Closed Marius1311 closed 1 year ago
Check out this pull request on
See visual diffs & provide feedback on Jupyter Notebooks.
Powered by ReviewNB
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:06Z ----------------------------------------------------------------
Please capitalize the title
Also would do real -> experimental
Marius1311 commented on 2023-06-13T11:26:26Z ----------------------------------------------------------------
Re capitalization: I would have to do that everywhere, will do that in a separate PR if required.
Marius1311 commented on 2023-06-13T11:26:41Z ----------------------------------------------------------------
Changed to "experimental"
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:07Z ----------------------------------------------------------------
This misses some introductory paragraph above.
Marius1311 commented on 2023-06-13T11:28:37Z ----------------------------------------------------------------
mh, ok, after the last review, I thought you wanted to have this message all the way to the top. Again, I would have to change this for all tutorials. Can be done on a separate PR.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:08Z ----------------------------------------------------------------
Use ~ for the TemporalProblem
Remove "CellRank's"
Marius1311 commented on 2023-06-13T11:31:59Z ----------------------------------------------------------------
ok.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:09Z ----------------------------------------------------------------
{mod}moscot
Cite moslin instead of the link to the repo?
Marius1311 commented on 2023-06-13T11:36:33Z ----------------------------------------------------------------
ok, fixed both.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:10Z ----------------------------------------------------------------
Would replace ... -> :
No need for -
in spatial-
"an scRNA-seq"
Can improve the accuracy instead?
Marius1311 commented on 2023-06-13T13:23:34Z ----------------------------------------------------------------
We know it improves the accuracy ;)
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:11Z ----------------------------------------------------------------
This points to master, should be main
. For consistency, I'd include this under the 1st paragraph.
Marius1311 commented on 2023-06-13T13:24:32Z ----------------------------------------------------------------
ok.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:11Z ----------------------------------------------------------------
Use subset_to_serum=True
.
Marius1311 commented on 2023-06-13T13:30:30Z ----------------------------------------------------------------
That gives https://github.com/theislab/cellrank/issues/1053
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:12Z ----------------------------------------------------------------
Line #4. # Experimental time points are saved in adata.obs["day"]
as categoricals. Let's convert the categories to be numerical, this is required by moscot later on.
Please shorten the comment, it's overwhelmingly long.
ok.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:14Z ----------------------------------------------------------------
Line #5. adata.obs["day"] = adata.obs["day"].cat.rename_categories({day: float(day) for day in adata.obs["day"].cat.categories})
Use
adata.obs["day"].astype(float).astype("category")
Marius1311 commented on 2023-06-13T14:07:10Z ----------------------------------------------------------------
done.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:15Z ----------------------------------------------------------------
Line #8. adata.obs["day_numerical"] = adata.obs["day"].astype(float)
This shouldn't be needed.
Looks nicer - the color legend is continuous rather than discrete.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:16Z ----------------------------------------------------------------
Line #10. # Subsample to speed up the analysis - this tutorial is meant to run in a couple of minutes on a laptop. It’s not a problem for CellRank to do any of the computations here on the full data, we’d just have to wait a bit longer.
Please shorten the the comment or add as a normal text.
done.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:16Z ----------------------------------------------------------------
Visualize the embedding
Marius1311 commented on 2023-06-13T13:34:02Z ----------------------------------------------------------------
ok.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:17Z ----------------------------------------------------------------
I don't think link to FLE or the citation is needed.
Marius1311 commented on 2023-06-13T13:34:31Z ----------------------------------------------------------------
removed the link.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:18Z ----------------------------------------------------------------
The link to neighbors doesn't work.
Marius1311 commented on 2023-06-13T13:35:13Z ----------------------------------------------------------------
I think I fixed it, should be "scanpy" rather than "sc".
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:19Z ----------------------------------------------------------------
"Waddington OT" -> "Waddington-OT" (as stylized in their publication)
Use {mod}moscot
instead of the hardcoded link.
No -
in uni-modal
Do not capitalize Transport
Marius1311 commented on 2023-06-13T13:36:37Z ----------------------------------------------------------------
ok.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:20Z ----------------------------------------------------------------
No -
in multi-modal
Please add nicer link to moscot, e.g.: "documentation <https://...>_
" .
Rather than linking to moslin
, I think you should link to the moscot tutorial that uses the LineageProblem.
Please add a link in "... inculding many tutorials" to moscot's tutorials.
