Open SteveTur opened 6 months ago
my0916
commented
6 months ago Hi,
I'm also suffering from a similar problem... https://github.com/aertslab/scenicplus/discussions/378#discussioncomment-9314238
I also tried to export to Cytoscape with the export_to_cytoscape function, but the exported cys file cannot be loaded with the message 'Cannot find the cysession.xml file'
Thank you for your help.
Best, Masa
yiyelinfeng
commented
6 months ago I have same problem, please help!
SeppeDeWinter
commented
6 months ago Hi @SteveTur, @my0916 and @yiyelinfeng
I am aware of the issue. I have added it to my to do list and will try to fix it whenever I have some time.
All the best,
Seppe
yiyelinfeng
commented
6 months ago Hi@ SeppeDeWinter,
my issue only just the .cys file cannot be loaded in cytoscape with the message 'Cannot find the cysession.xml file' when file -> import -> Network from file ... after import the SNENIC+ network layout. Because I want to check and change the network graph using Cytoscape. but can't load it.
thanks,
Best, Lin
DmitriiSeverinov
commented
5 months ago Hi @SeppeDeWinter , @yiyelinfeng @my0916 , @SteveTur ,
I had similar issue with blank output, what I found is that if I comment the following line in the create_nx_tables function, it works:
subset_eRegulons = [x + '_[^a-zA-Z0-9]' for x in subset_eRegulons]
So, here is my code how I did it:
from pycisTopic.diff_features import find_highly_variable_features
scplus_obj.uns['direct_e_regulon_metadata_filtered'] = scplus_mdata.uns['direct_e_regulon_metadata_filtered']
hvr = find_highly_variable_features(scplus_obj.to_df('ACC').loc[list(set(scplus_obj.uns['direct_e_regulon_metadata_filtered']['Region']))], n_top_features=3000, plot = False)
hvg = find_highly_variable_features(scplus_obj.to_df('EXP')[list(set(scplus_obj.uns['direct_e_regulon_metadata_filtered']['Gene']))].T, n_top_features=3000, plot = False)
from scenicplus.networks import create_nx_tables, create_nx_graph, plot_networkx, export_to_cytoscape
nx_tables = list()
def _format_df_nx(df, key, var):
"""
A helper function to format differential test results
"""
df.index = df['names']
df = pd.DataFrame(df['logfoldchanges'])
df.columns = [var+'_Log2FC_'+key]
df.index.name = None
return df
def _get_log2fc_nx(scplus_obj: 'SCENICPLUS',
variable,
features,
contrast: Optional[str] = 'gene'
):
"""
A helper function to derive log2fc changes
"""
if contrast == 'gene':
adata = anndata.AnnData(X=scplus_obj.X_EXP, obs=pd.DataFrame(
index=scplus_obj.cell_names), var=pd.DataFrame(index=scplus_obj.gene_names))
if contrast == 'region':
adata = anndata.AnnData(X=scplus_obj.X_ACC.T, obs=pd.DataFrame(
index=scplus_obj.cell_names), var=pd.DataFrame(index=scplus_obj.region_names))
adata.obs = pd.DataFrame(scplus_obj.metadata_cell[variable])
sc.pp.normalize_total(adata, target_sum=1e4)
sc.pp.log1p(adata)
adata = adata[:, features]
sc.tl.rank_genes_groups(
adata, variable, method='wilcoxon', corr_method='bonferroni')
groups = adata.uns['rank_genes_groups']['names'].dtype.names
diff_list = [_format_df_nx(sc.get.rank_genes_groups_df(
adata, group=group), group, variable) for group in groups]
return pd.concat(diff_list, axis=1)
def create_nx_tables(scplus_obj: 'SCENICPLUS',
eRegulon_metadata_key: str ='eRegulon_metadata',
subset_eRegulons: List = None,
subset_regions: List = None,
subset_genes: List = None,
add_differential_gene_expression: bool = False,
add_differential_region_accessibility: bool = False,
differential_variable: List =[]):
"""
A function to format eRegulon data into tables for plotting eGRNs.
