#Enrichment analysis of chromatin accessibility in transcription start sites\
#and its relationship to G4s. Accessible chromatin regions were\
#retrieved from ChIP-ATLAS (https://chip-atlas.org) using the following\
#parameters: antigen class: DNase-Seq; cell type class: blood; threshold for\
#significance: 500; antigen: DNase-Seq; cell type: all.

sortBed -i DNS.Bld.50.DNase-Seq.AllCell.bed | mergeBed\
> DNS.Bld.50.DNase-Seq.AllCell.sort.merged.only.bed



#A hg19 transcription start sites (TSS) file was generated as previously \
#described (Hänsel-Hertsch et al. 2016).

#  + Strand: Get start of first exons on each transcript
awk -v OFS="\t" '$7 == "+" {print $1, $4, $5, $7, $10, $12}' genes.gtf \
| sed 's/"//g' | sed 's/ //g' | sed 's/;//g' \
| sort -k 6,6 -k2,2n \
| groupBy -g 6 -c 1,2,5 -o first,first,first \
| awk -v OFS="\t" '{print $2, $3-1000, $3+1000, $4, $1, "+"}' > tss.plus.bed

# - strand: Get end of last exons on each transcript
awk -v OFS="\t" '$7 == "-" {print $1, $4, $5, $7, $10, $12}' genes.gtf \
| sed 's/"//g' | sed 's/ //g' | sed 's/;//g' \
| sort -k 6,6 -k3,3nr \
| groupBy -g 6 -c 1,3,5 -o first,first,first \
| awk -v OFS="\t" '{print $2, $3-1000, $3+1000, $4, $1, "-"}' > tss.minus.bed

# Merge promoters within genes
cat tss.minus.bed tss.plus.bed \
| cut -f1,2,3,4 \
| sort \
| uniq \
| awk -v OFS="\t" '{print $1 "_" $4, $2, $3, $4}' \
| sortBed \
| mergeBed \
| sed 's/_/\t/' \
| awk -v OFS="\t" '{print $1, $3, $4, $2}' \
| sortBed > hg19.gene_name.promoters.bed



#A maximum loop length of 7 nucleotides, 4 G-repeats \
#and a minimum of 3 Gs per G-repeat were considered to predict PQS in the human\
#genome (hg19). The fastaRegexFinder tool (Python) was used with the following\
#parameters to predict PQS: “--quiet -r '([gG]{3,}\w{1,7}){3,}[gG]{3,}'”.


#install fastaRegexFinder --> Python tool

fastaRegexFinder.py --quiet -r '([gG]{3,}\w{1,7}){3,}[gG]{3,}' -f\
genome.fa >> PQS_N1-7_hg19.bed"


#The 4 available OQS files were retrieved \
#(Chambers et al. 2015) and unified to a single OQS file using cat and the\
#bedtools functions mergeBed and sortBed (Quinlan and Hall 2010).

cat Na_K_minus_hits_intersect.bed Na_K_plus_hits_intersect.bed\
Na_PDS_minus_hits_intersect.bed Na_PDS_plus_hits_intersect.bed\
| mergeBed -i | sortBed -i > SortBed_on_G4_seq_hits.bed


#Bedtools intersect was used to retrieve ~300,000 PQS that overlap with OQS,\
(PQS/OQS).

bedtools intersect -u -a PQS_N1-7_hg19.bed -b SortBed_on_G4_seq_hits.bed\
> PQS_N1-7_hg19.OQS.bed


#Bedtools intersect was used to retrieve accessible \
#chromatin regions in TSS that are either positive or negative for PQS/OQS.

intersectBed -u -a DNS.Bld.50.DNase-Seq.AllCell.sort.merged.only.bed -b\
hg19.gene_name.promoters.bed\
> DNS.Bld.50.DNase-Seq.AllCell.sort.merged.only.int.TSS.bed

intersectBed -v -a DNS.Bld.50.DNase-Seq.AllCell.sort.merged.only.int.TSS.bed\
-b PQS_N1-7_hg19.in.OQS.sort.bed\
> DNS.Bld.50.DNase-Seq.AllCell.sort.merged.only.int.TSS.negative.for.PQS.OQS.bed

intersectBed -u -a DNS.Bld.50.DNase-Seq.AllCell.sort.merged.only.int.TSS.bed\
-b PQS_N1-7_hg19.in.OQS.sort.bed\
> DNS.Bld.50.DNase-Seq.AllCell.sort.merged.only.int.TSS.int.PQS.OQS.bed


#Bedtools shuffle was employed to generate 10 random distributions of\
#accessible chromatin regions in TSS that are either positive or negative for\
#PQS/OQS regions across the hg19 genome.

