klarman-cell-observatory / scplot Goto Github PK

Adapt scvelo plot
to show animated transitions, similar to http://hint.fm/wind/

Categorical types get mixed up in violin plots. It's also visible in the notebook:

It doesn't happen in composition plots. I suspect it's because violin calls __fix_color_by_data_type and composition does not. Conversion to string in __fix_color_by_data_type might lead to the mixing.

This is also related to the PR #3 . I think if you outsource color assignment to scanpy, everything will be easier and cleaner on the scplot side.

This would be super cool. I saw that it's missing in hvplot (holoviz/hvplot#225) but how about using plotly for this? I know it wouldn't be visually consistent with bokeh but better than nothing, no?

import chart_studio.plotly as py
import plotly.express as px

import scanpy as sc
import pandas as pd

adata = sc.datasets.paul15()
sc.pp.normalize_total(adata, target_sum=10000)
sc.pp.log1p(adata)
sc.pp.neighbors(adata)
sc.tl.umap(adata, n_components=3)

df = pd.DataFrame(adata.obsm['X_umap'], columns=['x', 'y', 'z'])
df['gene'] = adata[:, 'Gata1'].X
fig = px.scatter_3d(df, x='x', y='y', z='z', color='gene', width=1600, height=1000)
py.iplot(fig) # more options here https://plot.ly/python/3d-scatter-plots/