Open nickhir opened 1 year ago
Hi @nickhir
Thanks for reporting, this is quite peculiar indeed. Do you mind sharing the sce
object after running fitGAM
, and also let us know what version of tradeSeq
you are using?
Thanks, Koen
Sorry for getting back so late. I talked to the person who generated the data and he is fine with me sharing it. What do you think is the best way to transfer the data? The object is 60MB in size.
Hi @nickhir
Apologies for coming back to this so late. Feel free to send to koen.vdberge at gmail dot com if still relevant, with a description of which genes may be acting funny. You may also want to check out #209.
Hello,
I am using
tradeSeq
downstream ofslingshot
.slingshot
has identified 6 different trajectories, that upon closer inspection seem biologically meaningful. I wanted to usetradeSeq
to identify genes whose expression is associated with the pseudotime of the respective trajectory. Afterwards I used theplotSmoothers
function to visually inspect some of the results. Here I noticed for a couple of genes, that theassociationTest
detected a lot of genes as being associated with pseudotime, while visually it does not look like it. I have posted an extreme example below:Here,
associationTest
indicates, that all lineages have ap.value
of ~ 0, except the lineage 5. Visually it looks like only for lineage 4 there should be an association.This is the code I used to run the analysis:
I would greatly appreciate, if somebody has an idea why I observe such phenomena, or if I misunderstand the test that
tradeSeq
is doing.Cheers!