Parameters
---------
scplus_obj: SCENICPLUS
A SCENICPLUS object with eRegulons
eRegulon_metadata_key: str, optional
Key where the eRegulon metadata dataframe is stored
subset_eRegulons: list, optional
List of eRegulons to subset
subset_regions: list, optional
List of regions to subset
subset_genes: list, optional
List of genes to subset
add_differential_gene_expression: bool, optional
Whether to calculate differential gene expression logFC for a given variable
add_differential_region_accessibility: bool, optional
Whether to calculate differential region accessibility logFC for a given variable
differential_variable: list, optional
Variable to calculate differential gene expression or region accessibility.
Return
---------
A dictionary with edge feature tables ('TF2G', 'TF2R', 'R2G') and node feature tables ('TF', 'Gene', 'Region')
"""
er_metadata = scplus_obj.uns[eRegulon_metadata_key].copy()
if subset_eRegulons is not None:
# subset_eRegulons = [x + '_[^a-zA-Z0-9]' for x in subset_eRegulons]
er_metadata = er_metadata[er_metadata['Region_signature_name'].str.contains(
'|'.join(subset_eRegulons))]
if subset_regions is not None:
er_metadata = er_metadata[er_metadata['Region'].isin(subset_regions)]
if subset_genes is not None:
er_metadata = er_metadata[er_metadata['Gene'].isin(subset_genes)]
nx_tables = {}
nx_tables['Edge'] = {}
nx_tables['Node'] = {}
# Generate edge tables
r2g_columns = [x for x in er_metadata.columns if 'R2G' in x]
tf2g_columns = [x for x in er_metadata.columns if 'TF2G' in x]
nx_tables['Edge']['TF2R'] = er_metadata[er_metadata.columns.difference(
r2g_columns + tf2g_columns)].drop('Gene', axis=1).drop_duplicates()
nx_tables['Edge']['TF2R'] = nx_tables['Edge']['TF2R'][['TF', 'Region'] +
nx_tables['Edge']['TF2R'].columns.difference(['TF', 'Region']).tolist()]
nx_tables['Edge']['R2G'] = er_metadata[er_metadata.columns.difference(
tf2g_columns)].drop('TF', axis=1).drop_duplicates()
nx_tables['Edge']['R2G'] = nx_tables['Edge']['R2G'][['Region', 'Gene'] +
nx_tables['Edge']['R2G'].columns.difference(['Region', 'Gene']).tolist()]
nx_tables['Edge']['TF2G'] = er_metadata[er_metadata.columns.difference(
r2g_columns)].drop('Region', axis=1).drop_duplicates()
nx_tables['Edge']['TF2G'] = nx_tables['Edge']['TF2G'][['TF', 'Gene'] +
nx_tables['Edge']['TF2G'].columns.difference(['TF', 'Gene']).tolist()]
# Generate node tables
tfs = list(set(er_metadata['TF']))
nx_tables['Node']['TF'] = pd.DataFrame(
'TF', index=tfs, columns=['Node_type'])
nx_tables['Node']['TF']['TF'] = tfs
genes = list(set(er_metadata['Gene']))
genes = [x for x in genes if x not in tfs]
nx_tables['Node']['Gene'] = pd.DataFrame(
'Gene', index=genes, columns=['Node_type'])
nx_tables['Node']['Gene']['Gene'] = genes
regions = list(set(er_metadata['Region']))
nx_tables['Node']['Region'] = pd.DataFrame(
'Region', index=regions, columns=['Node_type'])
nx_tables['Node']['Region']['Region'] = regions
# Add gene logFC
if add_differential_gene_expression is True:
for var in differential_variable:
nx_tables['Node']['TF'] = pd.concat([nx_tables['Node']['TF'], _get_log2fc_nx(
scplus_obj, var, nx_tables['Node']['TF'].index.tolist(), contrast='gene')], axis=1)
nx_tables['Node']['Gene'] = pd.concat([nx_tables['Node']['Gene'], _get_log2fc_nx(
scplus_obj, var, nx_tables['Node']['Gene'].index.tolist(), contrast='gene')], axis=1)
if add_differential_region_accessibility is True:
for var in differential_variable:
nx_tables['Node']['Region'] = pd.concat([nx_tables['Node']['Region'], _get_log2fc_nx(
