 ))
ΔD = 26 (.00190 ( -4 ))
ΔD = -.1976"
If your current MC is 14%, double this figure.
The upshot is that if this is actually radially sawn and if it is a relatively typical piece of walnut, you're not going to see more than about 3/8" of total expansion/contraction across this MC range; and since the range is likely to be even smaller, the movement is likely to be even less. Here's where the safety factor comes in. Why cut it so close that a slight difference in one variable will ruin your whole calculation? Plan for the worst, which would be a total of .3952" (or more, if you think the piece will ever live in drier or wetter climates, or if the grain isn't strictly radial).
The actual dimensioning of parts -- panel widths, tongue & groove dimensions -- relies on the current MC of the piece. If it is at the high end (14%, and I hope not), you can expect most or all of the movement to be shrinkage, and much less to be expansion. In this case, you don't need to leave much (if any) headroom at the bottom of your side grooves for expansion, but you do want to allow for contraction of up to .3952" overall, or more depending on your safety factor, call it 1/2". That means that you need at least 1/4" tongues on the sides of your panel to keep them from pulling out of the grooves if your 14% wood dries to 6% at some point. So, to design for this worst-case scenario, you'd want tongues even wider, say 3/8", and grooves to match. And, you'd want the panel fairly close to bottomed out in the grooves, since it is not going to expand.
The opposite is true for wood at the low end of the MC range. If your wood is at 6% now (freshly KD stock), it can only expand from here unless you move to Phoenix or the Mojave Desert. So, you need to allow plenty of headroom in your grooves for expansion.
Obviously, you don't want to see huge gaps between frame and panel, so it's best to work with wood that is around the middle of the expected MC range, so the gaps can tighten up in the summer but not look overly large in the winter.
Another phenomenon of wood movement is that annual swings are highest in year 1 and decline significantly after that. So, if you make it to year two and the panel hasn't fallen out, you probably don't have any further worries. About the panel at least. 
Sorry for the nosebleed. It's actually pretty easy to predict this stuff from past experience with different woods and widths, but it doesn't hurt to have a grasp of all the dimensions of the problem and the calculations involved.
You can also work from the "Green to 6%" coefficients. Hoadley explains all that in Understanding Wood.
Ellis