mehta-lab
Fix the PC Viewer to support multi-dataset PHATE embedding
This PR cleans up the PC viewer to support multi-dataset view of PHATE and PC.
Here we ad a configurable .yml file to feed the FOV, tracking, and embeddings information. Additionally, we support configurable PHATE parameters and number of PCs.
Thank you @edyoshikun! I am getting the following error while using the script:
Error processing batch for /B/2/000000, track [65]: unhashable type: 'list' Error processing batch for /B/2/000000, track [65]: unhashable type: 'list' Error processing batch for /B/2/000000, track [65]: unhashable type: 'list' Error processing batch for /B/2/000000, track [65]: unhashable type: 'list' Processing FOV /C/2/000001 with tracks all ['/C/2/000001', '/C/2/000001', '/C/2/000001'] all Successfully cached 0 images Cached datasets: set() Cached dataset-FOV combinations: 0 Number of unique dataset-FOV-track combinations: 0 Saving image cache to /hpc/mydata/soorya.pradeep/scratch/cache/image_cache.pkl Successfully saved cache with 0 images
@Soorya19Pradeep was the freezing you observed happening on this branch?
What freezing are you refering to?
@Soorya19Pradeep was the freezing you observed happening on this branch?
What freezing are you refering to?
You reported that the viewer would freeze for the datasets your were working with.
@Soorya19Pradeep seems like this is now working. Can we close and merge this PR to main?
@
@Soorya19Pradeep was the freezing you observed happening on this branch?
What freezing are you refering to?
You reported that the viewer would freeze for the datasets your were working with.
I am using max projected data now and a smaller number of FOVs. It works well for this case, except it is slightly slow to respond sometimes. But it is workable.
@Soorya19Pradeep seems like this is now working. Can we close and merge this PR to main?
Thanks @edyoshikun . I currently just have one issue with the widget which has a very easy workaround. The plotly widget stores information in my web browser, which leads to it breaking if I run the same code again. For instance, when I use the lasso tool to annotate clusters once in a specific PC space, then I close down the port. Then rerun the code and try lasso again, it would not do a great job and shows a combination of something I am trying to do and from the past annotations. It is not very reproducible. I read that this is a feature from Plotly, and I can open a new browser window to fix this. I was just unsure if there is a workaround for that through the code.