scplus_obj, var, nx_tables['Node']['Region'].index.tolist(), contrast='region')], axis=1)
return nx_tables
nx_table = create_nx_tables(
scplus_obj = scplus_obj,
eRegulon_metadata_key ='direct_e_regulon_metadata_filtered',
subset_eRegulons = ['atf3', 'bach2b'],
subset_regions = hvr,
subset_genes = hvg,
add_differential_gene_expression = True,
add_differential_region_accessibility = True,
differential_variable = ['ident']
)
G, pos, edge_tables, node_tables = create_nx_graph(nx_table,
use_edge_tables = ['TF2R','R2G'],
color_edge_by = {'TF2R': {'variable' : 'TF', 'category_color' : {'atf3': 'Red', 'bach2b': 'Blue'}},
'R2G': {'variable' : 'importance_x_rho', 'continuous_color' : 'viridis', 'v_min': -1, 'v_max': 1}},
transparency_edge_by = {'R2G': {'variable' : 'importance_R2G', 'min_alpha': 0.1, 'v_min': 0}},
width_edge_by = {'R2G': {'variable' : 'importance_R2G', 'max_size' : 1.5, 'min_size' : 1}},
color_node_by = {'TF': {'variable': 'TF', 'category_color' : {'atf3': 'Red', 'bach2b': 'Blue'}},
'Gene': {'variable': 'ident_Log2FC_RSNs', 'continuous_color' : 'bwr'},
'Region': {'variable': 'ident_Log2FC_RSNs', 'continuous_color' : 'viridis'}},
transparency_node_by = {'Region': {'variable' : 'ident_Log2FC_RSNs', 'min_alpha': 0.1},
'Gene': {'variable' : 'ident_Log2FC_RSNs', 'min_alpha': 0.1}},
size_node_by = {'TF': {'variable': 'fixed_size', 'fixed_size': 30},
'Gene': {'variable': 'fixed_size', 'fixed_size': 15},
'Region': {'variable': 'fixed_size', 'fixed_size': 10}},
shape_node_by = {'TF': {'variable': 'fixed_shape', 'fixed_shape': 'ellipse'},
'Gene': {'variable': 'fixed_shape', 'fixed_shape': 'ellipse'},
'Region': {'variable': 'fixed_shape', 'fixed_shape': 'diamond'}},
label_size_by = {'TF': {'variable': 'fixed_label_size', 'fixed_label_size': 20.0},
'Gene': {'variable': 'fixed_label_size', 'fixed_label_size': 10.0},
'Region': {'variable': 'fixed_label_size', 'fixed_label_size': 0.0}},
layout='kamada_kawai_layout',
scale_position_by=250)
plt.figure(figsize=(20,20))
plot_networkx(G, pos)
plt.savefig('figures/TFs_target_genes/eRegulons.pdf')
export_to_cytoscape(G, pos, out_file = os.path.join('figures/network.cys'))
I hope it will work for you as well!
However, I didn't manage to solve the problem with import to cytoscape :(
Best, Dmitrii
yiyelinfeng
commented
5 months ago just change "export_to_cytoscape(G, pos, out_file = os.path.join('figures/network.cyjs'))", it works.
degrainger
commented
4 months ago Hi @SeppeDeWinter ,
Are you able to provide any update on this please? I am facing the same issue of create_nx_tables giving empty dataframes. I am also running into errors with @DmitriiSeverinov's code :/ Thanks!
Best, Dave
SeppeDeWinter
commented
3 months ago Hi @degrainger
I did not find the time yet to update this function, I'm sorry!
Best,
Seppe
degrainger
commented
3 months ago Not to worry, we can see you're a busy guy! 😄 I troubleshooted my way through the above solution and got it working in the end. Thanks for all your hard work! -Dave
SteveTur
commented
3 months ago Hi @SeppeDeWinter,
First of all, Seppe, no worries and thank you so much for all of the wonderful tools and all your help!
Hi @degrainger,
Could you share with us how you managed to troubleshoot this? Which version of SCENIC are you using to make it work? Is it the alpha version?
Thank you for your help!
Best,
Steven
degrainger
commented
3 months ago Hi @SteveTur,
Sure, I can share what I did but it mostly came from @DmitriiSeverinov's answer above.
