Issue with mapping step

[jipeifeng]

Hi: I have followed the pipeline to mapped another group of cell to the reference. Then, the result was so strange. (I have generated the reference using 'umap' to reduce dimension, works pretty well) The stream plot shows:

Then I mapped the other group of cells to the reference : 1. ko = sc.read_h5ad('data/ko.h5ad') 2. stream.set_workdir(ko, './stream') 3. stream.remove_mt_genes(ko) 4. ko.obs.rename(columns={'type':'label'}, inplace=True) 5. stream.add_cell_colors(ko) 6. sc.pp.subsample(ko, n_obs=10000) 7. stream.map_new_data(wt, ko, method='umap') 8. stream.plot_visualization_2D(wt,adata_new=ko, use_precomputed=False) 9. stream.plot_visualization_2D(wt,adata_new=ko,show_all_colors=True)

I got the results: stream.stream_plot(wt, adata_new=ko, root='S5')

Plot ko alone:

So, it is weird, can you please figure out the problem? Looking forward to your response, Thank you !

Peifeng

[jipeifeng]

I have also plotted the mapped cell separately following the same pipeline of generating the reference. Then , I got the result.

[huidongchen]

Hi,

sorry about the late response. The mapping result indeed doesn't look correct. For your KO, did you normalize your gene expression matrix before feeding them to the mapping procedure?