The process I used for the cub stacks I used went like this:
 
I used mdiscal (USGS ISIS3) with only Dark Current corrections (from the pre-developed model) applied. Mdiscal also automatically adjusts for smear and non-linearity.
Then, I multiplied the image by Responsivity, where Resp. = R * [C1 + (T * C2) + (T^2 * C3)]
For all WAC-B images,
R = 7853.2
C1 = 0.95557
C2 = 4.1897 x 10^-5
C3 = 0
T = CCD temperature of the images (found in cub header).
After this, multiply by Exposure Duration in units of seconds. This is also found in the header (note the header lists this time is ms).
 
The reason I multiply by responsivity and exposure duration is because of the equations found here, where
radiance L = [Linearity – Dark – Smear] / [ t * Resp].
 
By multiplying by Resp and t, I’m taking these variables out of the equation and am left with a DN value corrected for Linearity, Dark, and Smear only.