n=(1 2 3 4 5 6 7 8 9 10); for f in *bed; do for i in "${!n[@]}"; do bedtools shuffle -incl hg19.wgEncodeDukeMapabilityRegionsExcludable.whitelist.sort.bed -chromFirst -noOverlapping -seed ${n[i]} -i $f -g hg19.genome | sort -k1,1 -k2,2n > ${f%%.bed}.shuffle_${n[i]}.bed; done; done



#The plotEnrichment function of deepTools (Ramírez et al. 2016) was used to\
#measure the coverage of the individual bam files in 11 bed files per\
#condition (1 bed file + 10 randomised bed files). For each bam file,\
#enrichments in accessible chromatin regions in TSS that contain or lack\
#PQS/OQS were calculated using the average ratio of the real \
#coverage divided by 10 different coverages in the permutated accessible\
#chromatin regions in TSS that contain or lack PQS/OQS.

for f in *.bam; do samtools index -b $f; done

plotEnrichment -b *.bam --BED *bed -o plotEnrichment.png --variableScales\
--outRawCounts plotEnrichment.txt



#Similarity of predicted (PQS/OQS) G4-forming chromatin accessibility in\
#transcription start sites with G4 regions identified in chromatin. To assess\
#whether accessible chromatin regions in TSS contain PQS/OQS that also adopt\
#G4 secondary structure in chromatin, we first assembled (cat, sortBed,\
#mergeBed) various annotated chromatin G4 regions (chromatin G4Rs), which were\
#mapped by G4-ChIP-seq using BG4 (GSE145090, GSE99205, GSE152216) or G4P\
#(GSE133379).

#download the following files:
GSE133379_293T-G4P-hg19-rep1.narrowPeak
GSE133379_H1975-G4P-hg19-rep1.narrowPeak
GSE145090_20180108_K562_async_rep1-3.mult.5of8.bed
GSE133379_293T-G4P-hg19-rep2.narrowPeak
GSE133379_H1975-G4P-hg19-rep2.narrowPeak
GSE145090_HepG2_async_rep1-3.mult.2of9.bed
GSE133379_A549-G4P-hg19-rep1.narrowPeak
GSE133379_HeLa-S3-G4P-hg19-rep1.narrowPeak
GSE145090_HepG2_async_rep1-3.mult.6of9.bed
GSE133379_A549-G4P-hg19-rep2.narrowPeak
GSE145090_20180108_K562_async_rep1-3.mult.2of8.bed
GSE99205_common_HaCaT_G4_ChIP_peaks_RNase_treated.bed

#Retrieve all G4 regions identfied in breast cancer tissues.
#download https://github.com/sblab-bioinformatics/qG4-ChIP-seq-of-breast\
#-cancer-PDTX/blob/master/input_files/temp_concat_hg19_libraries.txt

cut -f 1,2,3 temp_concat_hg19_libraries.txt > temp_concat_hg19_libraries.bed

cat GSE133379_293T-G4P-hg19-rep1.narrowPeak\
GSE133379_H1975-G4P-hg19-rep1.narrowPeak\
GSE145090_20180108_K562_async_rep1-3.mult.5of8.bed\
GSE133379_293T-G4P-hg19-rep2.narrowPeak\
GSE133379_H1975-G4P-hg19-rep2.narrowPeak\
GSE145090_HepG2_async_rep1-3.mult.2of9.bed\
GSE133379_A549-G4P-hg19-rep1.narrowPeak\
GSE133379_HeLa-S3-G4P-hg19-rep1.narrowPeak\
GSE145090_HepG2_async_rep1-3.mult.6of9.bed\
GSE133379_A549-G4P-hg19-rep2.narrowPeak\
GSE145090_20180108_K562_async_rep1-3.mult.2of8.bed\
GSE99205_common_HaCaT_G4_ChIP_peaks_RNase_treated.bed\
temp_concat_hg19_libraries.bed | sortBed | mergeBed > insitu.G4Rs.bed


#Second, bedtools intersect was used to retrieve ~81,000 PQS that overlap with\
#chromatin G4Rs (PQS/chromatin.G4Rs).

intersectBed -u -a PQS_N1-7_hg19.bed -b insitu.G4Rs.bed > PQS.int.insituG4Rs.bed

#Third, bedtools intersect was used to retrieve accessible chromatin regions\
#in TSS that are positive for PQS/chromatin.G4Rs.

intersectBed -u -a DNS.Bld.50.DNase-Seq.AllCell.sort.merged.only.int.TSS.bed\
-b PQS.int.insituG4Rs.bed > OC.int.TSS.int.PQS.int.insituG4Rs.bed


#Fourth, we used intervene venn to assess the similarity between accessible\
#regions in TSS that are either a) positive or b) negative for PQS/OQS, and\
#c) positive for PQS/chromatin.G4Rs.

