Strategies for large-scale integration

[johnlee330]

Hey all, anyone have any recommendations regarding large-scale integration of spatial data? Suppose an order of ~100 spatial samples, would there be any feasible way to get them into the same integrated space?

Running the standard integration pipelines would crash out a session. Woudl probably need to utilize memory-efficient tools like sketch or BPcells

Would it be feasible doing sketch clustering, integrate on the sketch, project the sketch onto the full dataset?