1 Load environment

Code

import warnings

warnings.filterwarnings("ignore")

import matplotlib.pyplot as plt
import seaborn as sns
import scanpy as sc
import pandas as pd
import numpy as np
import random
import itertools

from tqdm import tqdm

import decoupler as dc
import sys

sys.setrecursionlimit(20000)

sys.path.append("./../../../../utilities_folder")
from utilities import load_object, intTable, plotGenesInTerm, getAnnGenes, run_ora_catchErrors

Set R environment with rpy2:

Code

import rpy2.rinterface_lib.callbacks
import anndata2ri
import logging

from rpy2.robjects import pandas2ri
import rpy2.robjects as ro

sc.settings.verbosity = 0
rpy2.rinterface_lib.callbacks.logger.setLevel(logging.ERROR)

pandas2ri.activate()
anndata2ri.activate()

%load_ext rpy2.ipython

Set up of graphical parameters for Python plots:

Code

%matplotlib inline
sc.set_figure_params(dpi = 300, fontsize = 20)

plt.rcParams['svg.fonttype'] = 'none'

cmap_up = sns.light_palette("red", as_cmap=True)
cmap_down = sns.light_palette("blue", as_cmap=True)
cmap_all = sns.light_palette("seagreen", as_cmap=True)

Set up of graphical parameters for R plots:

Code

default_units = 'in' 
default_res = 300
default_width = 10
default_height = 9

import rpy2
old_setup_graphics = rpy2.ipython.rmagic.RMagics.setup_graphics

def new_setup_graphics(self, args):
    if getattr(args, 'units') is not None:
        if args.units != default_units:
            return old_setup_graphics(self, args)
    args.units = default_units
    if getattr(args, 'res') is None:
        args.res = default_res
    if getattr(args, 'width') is None:
        args.width = default_width
    if getattr(args, 'height') is None:
        args.height = default_height        
    return old_setup_graphics(self, args)


rpy2.ipython.rmagic.RMagics.setup_graphics = new_setup_graphics

Here the cell were we inject the parameters using Quarto renderer:

Code

# Injected Parameters
N = 3

Code

# Injected Parameters
N = 10

Import R libraries:

Code

%%R
source('./../../../../utilities_folder/GO_helper.r')
loc <- './../../../../R_loc' # pointing to the renv environment

.libPaths(loc)

library('topGO')
library('org.Hs.eg.db')
library(dplyr)
library(ggplot2)

Set output folders:

Code

output_folder = './'
folder = './tables/cluster_' + str(N) + '/'

2 Load data

Here we load the dataframe:

Code

markers = pd.read_excel(folder + 'genes_in_cluster_' + str(N) + '.xlsx', index_col = 0)
markers

	logFC.celltypes_leveledhEGCLC	logFC.celltypes_leveledhPGCLC	logFC.celltypes_levelediMeLC	logCPM	LR	PValue	FDR	clusters
SERPINE1	13.498855	0	0	3.699170	952.961162	2.877701e-206	1.293672e-204	10
MAGEA4	13.019832	0	0	3.247550	783.518987	1.623797e-169	5.225319e-168	10
PPP1R17	12.975739	0	0	3.211685	770.603253	1.026946e-166	3.218391e-165	10
ZFP57	12.801596	0	0	3.041745	716.180908	6.506513e-155	1.803641e-153	10
MT1H	12.555808	0	0	2.810490	648.745475	2.724928e-140	6.413199e-139	10
...	...	...	...	...	...	...	...	...
AFP	10.308805	0	0	0.707135	296.923879	4.608047e-64	3.086517e-63	10
RTBDN	10.277904	0	0	0.650750	293.080804	3.127664e-63	2.066117e-62	10
TMEM187	10.248796	0	0	0.621090	290.120482	1.367276e-62	8.958198e-62	10
INHBA	10.160428	0	0	0.573238	282.355768	6.547756e-61	4.143000e-60	10
SV2B	10.146218	0	0	0.509373	279.646228	2.525704e-60	1.586396e-59	10

