changing color palette for pl.tsne

Hello! I’m trying to recolor some categorical variables in the scanpy.api.pl.tsne function but am having some trouble. Specifically, with continuous data, I’m fine using the color_map key word to change between scales like “viridis” and “Purples” but when trying to pass the palette key word for categorical data (sample labels, louvain lables), it doesn’t seem to update the colors in the plot. Perhaps I’m specifying the color palette incorrectly? Here are a few versions I’ve tried:

sc.pl.tsne(adata, 
           color=['louvain'], 
           #palette=['C0', 'C1', 'C2', 'C3', 'C4', 'C5', 'C6', 'C7', 'C8', 'C9'], 
           #palette=['tab:blue', 'tab:orange', 'tab:green', 'tab:red', 'tab:purple', 'tab:brown', 'tab:pink', 'tab:gray', 'tab:olive', 'tab:cyan'],
           #palette="Set3",
           palette=sns.color_palette("hls", 15),
           legend_fontsize="20")

But all of these attempts result in the same default cluster color scheme. Any suggestions?

Thanks!

[LuckyMD]

If you take a look at adata.uns['louvain_colors'] you will see this is an ndarray containing hex codes for the colours. So you can change colours manually by e.g., adata.uns['louvain_colors'][0] = '#ffff00'. The array is created the first time you call sc.pl.umap(adata, color='louvain') or any other plotting function with color='louvain'.

[elfore]

You can always choose a palTete like ‘Blues’, ‘Reds’, ‘binary’ that will give you a gradient from a clear to a darker color. Maybe that helps but I agree with Philipp, 120 clusters is a lot to visualize with different colors. … On Tue, May 28, 2019 at 4:21 PM Philipp A. @.***> wrote: Closed #156 <#156>. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#156?email_source=notifications&email_token=ABF37VNHLRL6TEP3I7BBLEDPXU5X7A5CNFSM4FAIXFAKYY3PNVWWK3TUL52HS4DFWZEXG43VMVCXMZLOORHG65DJMZUWGYLUNFXW5KTDN5WW2ZLOORPWSZGORVQ5OJI#event-2371999525>, or mute the thread https://github.com/notifications/unsubscribe-auth/ABF37VNDXU3HBFBIZ6LII73PXU5X7ANCNFSM4FAIXFAA . – Fidel Ramirez

Thanks for your great idea, I will give it a try.