OpenCell reader extension

[ieivanov]

The OpenCell acquisition starts with a large number of positions in the position list, but only some of those are actually imaged. When we save the data, we need to use a continuous position index starting at 0 because of a Micro-manager bug: micro-manager/micro-manager#1723, i.e. if we are imaging positions (5, 6, 13, 12) in the position list, they are saved with index (0, 1, 2, 3). In this case, we need to parse the correct position label from the tif filename.

I've implemented this as a special version of Micro-manager metadata. If we change the MM version on the Dragonfly microscope (unlikely) or the "raw_data" prefix (also very unlikely) this code will not work an will need to be modified. I've also switched the order of _gather_index_maps() and _set_mm_meta() to make this work. I didn't see unintended consequences here, but I think this may be making some of the tests fail. @ziw-liu can you help me take a look?

cc: @Soorya19Pradeep

[Soorya19Pradeep]

@ieivanov , thank you so much for implementing the reader. The iohub conversion works for both opencell data as well MM data now. I have tested it on data from DragonFly OpenCell imaging and MM HCS data.

[ziw-liu]

I've also switched the order of _gather_index_maps() and _set_mm_meta() to make this work. I didn't see unintended consequences here, but I think this may be making some of the tests fail. @ziw-liu can you help me take a look?

The _set_mm_meta() method sets self.frames based on MM summary metadata, which may be inaccurate for incomplete datasets, and _gather_index_maps() will overwrite the initial value based on the index map.

Also this line breaks single page TIFF conversion because it has an empty reader.stage_positions:

                name = self.reader.stage_positions[p].get("Label") or p

[ieivanov]

Thanks for catching this @ziw-liu. I commented out these lines so they don't overwrite the dimensions set by _gather_index_maps

[ziw-liu]

Code looks good! (I haven't tested the added functionality.)

It would be nice if this can be tested automatically somehow.

[ieivanov]

Thanks @ziw-liu. I opened a new issue for adding a test dataset