Closed carmenpetitjean closed 4 years ago
Hi ,
Thanks for trying STREAM. Indeed many changes including defaulting settings have been introduced since v0.4
. So in order to exactly reproduce what is shown in the tutorial notebook, you might need to use the same version.
Nevertheless, for your current analysis with v0.4.1
, the result actually is not bad except for the monocyte population. Can you try to increase the number of clusters to get a finer initial structure e.g.st.seed_elastic_principal_graph(adata,n_clusters=20)
or increase the number of neighbors, e.g st.dimension_reduction(adata,n_neighbors=200,n_components=4)
? This should help improve the structure learning for the disconnected part.
Increasing the number of clusters and the number of neighbours seemed to have improved analysis! Thank you!
Hi,
I am currently trying to follow the scATAC-seq using k-mers tutorial but have encountered difficulties reproducing your results (even using the same code), particularly for the dimension reduction plots and subway maps. I was wondering if these issues could be linked to the difference in STREAM versions used (your tutorial using v0.3.7 and I using v0.4.1) or if this could be due to any other reasons.
I've attached the plots that I have produced following the tutorial.
Thank you!