I ran SCENIC+ through the snakemake pipeline using version 1.0a1. I converted my mudata to scenicobject:
scplus_obj = mudata_to_scenicplus(
mdata = scplus_mdata,
path_to_cistarget_h5 = "ctx_results.hdf5",
path_to_dem_h5 = "dem_results.hdf5"
)Then in Dmitrii's answer, they have a slot called "direct_e_regulon_metadata_filtered" in their scenicplus object. I didn't have this so I made it manually by filtering for the TFs I was intererested in:
direct_md = scplus_obj.uns['eRegulon_metadata']
direct_filt = direct_md[direct_md['TF'].isin(['Etv2', 'Ebf1', 'Ebf2','Lmo2','Tal1','E2f6','Yy1'])]
scplus_obj.uns['direct_e_regulon_metadata_filtered'] = direct_filtI then ran Dmitrii's code but with a couple changes:
scplus_obj.uns['direct_e_regulon_metadata_filtered'] = scplus_mdata.uns['direct_e_regulon_metadata_filtered'])Make sure all your TFs are defined in the dictionaries and that all the variables match the same name as in your scplus_obj.uns columns as this was different in mine versus Dmitrii's
from pycisTopic.diff_features import find_highly_variable_features
#scplus_obj.uns['direct_e_regulon_metadata_filtered'] = scplus_mdata.uns['direct_e_regulon_metadata_filtered']
hvr = find_highly_variable_features(scplus_obj.to_df('ACC').loc[list(set(scplus_obj.uns['direct_e_regulon_metadata_filtered']['Region']))], n_top_features=1000, plot = False)
hvg = find_highly_variable_features(scplus_obj.to_df('EXP')[list(set(scplus_obj.uns['direct_e_regulon_metadata_filtered']['Gene']))].T, n_top_features=1000, plot = False)
from scenicplus.networks import create_nx_tables, create_nx_graph, plot_networkx, export_to_cytoscape
nx_tables = list()
def _format_df_nx(df, key, var):
"""
A helper function to format differential test results
"""
df.index = df['names']
df = pd.DataFrame(df['logfoldchanges'])
df.columns = [var+'_Log2FC_'+key]
df.index.name = None
return df
def _get_log2fc_nx(scplus_obj: 'SCENICPLUS',
variable,
features,
contrast: Optional[str] = 'gene'
):
"""
A helper function to derive log2fc changes
"""
if contrast == 'gene':
adata = anndata.AnnData(X=scplus_obj.X_EXP, obs=pd.DataFrame(
index=scplus_obj.cell_names), var=pd.DataFrame(index=scplus_obj.gene_names))
if contrast == 'region':
adata = anndata.AnnData(X=scplus_obj.X_ACC.T, obs=pd.DataFrame(
index=scplus_obj.cell_names), var=pd.DataFrame(index=scplus_obj.region_names))
adata.obs = pd.DataFrame(scplus_obj.metadata_cell[variable])
sc.pp.normalize_total(adata, target_sum=1e4)
sc.pp.log1p(adata)
adata = adata[:, features]
sc.tl.rank_genes_groups(
adata, variable, method='wilcoxon', corr_method='bonferroni')
groups = adata.uns['rank_genes_groups']['names'].dtype.names
diff_list = [_format_df_nx(sc.get.rank_genes_groups_df(
adata, group=group), group, variable) for group in groups]
return pd.concat(diff_list, axis=1)
def create_nx_tables(scplus_obj: 'SCENICPLUS',
eRegulon_metadata_key: str ='eRegulon_metadata',
subset_eRegulons: List = None,
subset_regions: List = None,
subset_genes: List = None,
add_differential_gene_expression: bool = False,
add_differential_region_accessibility: bool = False,
differential_variable: List =[]):
"""
A function to format eRegulon data into tables for plotting eGRNs.
Parameters
---------
scplus_obj: SCENICPLUS
A SCENICPLUS object with eRegulons
eRegulon_metadata_key: str, optional
Key where the eRegulon metadata dataframe is stored
subset_eRegulons: list, optional
List of eRegulons to subset
subset_regions: list, optional
List of regions to subset
subset_genes: list, optional
List of genes to subset
add_differential_gene_expression: bool, optional
Whether to calculate differential gene expression logFC for a given variable
add_differential_region_accessibility: bool, optional
Whether to calculate differential region accessibility logFC for a given variable
differential_variable: list, optional
Variable to calculate differential gene expression or region accessibility.
