def cluser_enrich(enr,gl,pval=0.05, top_clusters=20):
#2= Filter terms by p-Val
enr = enr[enr['p-Val']<pval]
#3= Make claster by kappa coeff
enr = erl.cluster(set(gl), enr, deep=2)
#3-1= Filter top clusters
enr = enr[enr['cluster']<top_clusters]
#4= Make clustered geneset
gs_clust,nt_cl = erl.cluster_genset(enr)
#5= Enrich clustered geneset
enr_clust = erl.enrich(gl,gs_clust)
# deduplicate index -- TODO!!! in package
nt_cl = nt_cl.loc[~nt_cl.index.duplicated(keep='first')]
#6= Add cluster to table
enr_clust = pd.concat([enr_clust,nt_cl.loc[enr_clust.index]['cluster']],axis=1, sort=False)
#7= Make graphs
G_gs = erl.make_graph_n(gl,enr, kappa=0.4)
G_cl = erl.make_graph_n(gl,enr_clust, kappa=0.01)
#8= Draw graphs
erl.draw_graph(G_gs, spring=150)
erl.draw_direct(G_cl)
#9= Draw barplot for clustered terms
enr.sort_values('cluster', axis=0, inplace = True)
cm = ('tab20' if max(enr['cluster'])>10 else 'tab10')
f, ax = plt.subplots(figsize=(8, 12))
sns.barplot(y=enr.index,
x='-log10(p-Val)',
ax = ax,
hue ='cluster',
dodge=False,
data = enr,
palette = cm)
ax.set_title('Top terms in clusters ')
exp_name='GO_KEGG2'
#~ gt = pd.read_table('tracks/MARGE/relativeRP/DN_RP_enhancers_genes.txt')
gl = [ 'CD3E', 'BLK', 'PTPN22', 'PAG1', 'CTLA4', 'PIK3CD', 'LAT2', 'CSK',
'CD247', 'CD3G', 'THEMIS', 'PSMB8', 'LCP2', 'GATA3', 'LAT', 'SLA2',
'SKAP1', 'TRAT1', 'BCL2', 'CD3D', 'THY1','RUNX1', 'BLK', 'PTPN22']
#setR = set(gl)
gs_fn ='EnrichrLibs/Reactome_2016.gmt'
gs = erl.read_gmt(gs_fn)
enr = erl.enrich(set(gl), gs)
#get_Enrichr(out_dir = 'EnrichrLibs')
# Several gene sets in one
gss = [
'GO_Biological_Process_2018',
'GO_Cellular_Component_2018',
'GO_Molecular_Function_2018',
'KEGG_2016',
'Reactome_2016'
]
# # -- transform
# rp = rpt.sort_values('lgFC_TLXvsRAG', axis=0, ascending=False)
# rp.drop_duplicates(subset='gene_name', inplace=True)
# Ap,Bp = 'TLX_rel_RP','RAG_rel_RP'
# cols = ['gene_name', Ap, Bp]
# rp = rp[cols]
# rp = rp.set_index(keys=rp.columns[0])
# ### Annotations: genes, tss, T-all oncogenes etc.
# In[9]:
# Load T-ALL ocnogenes
tall = erl.read_gmt(join(DATADIR, 'gene_lists/Cancermine/T-ALL.gmt'))
df_mut = pd.read_csv(
join(
DATADIR,
'gene_lists/COSMIC/Genes_mutation_HUMAN_Acute-lymphoblastic-leukaemia.csv'
))
df_mut['Gname'] = df_mut['Gene name'].apply(lambda x: x.split('_')[0])
tall_onc = tall['T-ALL all']
tall_mut = list(df_mut['Gname'].unique())
# Load genes body regions
genes = pb.BedTool(
join(DATADIR, 'tracks/annot_tracks/references/mm9/mm9.refGene.bed'))
# In[9]:
# List of gene sets as above
#~ gss = [
#~ 'GO_Biological_Process_2018',
#~ 'GO_Cellular_Component_2018',
#~ 'GO_Molecular_Function_2018',
#~ 'KEGG_2016',
#~ 'Reactome_2016'
#~ ]
# ### Batch enrichment
# In[10]:
enrr = erl.enrich_gs(gl, gss, path_lib=lib_dir)
# ### Plots
# In[11]:
#~ enrr.sort_values('p-Val', axis=0, inplace = True)
#~ ds = enrr.head(20)
#~ f, ax = plt.subplots()
#~ sns.barplot(y=ds.index,
#~ x='-log10(p-Val)',
#~ ax = ax,
#~ color="Red",
#~ data = ds)
#~ ax.set_title('All terms')
lib_dir='_tmp'
gss = [
'GO_Biological_Process_2018',
#~ 'GO_Cellular_Component_2018',
#~ 'GO_Molecular_Function_2018',
'KEGG_2016',
#'Reactome_2016'
]
#1= Enrich all terms
enr = erl.enrich_gs(gl,gss, path_lib=lib_dir)
#2= Filter terms by p-Val
enr = enr[enr['p-Val']<0.005]
#3= Make claster by kappa coeff
enr = erl.cluster(set(gl), enr, deep=2)
#3-1= Filter top clusters
top_clusters = 18
enr = enr[enr['cluster']<top_clusters]
