Multichannel tif cli

[kbestak]

Hi @Yu-AnChen,

As we discussed quite a while ago, here is the CLI for registration of multiple-cycle multichannel ome tif files.

The main changes are the palom/cli/align_multiple_cycles.py script with an example usecase and parameters added to the README.md. In addition, I've updated the pyproject.toml file accordingly.

Let me know what you think and if this is of interest.

We're in the process of adding Palom in the expansion of nf-core/mcmicro as a parallel option to ASHLAR for users with prestitched images.
In relation, as raised here: #14, one prerequisite for a fully standardized nf-core module is a license, would you mind updating Palom with one?
Second, the standardization would require a Docker image, so either pushing Palom to Biocontainers, or having a CI that pushes a Docker image on the github image repository would be very helpful (there is an old PR still open from Florian Wuennemann addressing this). We would be willing to provide support in pushing this if needed.

All the best

[Yu-AnChen]

Thanks so much! Will review and test this and the #7 early next week.

[kbestak]

#7 is very outdated, I could set up a new github workflow if needed, let me know!

[Yu-AnChen]

When the input image is large in X-Y dimensions, using the second level (level=1) for coarse alignment will likely be inefficient and may not give good results depending on how many key-points were used. Selecting a pyramid level with reasonable size (~1000-2000 pixels in X/Y) would be more appropriate in my experiences. Happy to hear your experiences and thoughts! There's a handy function in the reader object to select a level when given a desired thumbnail image size -

palom/palom/reader.py

Lines 93 to 98 in 33f5a05

	def get_thumbnail_level_of_size(self, size: float) -> int:
	shapes = [
	np.abs(np.mean(level.shape[1:3]) - size)
	for level in self.pyramid
	]
	return np.argmin(shapes)

[Yu-AnChen]

Should check for at least 2 images in img_list

[Yu-AnChen]

Thanks @kbestak - it looks great overall 👍 Just would like to learn a bit more of your cases and experiences. And thanks for the kind offer of the CI to auto-build and push the docker image. Yes, if you have an recent template that you are using, please help me update #7 or start a new PR :)

[kbestak]

Thanks a lot for the feedback (and apologies for the slow response...)!
The reference thumbnail was a good point to raise that I overlooked to reintegrate, so it's applied now, and I've added the img_list check.