Loading the stitched-full view in the new spatial transcriptomics project works much better now:
Loading it from filesystem now takes 127028 ms (~ 2 min), so much better than before.
Once loaded browsing of all the spots is very smooth.
Still, it would be interesting to now what the bottlenecks are in loading the view, and if it could be further sped up by improving the table loading (which only takes a couple seconds in python).
Loading the
stitched-full
view in the new spatial transcriptomics project works much better now:Still, it would be interesting to now what the bottlenecks are in loading the view, and if it could be further sped up by improving the table loading (which only takes a couple seconds in python).
If table loading turns out to be a bottleneck now, we could try using parquet (example data with parquet tables is here: https://github.com/mobie/spatial-transcriptomics-example-project/tree/parquet)