Return
---------
A dictionary with edge feature tables ('TF2G', 'TF2R', 'R2G') and node feature tables ('TF', 'Gene', 'Region')
"""
er_metadata = scplus_obj.uns[eRegulon_metadata_key].copy()
if subset_eRegulons is not None:
# subset_eRegulons = [x + '_[^a-zA-Z0-9]' for x in subset_eRegulons]
er_metadata = er_metadata[er_metadata['Region_signature_name'].str.contains(
'|'.join(subset_eRegulons))]
if subset_regions is not None:
er_metadata = er_metadata[er_metadata['Region'].isin(subset_regions)]
if subset_genes is not None:
er_metadata = er_metadata[er_metadata['Gene'].isin(subset_genes)]
nx_tables = {}
nx_tables['Edge'] = {}
nx_tables['Node'] = {}
# Generate edge tables
r2g_columns = [x for x in er_metadata.columns if 'R2G' in x]
tf2g_columns = [x for x in er_metadata.columns if 'TF2G' in x]
nx_tables['Edge']['TF2R'] = er_metadata[er_metadata.columns.difference(
r2g_columns + tf2g_columns)].drop('Gene', axis=1).drop_duplicates()
nx_tables['Edge']['TF2R'] = nx_tables['Edge']['TF2R'][['TF', 'Region'] +
nx_tables['Edge']['TF2R'].columns.difference(['TF', 'Region']).tolist()]
nx_tables['Edge']['R2G'] = er_metadata[er_metadata.columns.difference(
tf2g_columns)].drop('TF', axis=1).drop_duplicates()
nx_tables['Edge']['R2G'] = nx_tables['Edge']['R2G'][['Region', 'Gene'] +
nx_tables['Edge']['R2G'].columns.difference(['Region', 'Gene']).tolist()]
nx_tables['Edge']['TF2G'] = er_metadata[er_metadata.columns.difference(
r2g_columns)].drop('Region', axis=1).drop_duplicates()
nx_tables['Edge']['TF2G'] = nx_tables['Edge']['TF2G'][['TF', 'Gene'] +
nx_tables['Edge']['TF2G'].columns.difference(['TF', 'Gene']).tolist()]
# Generate node tables
tfs = list(set(er_metadata['TF']))
nx_tables['Node']['TF'] = pd.DataFrame(
'TF', index=tfs, columns=['Node_type'])
nx_tables['Node']['TF']['TF'] = tfs
genes = list(set(er_metadata['Gene']))
genes = [x for x in genes if x not in tfs]
nx_tables['Node']['Gene'] = pd.DataFrame(
'Gene', index=genes, columns=['Node_type'])
nx_tables['Node']['Gene']['Gene'] = genes
regions = list(set(er_metadata['Region']))
nx_tables['Node']['Region'] = pd.DataFrame(
'Region', index=regions, columns=['Node_type'])
nx_tables['Node']['Region']['Region'] = regions
# Add gene logFC
if add_differential_gene_expression is True:
for var in differential_variable:
nx_tables['Node']['TF'] = pd.concat([nx_tables['Node']['TF'], _get_log2fc_nx(
scplus_obj, var, nx_tables['Node']['TF'].index.tolist(), contrast='gene')], axis=1)
nx_tables['Node']['Gene'] = pd.concat([nx_tables['Node']['Gene'], _get_log2fc_nx(
scplus_obj, var, nx_tables['Node']['Gene'].index.tolist(), contrast='gene')], axis=1)
if add_differential_region_accessibility is True:
for var in differential_variable:
nx_tables['Node']['Region'] = pd.concat([nx_tables['Node']['Region'], _get_log2fc_nx(
scplus_obj, var, nx_tables['Node']['Region'].index.tolist(), contrast='region')], axis=1)
return nx_tables
nx_table = create_nx_tables(
scplus_obj = scplus_obj,
eRegulon_metadata_key ='direct_e_regulon_metadata_filtered',
subset_eRegulons = ['Etv2', 'Ebf1', 'Ebf2','Lmo2','Tal1','E2f6','Yy1'],
subset_regions = hvr,
subset_genes = hvg,
add_differential_gene_expression = True,
add_differential_region_accessibility = True,
differential_variable = ['celltypeNew']
)
G, pos, edge_tables, node_tables = create_nx_graph(nx_table,
use_edge_tables = ['TF2R','R2G'],
color_edge_by = {'TF2R': {'variable' : 'TF', 'category_color' : {'Etv2': 'Purple', 'Ebf1': 'Red', 'Ebf2' : 'Orange','Lmo2' : 'Green','Tal1' :'Blue','E2f6':'Cyan','Yy1':'Grey'}},
'R2G': {'variable' : 'importance_x_rho', 'continuous_color' : 'viridis', 'v_min': -1, 'v_max': 1}},
transparency_edge_by = {'R2G': {'variable' : 'importance_R2G', 'min_alpha': 0.1, 'v_min': 0}},
