notebook 5: add export of marker gene lists per cluster / cell type
after final clustering, marker genes driving the clusters need to be exported in a txt format file for users to check for biological meaning
import scanpy
method_ranking='wilcoxon'
scanpy.tl.rank_genes_groups(adata, groupby = 'clustering', method=method_ranking, n_genes = 20, key_added = method_ranking)
import pandas as pd
local_path = "/mnt/workspace/mlooso/poe16/"
data="pipeline_output/results/poe16cite_1/"
for c in adata.obs['clustering'].cat.categories:
d = {'gene' : adata.uns[method_ranking]['names'][c],
'logfc' : adata.uns[method_ranking]['logfoldchanges'][c],
'pval' : adata.uns[method_ranking]['pvals'][c],
'qval' : adata.uns[method_ranking]['pvals_adj'][c],
'score': adata.uns[method_ranking]['scores'][c]}
df = pd.DataFrame(d).set_index('gene')
df = pd.merge(df, adata.var, left_index=True, right_index=True)
is_lfc_greater_than = abs(d['logfc'])>0.01
df[is_lfc_greater_than].to_csv(path_or_buf=local_path+data+'03-markergenes_cluster_{c}.tsv'.format(c = c),
sep = '\t',
header = True,
index=True)
Edited by Hendrik Schultheis