#4= Make clustered geneset
gs_clust,nt_cl = erl.cluster_genset(enr)
#5= Enrich clustered geneset
## Enrichment analysis
# List of gene sets
gss = [
#~ 'GO_Biological_Process_2018',
#~ 'GO_Cellular_Component_2018',
#~ 'GO_Molecular_Function_2018',
#~ 'KEGG_2016',
#~ 'Reactome_2016',
'Cancer_Cell_Line_Encyclopedia',
'NCI-60_Cancer_Cell_Lines',
'MSigDB_Computational',
'MSigDB_Oncogenic_Signatures'
]
enr = erl.enrich_gs(gl, gss, path_lib='../data/EnrichrLibs')
# For futher analysis it is convinient to filter terms by p-value
enr_a = enr[enr['p-Val'] < 0.05]
G, enr_out, nt = erl.make_graph(gl, enr_a, kappa=0.4)
enr_out.sort_values('cluster', axis=0, inplace=True)
# --- Plot ---
cm = 'tab20'
erl.draw_graph(G, spring=500, pval_prcnt=0.7, palette=cm)
ds = enr_out.head(40)
f, ax = plt.subplots(figsize=(12, 12))
setR = set(list(gt.iloc[:, 0]))
gss = [
#~ 'Pierre_gene_sets.gmt',
'GO_Biological_Process_2018.gmt',
'GO_Cellular_Component_2018.gmt',
'GO_Molecular_Function_2018.gmt',
'KEGG_2016.gmt',
'Reactome_2016.gmt'
]
enrr = pd.DataFrame()
for gs in gss:
pl = erl.read_gmt('EnrichrLibs/' + gs)
enr = erl.enrich(setR, pl)
print(len(enr))
enrr = pd.concat([enrr, enr])
#~ ds = enr.head(20)
#~ f, ax = plt.subplots(figsize=(16.5, 6.5))
#~ sns.barplot(y=ds.index,
#~ x='-log10(p-Val)',
#~ ax = ax,
#~ color="Red",
#~ data = ds)
#~ ax.set_title(gs.split(".")[0])
# p-Value filter
enrr = enrr[enrr['p-Val'] < 0.005]
'MSigDB_Computational',
'MSigDB_Oncogenic_Signatures',
#'Mouse_Gene_Atlas',
'NCI-60_Cancer_Cell_Lines',
#'NCI-Nature_2016',
#'OMIM_Disease',
'RNA-Seq_Disease_Gene_and_Drug_Signatures_from_GEO',
'Reactome_2016',
#'WikiPathways_2016'
]
# == UP analysis
upTLX_list = [x.upper() for x in upTLX_list]
enrUP_tlx = erl.enrich_gs(upTLX_list, gss)
# --- Plot ---
enrUP_tlx.sort_values('p-Val', axis=0, inplace=True)
ds = enrUP_tlx.head(20)
f, ax = plt.subplots()
sns.barplot(y=ds.index, x='-log10(p-Val)', ax=ax, color="Red", data=ds)
ax.set_title('UP_dRP_Tlx_peaks')
if SAVE:
plt.savefig(pp, format='pdf')
# == DN analysis
dnTLX_list = [x.upper() for x in dnTLX_list]
'ChEA_2016',
'Disease_Perturbations_from_GEO_down',
'Disease_Perturbations_from_GEO_up',
'Disease_Signatures_from_GEO_down_2014',
'Disease_Signatures_from_GEO_up_2014',
'GO_Biological_Process_2018',
'GO_Cellular_Component_2018',
'GO_Molecular_Function_2018',
'Human_Phenotype_Ontology',
'KEGG_2016',
'MSigDB_Computational',
'MSigDB_Oncogenic_Signatures',
'Mouse_Gene_Atlas',
'NCI-60_Cancer_Cell_Lines',
'NCI-Nature_2016',
#'OMIM_Disease',
'RNA-Seq_Disease_Gene_and_Drug_Signatures_from_GEO',
'Reactome_2016',
'WikiPathways_2016'
]
enrDD = erl.enrich_gs(gup, gss)
enrDD.sort_values('p-Val', axis=0, inplace=True)
ds = enrDD.head(20)
f, ax = plt.subplots(figsize=(16.5, 6.5))
sns.barplot(y=ds.index, x='-log10(p-Val)', ax=ax, color="Red", data=ds)
ax.set_title('All terms')
plt.show()
# my libs
import EnrichRLib as erl
# Project settings
from os.path import join
WORKDIR = '/home/sergio/Res_CIML/TLX3_project'
SCRIPTS = join(WORKDIR,'scripts')
DATADIR = join(WORKDIR,'data')
WGS = join(DATADIR,'tracks/WGS-WES/Germline')
RP = join(DATADIR,'tracks/MARGE/relativeRP/bam_input')
# Load genes body regions
genes = pb.BedTool(join(DATADIR,'tracks/annot_tracks/references/mm9/mm9.refGene.bed'))
tall = erl.read_gmt(join(DATADIR,'gene_lists/Cancermine/T-ALL.gmt'))
# Genes of interests
gl = ['BCL11B',
'EZH2',
'RUNX1',
'FBXW7',
'FOS',
'ETV6',
'RPL5',
'RB1',
'GATA3',
'LEF1',
'TET1',
'PTEN',