How to avoid influence from other part of complex

Greeting there,

I’m working on a large sub-complex which is a part of a pretty large complex that is purified from cell directly.

I used box sizes just capture the largest dimension of my sub-complex when I extract from micrographs after manual picking (definitely will sacrifice high resolution signal), instead of using box sizes that double the diameter of sub-complex. Because most of the left part of the large complex is highly repetitive, which will dominate the 2D classification. Still the signal of my sub-complex is overlap with the left part the large complex in most of views and the 2D classification is biased.

So I’m seeking advice here to see anyone has good solution. 1, My plan is that extract from micrographs using even smaller box sizes to see whether I can avoid the bias from the highly repetitive part, re-extract from micrographs using double sizes of the diameter of sub-complex after select 2D classes. But I’m not sure whether it works. 2, Will it also influence my reconstruction? I’m thinking about to use a low resolution reference for reconstruction, is that a good idea? And I didn’t find the option that reconstruction with reference.

Thanks and cheers,

Yanhe