172 rows × 8 columns

Code

allGenes_series = pd.read_csv('./tables/all_bkg_genes.csv')
allGenes = allGenes_series['0'].tolist()

Here we load the dictionary that associates to each GO term its genes:

Code

GO2gene = load_object('./../../../../data/GO2gene_complete.pickle')

3 Markers of cluster

We filter genes for the cluster under investigation based on the p-value adjusted that we then convert in -log(p-value adjusted):

Code

markers = markers[markers.FDR < 0.01]
markers['-log10(FDR)'] = -np.log10(markers.FDR)
markers = markers.replace(np.inf, markers[markers['-log10(FDR)'] != np.inf]['-log10(FDR)'].max())
markers

	logFC.celltypes_leveledhEGCLC	logFC.celltypes_leveledhPGCLC	logFC.celltypes_levelediMeLC	logCPM	LR	PValue	FDR	clusters	-log10(FDR)
SERPINE1	13.498855	0	0	3.699170	952.961162	2.877701e-206	1.293672e-204	10	203.888176
MAGEA4	13.019832	0	0	3.247550	783.518987	1.623797e-169	5.225319e-168	10	167.281887
PPP1R17	12.975739	0	0	3.211685	770.603253	1.026946e-166	3.218391e-165	10	164.492361
ZFP57	12.801596	0	0	3.041745	716.180908	6.506513e-155	1.803641e-153	10	152.743850
MT1H	12.555808	0	0	2.810490	648.745475	2.724928e-140	6.413199e-139	10	138.192925
...	...	...	...	...	...	...	...	...	...
AFP	10.308805	0	0	0.707135	296.923879	4.608047e-64	3.086517e-63	10	62.510531
RTBDN	10.277904	0	0	0.650750	293.080804	3.127664e-63	2.066117e-62	10	61.684845
TMEM187	10.248796	0	0	0.621090	290.120482	1.367276e-62	8.958198e-62	10	61.047779
INHBA	10.160428	0	0	0.573238	282.355768	6.547756e-61	4.143000e-60	10	59.382685
SV2B	10.146218	0	0	0.509373	279.646228	2.525704e-60	1.586396e-59	10	58.799588

172 rows × 9 columns

3.0.1 All regulated

Code

all_sign = markers.index.tolist()
allSelected = allGenes_series['0'].isin(all_sign).astype('int').tolist()

4 topGO

4.1 All significant

Code

%%R -i allSelected -i allGenes

allGenes_v <- c(allSelected)
#print(allGenes_v)
names(allGenes_v) <- allGenes
allGenes_v <- unlist(allGenes_v)

geneNames <- c(allGenes)

ann_org_BP <- topGO::annFUN.org(whichOnto='BP', feasibleGenes=names(allGenes_v), 
                           mapping='org.Hs.eg', ID='symbol')

ann_org_MF <- topGO::annFUN.org(whichOnto='MF', feasibleGenes=names(allGenes_v), 
                           mapping='org.Hs.eg', ID='symbol')

ann_org_CC <- topGO::annFUN.org(whichOnto='CC', feasibleGenes=names(allGenes_v), 
                           mapping='org.Hs.eg', ID='symbol')

selection <- function(allScores){return (as.logical(allScores))}

Biological Process

Code

%%R
#print(lapply(ann_org_BP, count_genes))

GOdata <- new("topGOdata",
  ontology="BP",
  allGenes=allGenes_v,
  annot=annFUN.GO2genes,
  GO2genes=ann_org_BP,
  geneSel = selection,
  nodeSize=10)

Code

%%R -o results

results <- runTest(GOdata, algorithm="weight01",statistic="fisher")

Code

scores = ro.r.score(results)
score_names = ro.r(
'''
names(results@score)
'''
)
go_data = ro.r.GOdata

genesData = ro.r(
'''
geneData(results)
'''
)
genesData

array([10903,   159,    10,  2609], dtype=int32)

Code

#num_summarize = min(100, len(score_names))
results_table = ro.r.GenTable(go_data, weight=results,
        orderBy="weight", topNodes=len(scores))

Code

results_table_py = ro.conversion.rpy2py(results_table)