width_edge_by = {'R2G': {'variable' : 'importance_R2G', 'max_size' : 1.5, 'min_size' : 1}},
color_node_by = {'TF': {'variable': 'TF', 'category_color' : {'Etv2': 'Purple', 'Ebf1': 'Red', 'Ebf2' : 'Orange','Lmo2' : 'Green','Tal1' :'Blue','E2f6':'Cyan','Yy1':'Grey'}},
'Gene': {'variable': 'celltypeNew_Log2FC_early_angioblast', 'continuous_color' : 'bwr'},
'Region': {'variable': 'celltypeNew_Log2FC_early_angioblast', 'continuous_color' : 'viridis'}},
transparency_node_by = {'Region': {'variable' : 'celltypeNew_Log2FC_early_angioblast', 'min_alpha': 0.1},
'Gene': {'variable' : 'celltypeNew_Log2FC_early_angioblast', 'min_alpha': 0.1}},
size_node_by = {'TF': {'variable': 'fixed_size', 'fixed_size': 30},
'Gene': {'variable': 'fixed_size', 'fixed_size': 15},
'Region': {'variable': 'fixed_size', 'fixed_size': 10}},
shape_node_by = {'TF': {'variable': 'fixed_shape', 'fixed_shape': 'ellipse'},
'Gene': {'variable': 'fixed_shape', 'fixed_shape': 'ellipse'},
'Region': {'variable': 'fixed_shape', 'fixed_shape': 'diamond'}},
label_size_by = {'TF': {'variable': 'fixed_label_size', 'fixed_label_size': 20.0},
'Gene': {'variable': 'fixed_label_size', 'fixed_label_size': 10.0},
'Region': {'variable': 'fixed_label_size', 'fixed_label_size': 0.0}},
layout='kamada_kawai_layout',
scale_position_by=250)
plt.figure(figsize=(20,20))
plot_networkx(G, pos)
plt.savefig('Test2.pdf')
export_to_cytoscape(G, pos, out_file = os.path.join('Test2_network.cyjs'))Let me know how you get on or if you get errors - I may have encountered them in my troubleshooting. Good luck! -Dave
SteveTur
commented
3 months ago Hi Dave,
Thank you so much for this update! I will let you know if it works for me. :)
Best,
Steven
SteveTur
commented
3 months ago Hi @SeppeDeWinter, @degrainger, @DmitriiSeverinov,
I managed to make it work too! However, I am still confused with some parameters in this function. How do you choose the HVR (highly variable regions) and HVG (highly variable genes) clearly? The shape and number of connections change a lot depending on those selections.
Here's how I've been approaching it: I look at the specificity score of each cell type and the correlation heatmap. Then, for each cell, I plot the network of the TFs involved in the same cell type depending on SS and heatmap. However, it's not clear if you can specify the cell type when plotting the network, and how exactly the HVR/HVG parameters work. For example, if I have an eRegulon gene-based with eRegulon+/+(120g), does the HVG parameter influence those 120 genes? Is it the same for regions?
Another point is when you plot the network of your chosen TF, you don't specify whether it is extended or direct. Does it consider both if you have both for each TF?
If you can share how you deal with these issues, it would be very helpful!
Best regards, Steven
degrainger
commented
3 months ago Hi @SeppeDeWinter, @degrainger, @DmitriiSeverinov,
I managed to make it work too! However, I am still confused with some parameters in this function. How do you choose the HVR (highly variable regions) and HVG (highly variable genes) clearly? The shape and number of connections change a lot depending on those selections.
Here's how I've been approaching it: I look at the specificity score of each cell type and the correlation heatmap. Then, for each cell, I plot the network of the TFs involved in the same cell type depending on SS and heatmap. However, it's not clear if you can specify the cell type when plotting the network, and how exactly the HVR/HVG parameters work. For example, if I have an eRegulon gene-based with eRegulon+/+(120g), does the HVG parameter influence those 120 genes? Is it the same for regions?