Marius1311 commented on 2023-06-13T13:40:45Z ----------------------------------------------------------------
all done.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:21Z ----------------------------------------------------------------
Add 3 ~
you can use the ... or the LineageProblem, respectively.
Marius1311 commented on 2023-06-13T13:41:49Z ----------------------------------------------------------------
done.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:22Z ----------------------------------------------------------------
Visualilze -> Visualize
Marius1311 commented on 2023-06-13T13:42:13Z ----------------------------------------------------------------
done.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:23Z ----------------------------------------------------------------
local PCAs, that are computed separately
"comptes" -> "computes"
we'll use -> we use
Marius1311 commented on 2023-06-13T13:43:28Z ----------------------------------------------------------------
done.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:24Z ----------------------------------------------------------------
"Early-" -> "early"
Marius1311 commented on 2023-06-13T13:43:50Z ----------------------------------------------------------------
ok.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:24Z ----------------------------------------------------------------
"at the left marginal" -> "on the source marginal"
"regularitatition" -> "regularization"
Higher entropic regularization speeds up
Marius1311 commented on 2023-06-13T13:45:31Z ----------------------------------------------------------------
ok.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:25Z ----------------------------------------------------------------
"to a Markov chain description" -> "to a Markov chain describing a biological sytem, we do the following:"
"arrive at a Markov-chain description" -> "and construct the Markov chain"
I'd add more information how the sparsification is done.
Marius1311 commented on 2023-06-13T13:48:05Z ----------------------------------------------------------------
ok, added a half sentence on sparsification.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:26Z ----------------------------------------------------------------
Visualize the recovered dynamics
Marius1311 commented on 2023-06-13T13:48:34Z ----------------------------------------------------------------
ok.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:27Z ----------------------------------------------------------------
Would do: by sampling {meth}random walks <cellrank.kernels.TransportMapKernel.plot_random_walks>
.
Marius1311 commented on 2023-06-13T13:49:49Z ----------------------------------------------------------------
ok, good suggestion.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:28Z ----------------------------------------------------------------
Please add a new paragraph (within the same cell) starting with: "Another way to visualize..."
Marius1311 commented on 2023-06-13T13:55:55Z ----------------------------------------------------------------
ok.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:29Z ----------------------------------------------------------------
Can we please remove the vertical bars?
Marius1311 commented on 2023-06-13T13:56:33Z ----------------------------------------------------------------
I'm afraid I don't know how.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:30Z ----------------------------------------------------------------
Line #12. locs, labels = plt.xticks()
Can you try getting the ticks from ax
?it will save us an import.
good point, found a solution.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:31Z ----------------------------------------------------------------
Line #16. plt.show()
Not need for this.
done.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:32Z ----------------------------------------------------------------
This is not really needed here, would remove.
Marius1311 commented on 2023-06-13T14:03:32Z ----------------------------------------------------------------
ok.
View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:15:33Z ----------------------------------------------------------------
Too many "to"s in "to learn how to use the transition matrix to"
"we recommend..." -> "we recommend to:" + "go through"/"take a look"/"read the Waddington-OT"
"parameter values" -> "parameters"
Would add the citations for WOT/moscot when you mention them, not at the end + use {mod}moscot
.
Marius1311 commented on 2023-06-13T14:06:25Z ----------------------------------------------------------------
ok, done almost all.
@Marius1311 generally, would prefer these 2 things:
100_initial_terminal.ipynb
or 200_experimental_time.ipynb
so that globs will sort it by their name100_
, so that we know to which tutorials they belong.View / edit / reply to this conversation on ReviewNB
michalk8 commented on 2023-06-13T00:23:17Z ----------------------------------------------------------------
All the headings, except the 1st one (CellRank meets real time) are too high (should be lower than the 1st one).
Marius1311 commented on 2023-06-13T11:31:04Z ----------------------------------------------------------------
ok, changed that.
Re capitalization: I would have to do that everywhere, will do that in a separate PR if required.
View entire conversation on ReviewNB
mh, ok, after the last review, I thought you wanted to have this message all the way to the top. Again, I would have to change this for all tutorials. Can be done on a separate PR.
View entire conversation on ReviewNB
Looks nicer - the color legend is continuous rather than discrete.
View entire conversation on ReviewNB
Adds a tutorial showcasing how CellRank interfaced with moscot to analyze time-course data.