Code

scores_py = ro.conversion.rpy2py(scores)
score_names = [i for i in score_names]

Code

scores_df = pd.DataFrame({'Scores': scores_py, 'GO.ID': score_names})
results_table_py = results_table_py.merge(scores_df, left_on = 'GO.ID', right_on = 'GO.ID')
results_table_py

	GO.ID	Term	Annotated	Significant	Expected	weight	Scores
0	GO:0030728	ovulation	12	3	0.17	0.00061	0.000608
1	GO:0007194	negative regulation of adenylate cyclase...	12	3	0.17	0.00061	0.000608
2	GO:0022602	ovulation cycle process	28	4	0.41	0.00068	0.000679
3	GO:0021954	central nervous system neuron developmen...	61	5	0.89	0.00076	0.000763
4	GO:0034220	ion transmembrane transport	653	16	9.52	0.00082	0.000824
...	...	...	...	...	...	...	...
5697	GO:2001256	regulation of store-operated calcium ent...	11	0	0.16	1.00000	1.000000
5698	GO:2001257	regulation of cation channel activity	104	0	1.52	1.00000	1.000000
5699	GO:2001258	negative regulation of cation channel ac...	23	0	0.34	1.00000	1.000000
5700	GO:2001259	positive regulation of cation channel ac...	43	0	0.63	1.00000	1.000000
5701	GO:2001267	regulation of cysteine-type endopeptidas...	13	0	0.19	1.00000	1.000000

5702 rows × 7 columns

Code

results_table_py = results_table_py[results_table_py['Scores'] < 0.05]
results_table_py = results_table_py[results_table_py['Annotated'] < 200]
results_table_py = results_table_py[results_table_py['Annotated'] > 15]

Code

results_table_py['-log10(pvalue)'] = - np.log10(results_table_py.Scores)
results_table_py['Significant/Annotated'] = results_table_py['Significant'] / results_table_py['Annotated']

Code

intTable(results_table_py, folder = folder, fileName = 'GO_BP_all.xlsx', save = True)

Code

%%R -i folder
Res <- GenTable(GOdata, weight=results,
        orderBy="weight", topNodes=length(score(results)))
#print(Res[0:10,])
colnames(Res) <- c("GO.ID", "Term", "Annotated", "Significant", "Expected", "Statistics")
Res$ER <- Res$Significant / Res$Expected

image = bubbleplot(Res, Ont = 'BP', fillCol = 'forestgreen')
ggsave(file=paste0(folder, "TopGO_results_BP.pdf"), plot=image, width=12, height=4)

bubbleplot(Res, Ont = 'BP', fillCol = 'forestgreen')

Code

%%R -i markers
image = plotGenesInTerm_v1(Res, GOdata, SE = markers, nterms=15, ngenes=12,
                             fillCol='forestgreen', log = TRUE)

ggsave(file=paste0(folder, "Genes_in_Term_results_BP.pdf"), plot=image, width=12, height=4)

plotGenesInTerm_v1(Res, GOdata, SE = markers, nterms=15, ngenes=12,
                             fillCol='forestgreen', log = TRUE)

Code

%%R -i markers -i folder
saveGenesInTerm(Res, GOdata, nterms = 20, path = paste0(folder,'GO_BP_genesInTerm_all.xlsx'), SE = markers)

Molecular Function

Code

%%R

GOdata <- new("topGOdata",
  ontology="MF",
  allGenes=allGenes_v,
  annot=annFUN.GO2genes,
  GO2genes=ann_org_MF,
  geneSel = selection,
  nodeSize=10)

Code

%%R -o results

results <- runTest(GOdata, algorithm="weight01",statistic="fisher")

Code

scores = ro.r.score(results)
score_names = ro.r(
'''
names(results@score)
'''
)
go_data = ro.r.GOdata

genesData = ro.r(
'''
geneData(results)
'''
)
genesData

array([11203,   159,    10,   416], dtype=int32)

Code

#num_summarize = min(100, len(score_names))
results_table = ro.r.GenTable(go_data, weight=results,
        orderBy="weight", topNodes=len(scores))

Code

results_table_py = ro.conversion.rpy2py(results_table)

