We are working on a some 3D reconstruction of big protein from the negative staining data. Our particles and 2D classes are also nice and sharp looking. If we keep the default setting (on) of the parameter “ignore the dc component” during Ab-initio job, we see only the stain and noise. While if we turn this parameter off, we reconstruct the the meaningful volume map. Therefore our question is how bad is it to keep DC during our processing negative data? Maybe you can give us an idea what the range of zero frequencies is ignoring by default?
Not sure what you mean by range of zero frequencies? The DC component is the one sample at zero frequency, it’s just the average brightness of the whole image. Maybe your particles include some with wildly different contrasts and including the DC component (so cross-correlation becomes sensitive to brightness) is some of those to effectively be downweighted. When you run 2D classification are there any obvious things to exclude?
Dear @DanielAsarnow, thank you so much for the reply! Yes, we were totally wrong about the range. I think you are right about wildly different contrasts, since the dye is a mix of different heavy metal salts.
Yes, we have some nice 2D classes and some obvious garbage.
Then maybe you know why is it important to ignore DC during Ab-initio and if it is acceptable to keep in our case if it is indeed wildly different contrasts?