I switched it to the command line version of sourcetracker2-2.0.1-dev to solve "ipcluster remains open" error in sourcetracker2-2.0.1. As I found the issue #114 here, I tested sourcetracker2-2.0.1 and sourcetracker2-2.0.1-dev (mean and sum for collapsing source samples) on my data.
The mixing proportion results are kind of similar among all these three approaches, however, the standard deviation results are different between sourcetracker2-2.0.1 (at the level of 1e-3) and sourcetracker2-2.0.1-dev (at the level of 1e-4). I wonder whether there is any change between these two version of sourcetracker2 in terms of standard deviation calculation.
Panda-smile
commented
8 months ago
Hi all,
I switched it to the command line version of sourcetracker2-2.0.1-dev to solve "ipcluster remains open" error in sourcetracker2-2.0.1. As I found the issue #114 here, I tested sourcetracker2-2.0.1 and sourcetracker2-2.0.1-dev (mean and sum for collapsing source samples) on my data.
The mixing proportion results are kind of similar among all these three approaches, however, the standard deviation results are different between sourcetracker2-2.0.1 (at the level of 1e-3) and sourcetracker2-2.0.1-dev (at the level of 1e-4). I wonder whether there is any change between these two version of sourcetracker2 in terms of standard deviation calculation.
Thanks!
Dot