intervene venn -i OC.int.TSS.int.PQS.int.insituG4Rs.bed\
OC.int.TSS.int.PQS.OQS.bed OC.int.TSS.negative.for.PQS.OQS.bed\
--names=OC.TSS.PQS.insituG4Rs,OC.TSS.PQS.OQS,OC.TSS.Neg.PQS.OQS --save-overlaps


#Enrichment analysis of transcription start sites containing either cancer or\
#normal blood-related accessible chromatin regions.Accessible chromatin regions\
#were retrieved from ChIP-ATLAS (https://chip-atlas.org) using the following\
#parameters: antigen class: DNase-Seq; cell type class: blood; threshold for\
#significance: 500; antigen: All; cell type: eythrocyte.progenitors (SRX062346\
#,SRX027080, SRX2187113, SRX201280, SRX089247, SRX027082, SRX027081, SRX015226,\
#SRX026926, SRX026923, SRX026927, SRX015228, SRX027077, SRX015227, SRX026920,\
#SRX026922, SRX062365, SRX026915, SRX214040, SRX026921, SRX062345, SRX040374,\
#SRX062347, SRX4457261, SRX4457260, SRX4457262, SRX4457263, SRX4457266,\
#SRX4457267, SRX4457265, SRX4457264), Erythroid cells (SRX2105479, SRX2105477,\
#SRX2105478, SRX838173, SRX2370592, SRX2370809, SRX2370805, SRX2370710,\
#SRX2370651, SRX2370789, SRX2370806, SRX2370593, SRX2370676, SRX2370711,\
#SRX2370602, SRX2370601, SRX2370642, SRX2370810, SRX2370641, SRX2370788,\
#SRX2370652), lymphocytes (SRX201251, SRX201254, SRX193595, SRX20130\
#, SRX069095, SRX201250, SRX193593, SRX201249, SRX193594, SRX069185, SRX201270,\
#SRX201259, SRX193612, SRX201263, SRX037118, SRX201262, SRX193613, SRX201300,\
#SRX5718539, SRX5718538, SRX040389, SRX100956, SRX055204, SRX040414, SRX204403,\
#SRX055163, SRX213515, SRX040396, SRX055172, SRX204402, SRX040412, SRX189431,\
#SRX040388, SRX055153, SRX100962, SRX201299, SRX040415, SRX193609, SRX055203,\
#SRX055155, SRX204404, SRX214041, SRX055180, SRX055190, SRX055152, SRX040413,\
#SRX089240, SRX055189, SRX204362, SRX252603, SRX201292, SRX189402, SRX055164,\
#SRX201276, SRX201275, SRX193597, SRX189409, SRX189387, SRX055157, SRX055171,\
#SRX055156, SRX342324) and monocytes (SRX2370645, SRX2370646, SRX581585,\
#SRX581580, SRX581582, SRX581593, SRX581578, SRX581579, SRX581576, SRX581577,\
#SRX581589, SRX581581, SRX581584, SRX581595, SRX581594, SRX581590, SRX581588,\
#SRX201301, SRX055167, SRX040416, SRX252602, SRX055205, SRX069106, SRX581591,\
#SRX581587, SRX581586). DNase-seq peaks of each experiment (SRX…) within either\
#the eythrocyte.progenitors or lymphocytes categories were sorted (sortBed),\
#merged (mergeBed) and filtered for unique peaks that only appear in the\
#respective categories (intersectBed). Lymphocyte and erythrocyte progenitors\
#TSSs were selected from overlaps with their specific unique accessible\
#chromatin peaks (intersectBed). We performed the same filtering process to\
#reveal monocyte related TSS but due to the substantially reduced number of\
#TSS (259) relative to lymphocyte (1971 TSS) and erythrocyte progenitors\
#(878 TSS), ignored them for further analysis.

cat\
DNS.Bld.50.AllAg.CD34PULUS.bed\
DNS.Bld.50.AllAg.CD34PULUS_cells.bed\
DNS.Bld.50.AllAg.Erythroid_Cells.bed\
DNS.Bld.50.AllAg.Hematopoietic_Stem_Cells.bed\
DNS.Bld.50.AllAg.Peripheral_blood_stem_cells.bed | sortBed | mergeBed\
> eythrocyte.progenitors_DHS_hg19.bed