Code

scores_py = ro.conversion.rpy2py(scores)
score_names = [i for i in score_names]

Code

scores_df = pd.DataFrame({'Scores': scores_py, 'GO.ID': score_names})
results_table_py = results_table_py.merge(scores_df, left_on = 'GO.ID', right_on = 'GO.ID')
results_table_py = results_table_py[results_table_py['Scores'] < 0.05]
results_table_py = results_table_py[results_table_py['Annotated'] < 200]
results_table_py = results_table_py[results_table_py['Annotated'] > 15]

intTable(results_table_py, folder = folder, fileName = 'GO_MF_all.xlsx', save = True)

Code

%%R
Res <- GenTable(GOdata, weight=results,
        orderBy="weight", topNodes=length(score(results)))
#print(Res[0:10,])
colnames(Res) <- c("GO.ID", "Term", "Annotated", "Significant", "Expected", "Statistics")
Res$ER <- Res$Significant / Res$Expected

image = bubbleplot(Res, Ont = 'MF', fillCol = 'forestgreen')

ggsave(file=paste0(folder, "TopGO_results_MF.pdf"), plot=image, width=12, height=4)

bubbleplot(Res, Ont = 'MF', fillCol = 'forestgreen')

Code

%%R -i markers
image = plotGenesInTerm_v1(Res, GOdata, SE = markers, nterms=15, ngenes=12,
                             fillCol='forestgreen', log = TRUE)

ggsave(file=paste0(folder, "Genes_in_Term_results_MF.pdf"), plot=image, width=12, height=4)

plotGenesInTerm_v1(Res, GOdata, SE = markers, nterms=15, ngenes=12,
                             fillCol='forestgreen', log = TRUE)

Code

%%R -i markers -i folder
saveGenesInTerm(Res, GOdata, nterms = 20, path = paste0(folder,'GO_MF_genesInTerm_all.xlsx'), SE = markers)

Cellular Compartment

Code

%%R

GOdata <- new("topGOdata",
  ontology="CC",
  allGenes=allGenes_v,
  annot=annFUN.GO2genes,
  GO2genes=ann_org_CC,
  geneSel = selection,
  nodeSize=10)

Code

%%R -o results

results <- runTest(GOdata, algorithm="weight01",statistic="fisher")

Code

scores = ro.r.score(results)
score_names = ro.r(
'''
names(results@score)
'''
)
go_data = ro.r.GOdata

genesData = ro.r(
'''
geneData(results)
'''
)
genesData

array([11323,   161,    10,   336], dtype=int32)

Code

#num_summarize = min(100, len(score_names))
results_table = ro.r.GenTable(go_data, weight=results,
        orderBy="weight", topNodes=len(scores))

Code

results_table_py = ro.conversion.rpy2py(results_table)

Code

scores_py = ro.conversion.rpy2py(scores)
score_names = [i for i in score_names]

Code

scores_df = pd.DataFrame({'Scores': scores_py, 'GO.ID': score_names})
results_table_py = results_table_py.merge(scores_df, left_on = 'GO.ID', right_on = 'GO.ID')

Code

results_table_py = results_table_py[results_table_py['Scores'] < 0.05]
results_table_py = results_table_py[results_table_py['Annotated'] < 200]
results_table_py = results_table_py[results_table_py['Annotated'] > 15]

intTable(results_table_py, folder = folder, fileName = 'GO_CC_all.xlsx', save = True)

Code

%%R
Res <- GenTable(GOdata, weight=results,
        orderBy="weight", topNodes=length(score(results)))
#print(Res[0:10,])
colnames(Res) <- c("GO.ID", "Term", "Annotated", "Significant", "Expected", "Statistics")
Res$ER <- Res$Significant / Res$Expected
image = bubbleplot(Res, Ont = 'CC', fillCol = 'forestgreen')

ggsave(file=paste0(folder, "TopGO_results_CC.pdf"), plot=image, width=12, height=4)

bubbleplot(Res, Ont = 'CC', fillCol = 'forestgreen')

