Don't fail when KDE bandwidth is 0 #3
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Hi! This is the other error I was running into (the one due to which I originally decided to split the computation to individual genes, which lead me to discover the first reported error).
This one is pretty straightforward: when the data is "too simple", the KDE estimator is failing because the automatically selected bandwidth is zero. These genes are otherwise uninteresting (they get binarized to
NaNanyway asZeroInf; maybeDiscarededwould be even more appropriate here), but it's hard to filter them out beforehand. I originally used a "at least X non-zero samples" criterion to avoid this issue, but here X should ideally depend on the number and variance of observations overall and I couldn't figure out a formula that was reliable enough to always remove all problematic genes.Hence, in this PR, I try to compute the bandwidth (using the same method) before using the KDE estimator itself. If it fails, I print a warning and return
NaN, otherwise everything works as usual. Hopefully this should only impact very low-quality genes anyway. The main reason for fixing this is to avoid failing a large computation when only one or two genes are problematic (and having to search for these genes through other means).(Also: I am branching off of PR #2, since I need the single-gene bug fixed in order for my test to work, but otherwise these should be two separate things)