cat\
DNS.Bld.50.AllAg.B-cell_BRACKETLCD19PULUSBRACKETR.bed\
DNS.Bld.50.AllAg.B-Lymphocytes.bed\
DNS.Bld.50.AllAg.CD20PULUS_B_cells.bed\
DNS.Bld.50.AllAg.CD20PULUS.bed\
DNS.Bld.50.AllAg.CD3PULUS.bed\
DNS.Bld.50.AllAg.CD4-Positive_T-Lymphocytes.bed\
DNS.Bld.50.AllAg.CD4PULUS.bed\
DNS.Bld.50.AllAg.CD4PULUS_Th1.bed\
DNS.Bld.50.AllAg.CD56PULUS.bed\
DNS.Bld.50.AllAg.CD8PULUS.bed\
DNS.Bld.50.AllAg.PBMC.bed\
DNS.Bld.50.AllAg.Th17_Cells.bed\
DNS.Bld.50.AllAg.Th1_Cells.bed\
DNS.Bld.50.AllAg.Th2_Cells.bed\
DNS.Bld.50.AllAg.Treg.bed\
DNS.Bld.50.AllAg.Treg_Wb83319432.bed | sortBed | mergeBed\
> lymphocytes_DHS_hg19.bed

cat\
DNS.Bld.50.AllAg.Dendritic_Cells.bed\
DNS.Bld.50.AllAg.Macrophages.bed\
DNS.Bld.50.AllAg.Monocytes.bed\
DNS.Bld.50.AllAg.Monocytes-CD14PULUS.bed | sortBed | mergeBed\
> monocytes_DHS_hg19.bed

cat monocytes_DHS_hg19.bed lymphocytes_DHS_hg19.bed | sortBed | mergeBed\
> tmp && intersectBed -v -a eythrocyte.progenitors_DHS_hg19.bed -b tmp\
> uniqe.eythrocyte.progenitors_DHS_hg19.bed && rm tmp && intersectBed\
-u -a hg19.gene_name.promoters.bed -b uniqe.eythrocyte.progenitors_DHS_hg19.bed\
> TSS.int.unique.eythrocyte.progenitors_DHS_hg19.bed

cat monocytes_DHS_hg19.bed eythrocyte.progenitors_DHS_hg19.bed | sortBed\
| mergeBed > tmp && intersectBed -v -a lymphocytes_DHS_hg19.bed -b tmp\
> uniqe.lymphocytes_DHS_hg19.bed && rm tmp && intersectBed -u -a\
hg19.gene_name.promoters.bed -b uniqe.lymphocytes_DHS_hg19.bed\
> TSS.int.unique.lymphocytes_DHS_hg19.bed


#To reveal TSS that contain pan cancer accessible chromatin, a genomic\
#coordinate (bed, genome alignment to hg38) pan cancer ATAC-seq file was\
#retrieved from https://gdc.cancer.gov/about-data/publications/ATACseq-AWG,\
#sorted (sortBed),lifted to hg19 (https://genome.ucsc.edu/cgi-bin/hgLiftOver),\
#sorted and TSS (intersectBed) captured to reveal cancer-specific TSS. To\
#compare a similar number of pan cancer related TSS with blood related TSS\
#(Lymphocyte and erythrocyte progenitors), we selected the top 5,000 enriched\
#ATAC-seq peaks by its enrichment score (The normalized peak score of the given\
#peak (i.e. "score-per-million", see methods of Corces & Granja et al Science\
#2018 publication)). Bedtools shuffle was employed to generate, respectively,\
#10 random distributions of the 3 TSS categories (top 5K pan cancer,\
#lymphocytes, erythrocyte progenitors). The plotEnrichment function of\
#deepTools (Ramírez et al. 2016) was used to measure the coverage #of the\
# individual bam files in 11 bed files per condition (1 bed file + 10\
#randomised bed files). For each bam file, enrichments in the 3 TSS\
#categories were calculated using the average ratio of the real coverage\
#divided by 10 different coverages in the permutated TSS datasets.


#download https://api.gdc.cancer.gov/data/116ebba2-d284-485b-9121-faf73ce0a4ec

#-->TCGA-ATAC_PanCancer_PeakSet.txt

intersectBed -u -a hg19.gene_name.promoters.bed -b\
TCGA-ATAC_PanCancer_PeakSet.top5k.sort.clean.hg19.sort.bed\
> TSS.plus.minus.1kbp.int.TCGA.ATAC.PanCancer.top5kpeaks.bed

n=(1 2 3 4 5 6 7 8 9 10); for f in\
TSS.plus.minus.1kbp.int.TCGA.ATAC.PanCancer.top5kpeaks.bed\
TSS.int.unique.lymphocytes_DHS_hg19.bed\
TSS.int.unique.eythrocyte.progenitors_DHS_hg19.bed; do for\
i in "${!n[@]}"; do bedtools shuffle -incl\
hg19.wgEncodeDukeMapabilityRegionsExcludable.whitelist.sort.bed -chromFirst\
-noOverlapping -seed ${n[i]} -i $f -g hg19.genome | sort -k1,1 -k2,2n\
> ${f%%.bed}.shuffle_${n[i]}.bed; done; done

plotEnrichment -b *.bam --BED *bed -o plotEnrichment.png --variableScales\
--outRawCounts plotEnrichment.txt