Code

%%R -i markers
image = plotGenesInTerm_v1(Res, GOdata, SE = markers, nterms=12, ngenes=12,
                             fillCol='forestgreen', log = TRUE)

ggsave(file=paste0(folder, "Genes:_in_Term_results_CC.pdf"), plot=image, width=12, height=4)

plotGenesInTerm_v1(Res, GOdata, SE = markers, nterms=12, ngenes=12,
                             fillCol='forestgreen', log = TRUE)

Code

%%R -i markers -i folder
saveGenesInTerm(Res, GOdata, nterms = 20, path = paste0(folder,'GO_CC_genesInTerm_all.xlsx'), SE = markers)

Pathway annotation

4.1.0.1 Reactome

Code

curated = msigdb[msigdb['collection'].isin(['reactome_pathways'])]
curated = curated[~curated.duplicated(['geneset', 'genesymbol'])]

aggregated = curated[["geneset", "genesymbol"]].groupby("geneset").count().rename(columns={"genesymbol": "gene_count"})
curated = curated[~curated.geneset.isin(aggregated[aggregated.gene_count > 200].index.tolist())].copy()
curated = curated[~curated.geneset.isin(aggregated[aggregated.gene_count < 15].index.tolist())].copy()

Code

rank = pd.DataFrame(markers['-log10(FDR)'])

rank_copy = rank.copy()
rank_copy['pval'] = markers.loc[rank.index].FDR

Code

rank_copy

	-log10(FDR)	pval
SERPINE1	203.888176	1.293672e-204
MAGEA4	167.281887	5.225319e-168
PPP1R17	164.492361	3.218391e-165
ZFP57	152.743850	1.803641e-153
MT1H	138.192925	6.413199e-139
...	...	...
AFP	62.510531	3.086517e-63
RTBDN	61.684845	2.066117e-62
TMEM187	61.047779	8.958198e-62
INHBA	59.382685	4.143000e-60
SV2B	58.799588	1.586396e-59

172 rows × 2 columns

Code

results_table_py = run_ora_catchErrors(mat=rank.T, net=curated, source='geneset', target='genesymbol', verbose=False, n_up=len(rank), n_bottom=0)
len(results_table_py)

No significant term was found

0

Code

intTable(results_table_py, folder = folder, fileName = 'Reactome_all.xlsx', save = True)

Code

if len(results_table_py) > 0:
    results_table_py = getAnnGenes(results_table_py, GO2gene['reactome_pathways'], rank_copy)
    _, df = plotGenesInTerm(results = results_table_py, GO2gene = GO2gene['reactome_pathways'], DEGs = rank_copy, n_top_terms = 10, cmap = cmap_all)

Code

if len(results_table_py) > 0:
    intTable(df, folder = folder, fileName = 'genesInTerm_Reactome_all.xlsx', save = True)

4.1.0.2 KEGG

Code

curated = msigdb[msigdb['collection'].isin(['kegg_pathways'])]
curated = curated[~curated.duplicated(['geneset', 'genesymbol'])]

aggregated = curated[["geneset", "genesymbol"]].groupby("geneset").count().rename(columns={"genesymbol": "gene_count"})
curated = curated[~curated.geneset.isin(aggregated[aggregated.gene_count > 200].index.tolist())].copy()
curated = curated[~curated.geneset.isin(aggregated[aggregated.gene_count < 15].index.tolist())].copy()

Code

results_table_py = run_ora_catchErrors(mat=rank.T, net=curated, source='geneset', target='genesymbol', verbose=False, n_up=len(rank), n_bottom=0)

No significant term was found

Code

intTable(results_table_py, folder = folder, fileName = 'KEGG_all.xlsx', save = True)

Code

if len(results_table_py) > 0:
    results_table_py = getAnnGenes(results_table_py, GO2gene['kegg_pathways'], rank_copy)
    _, df = plotGenesInTerm(results_table_py, GO2gene['kegg_pathways'], rank_copy, n_top_terms = 10, n_top_genes = 15, cmap = cmap_all)

Code

if len(results_table_py) > 0:
    intTable(df, folder = folder, fileName = 'genesInTerm_KEGG_all.xlsx', save = True)