confusion about the alignment results

[cynthier]

Hi Dinithins,

I am confused about the alignment results. Some gene expression patterns don’t align well with the results. For instance, Klf12 shows higher expression in the reference data, yet the alignment indicates a perfect match. Meanwhile, E2f8 appears mismatched in several early bins rather than the later ones, which is unexpected. Could you help explain this? Only based on the alignment results, it seems hard to determine whether these genes genuinely influence the trajectory.

Klf12:MMMMMMMMMMMMMM

E2f8:IDIDIDMMMMMMMMMMM

Best,
Cynthier

[dinithins]

Hi @cynthier,

Thanks for the question.

For KLF12, even though the average expression is higher in the reference than that of the query, their distributions seem to be still matching (Note: the distributional distance accounts for both mean and variance of expression), hence it is a 100% match along pseudotime with no statistically significant mismatch.

For E2F8, this is actually a challenging instance for the current distributional distance computation in genes2genes due to the extremely low number of cells that express the gene in early pseudotime. It seems to be sensitive to the slightest change in the expression of the outlier cells, hence giving a mismatch for the early time points. I would disregard this early mismatch due to lack of evidence.