Another point is when you plot the network of your chosen TF, you don't specify whether it is extended or direct. Does it consider both if you have both for each TF?
If you can share how you deal with these issues, it would be very helpful!
Best regards, Steven
Hey @SteveTur,
Congrats, glad you got it working. I also considered similar things when performing my analysis. Firstly, I used the highly variable genes / regions to show the most variable genes are regulated by TFs known to be important in my field of study. My data are currently limited to clusters of cells along a single differentiation path and so I wanted to make a figure demonstrating known TFs regulate many of the most variable features along this process.
However, when wanting to look at potentially novel regulators, and their target regions/genes, I thought it would be better to select regions and genes differently. Instead, I selected the top 30-50 genes and regions as ranked by triplet score for each regulon. I thought this would give a better representation of the GRN.
Let me know what you think about that, happy to share the code if you're interested.
Dave
SteveTur
commented
3 months ago Hi @degrainger,
Thank you for this advice, it seems to be a good start. I am gonna try this first to see how the results look!
I will be happy to see the code you wrote to do this analysis!
Thank you so much for your help :)
Best,
Steven
ccruizm
commented
2 months ago Hello @degrainger, Your solution (alongside what @DmitriiSeverinov mentioned) also worked for me! Could you please share the code you mentioned here? I would also like to have a better idea of how to prioritize the TF when building the GRNs. Thanks in advance!
Firstly, I used the highly variable genes / regions to show the most variable genes are regulated by TFs known to be important in my field of study. My data are currently limited to clusters of cells along a single differentiation path and so I wanted to make a figure demonstrating known TFs regulate many of the most variable features along this process. However, when wanting to look at potentially novel regulators, and their target regions/genes, I thought it would be better to select regions and genes differently. Instead, I selected the top 30-50 genes and regions as ranked by triplet score for each regulon. I thought this would give a better representation of the GRN.
Hello,
It seems that the function to plot the GRN on python doesn't work anymore since the recent updates:
from scenicplus.networks import create_nx_tables, create_nx_graph, plot_networkx, export_to_cytoscape G, pos, edge_tables, node_tables = create_nx_graph( nx_tables, use_edge_tables = ['TF2R','R2G'], color_edge_by = { 'TF2R': {'variable' : 'TF', 'category_color' : category_color}, 'R2G': {'variable' : 'R2G_rho', 'continuous_color' : 'viridis', 'v_min': -1, 'v_max': 1} }, transparency_edge_by = { 'R2G': {'variable' : 'R2G_importance', 'min_alpha': 0.1, 'v_min': 0} }, width_edge_by = { 'R2G': {'variable' : 'R2G_importance', 'max_size' : 1.5, 'min_size' : 1} }, color_node_by = { 'TF': {'variable': 'TF', 'category_color' : category_color}, 'Gene': {'variable': 'GEX_celltype_Log2FC_6', 'continuous_color' : 'bwr'}, 'Region': {'variable': 'GEX_celltype_Log2FC_6', 'continuous_color' : 'viridis'} }, transparency_node_by = { 'Region': {'variable' : 'GEX_celltype_Log2FC_6', 'min_alpha': 0.1}, 'Gene': {'variable' : 'GEX_celltype_Log2FC_6', 'min_alpha': 0.1} }, size_node_by = { 'TF': {'variable': 'fixed_size', 'fixed_size': 30}, 'Gene': {'variable': 'fixed_size', 'fixed_size': 15}, 'Region': {'variable': 'fixed_size', 'fixed_size': 10} }, shape_node_by = { 'TF': {'variable': 'fixed_shape', 'fixed_shape': 'ellipse'}, 'Gene': {'variable': 'fixed_shape', 'fixed_shape': 'ellipse'}, 'Region': {'variable': 'fixed_shape', 'fixed_shape': 'diamond'} }, label_size_by = { 'TF': {'variable': 'fixed_label_size', 'fixed_label_size': 10.0}, 'Gene': {'variable': 'fixed_label_size', 'fixed_label_size': 10.0}, 'Region': {'variable': 'fixed_label_size', 'fixed_label_size': 0.0} }, layout='kamada_kawai_layout', scale_position_by=250 )
plt.figure(figsize=(50,50)) plot_networkx(G, pos)
Only blank output are coming out for some reason. We never got any problem before the recent changes. What should we do?
Thank you for your help,
Best,
Steven