单细胞｜RNA速率下游 · Dynamo

数据集：cellDancer文件 PancreaticEndocrinogenesis_cellDancer_estimation.csv

import random
import numpy as np
import pandas as pd
import numpy as np
import scipy
import warnings
warnings.filterwarnings('ignore')
import dynamo as dyn
from celldancer.utilities import export_velocity_to_dynamo

1.将cellDancer的预测输入dynamo

# First create dynamo adata
adata_dyn = dyn.sample_data.pancreatic_endocrinogenesis()
pancreas_genes = ["Hes1","Nkx6-1","Nkx2-2","Neurog3","Neurod1","Pax4","Pax6",
"Arx","Pdx1","Ins1","Ins2","Ghrl","Ptf1a","Iapp","Isl1", "Sox9","Gcg"]
dyn.pp.recipe_monocle(adata_dyn, n_top_genes=4000, 
                      fg_kwargs={"shared_count": 20}, 
                      genes_to_append=pancreas_genes)
dyn.tl.dynamics(adata_dyn, model='static')

# replace adata.layers['velocity_S'], adata.var['use_for_dynamics'], and adata.var['use_for_transition'] with cellDancer outputs.
cellDancer_df = pd.read_csv('your_path/PancreaticEndocrinogenesis_cellDancer_estimation.csv')
adata = export_velocity_to_dynamo(cellDancer_df,adata_dyn)

2.将 RNA 速率投影到嵌入空间上

pancreas_cluster_cmap = {'Ductal': '#3361A5','Ngn3 low EP': '#248AF3',
                         'Ngn3 high EP': '#14B3FF',
                         'Pre-endocrine': '#88CEEF','Alpha': '#ff4800',
                         'Beta': '#B81136','Delta': 'green',
                         'Epsilon': '#03B3B0'}

dyn.tl.reduceDimension(adata, n_pca_components=30)
dyn.tl.cell_velocities(adata, method="pearson", 
                       other_kernels_dict={"transform": "sqrt"})
dyn.tl.cell_velocities(adata, basis="pca")

dyn.pl.streamline_plot(adata, color=["clusters"], basis="umap", 
                       color_key = pancreas_cluster_cmap, 
                       show_legend="on data", 
                       show_arrowed_spines=True)

 

3.学习和可视化 UMAP 上的矢量场

dyn.vf.VectorField(adata, basis="pca", pot_curl_div=True)
dyn.vf.VectorField(adata, basis="umap", pot_curl_div=True)

dyn.vf.speed(adata, basis="pca")
dyn.vf.divergence(adata, basis="pca")
dyn.vf.acceleration(adata, basis="pca")
dyn.vf.curl(adata, basis="umap")

dyn.pl.topography(adata, color=['clusters'], basis='umap', 
                  background='white',
                  streamline_color='black', 
                  color_key = pancreas_cluster_cmap, 
                  show_legend='on data', terms=("streamline"))

 

adata_load_fix_points=adata.copy()
Xss, ftype, conf = adata_load_fix_points.uns['VecFld_umap']['Xss'],\
                   adata_load_fix_points.uns['VecFld_umap']['ftype'],\
                   adata_load_fix_points.uns['VecFld_umap']['confidence']

fixed_points = [10,7,16,12]

adata_load_fix_points.uns['VecFld_umap']['Xss'] = Xss[fixed_points]
adata_load_fix_points.uns['VecFld_umap']['ftype'] = ftype[fixed_points]
adata_load_fix_points.uns['VecFld_umap']['confidence']=conf[fixed_points]

dyn.pl.topography(
    adata_load_fix_points,
    markersize=500,
    basis="umap",
    fps_basis="umap",
    color_key = pancreas_cluster_cmap,
    color=["clusters"],
    show_legend='on data',
    save_show_or_return='show'
)

 

4.Jacobian分析检测基因调控

dyn.vf.jacobian(adata, regulators=['Arx','Pax4'])

dyn.pl.jacobian(
    adata,
    basis="umap",
    regulators=['Arx'],
    effectors=['Pax4'],
    alpha=1,
)

dyn.pl.jacobian(
    adata,
    basis="umap",
    regulators=['Pax4'],
    effectors=['Arx'],
    alpha=1,
)

 

在UMAP上绘制特定基因的表达

dyn.pl.umap(adata,  color=["Arx",'Pax4'],save_show_or_return='show')

 

参考：Downstream analysis using Dynamo — cellDancer documentation (guangyuwanglab2021.github.io)
Cell, 2022, 185(4):690-711.e45.