Yes, color only works on 1x1 binning. What software do you use?
Depending on your processing software, it goes something like this:
1. Calibrate all images (darks and flats, using color master darks and color master flats).
2. Convert Bayer (convert to color, or convert to individual color frames and process them separately as R, G, and B, your choice). Your software may, or may not, show the images as color at this point.
3. Register the stack (either color, or register the R, G, and B's as separate stacks).
4. Normalize each stack
5. data/noise rejection algorithm on the stack(s).
6. Combine the stack(s), using mean, median, sigma, or whatever you prefer. Choice depends on your data quality, number of frames.
7. Save frame(s).
8. If you separated R,G,B, combine to create color image.
9. Post processing: sharpen, deconvolution, DDP, color adjustments, etc., depending on your software packages and preferences.
I used to process my ST2000XCM pictures by separating the Bayer matrix into separate R,G,and B, stacks, and sometimes creating an artificial luminance to process for detail. That's all more work, though, and it's probably best to start by processing full color frames.