Closed HelloWorldLTY closed 1 year ago
Thank you so much for raising the issues!
One probably needs to first install the "pbmcsca" dataset using InstallData("pbmcsca")
, before loading the data - sorry about that! Corrections have been made in the Tutorial.
The error "prop.align too large!" indicates insufficient size for the maximal correspondence subsets identified in the previous step, which can be fixed by increasing the parameter k1
in function findMutualNN()
. To improve automation in this step, we have updated the corresponding codes in our Tutorial - a suitable k1
is now determined automatically, without user specification.
For our partial alignability test, we incorporate a sample splitting procedure to ensure its statistical validity, which introduces randomness to our test result. However, one may benefit from such randomness by repeating the test under various random seeds, to assess the robustness/stability of the test results with respect to resampling.
You are right, the final integrated datasets are returned regardless of the significance of the p-value. Our test essentially indicates the goodness-of-fit of the data generative model motivating the proposed alignment procedure, but does not necessarily affect the execution of the alignment algorithm. However, we do recommend not using the integrated data if a significant p-value is returned.
Please let me know if you have further questions/comments! Thank a lot!
Thanks for your reply! I will try your method based on the updated version again.
Of course! Please let me know if you have further questions!
On Aug 18, 2023, at 12:04 PM, HelloWorldLTY @.***> wrote:
Thanks for your reply! I will try your method based on the updated version again.
— Reply to this email directly, view it on GitHub https://github.com/rongstat/SMAI/issues/1#issuecomment-1684127402, or unsubscribe https://github.com/notifications/unsubscribe-auth/AWOXT4RBMHVZZUTXSYFEAMLXV6G7FANCNFSM6AAAAAA3U46ZQY. You are receiving this because you modified the open/close state.
Thanks a lot.
Hi, I tried SMAI to test if two datasets are alignable or not, and had a series of findings.
The first thing I intend to share are two bugs. One for example dataset: I have installed seuratDisk
Another one is for my own dataset:
Could you please help me address them? Thanks a lot.
Moreover, another question is from the algorithm 1. I found that different seeds generated different p values. I think this algorithm does not contrain permutation, so there should be no randomness in p-value.
And if we found that two datasets are not alignable, why this algorithm still "aligns" them?
Thanks a lot.