Homec4science
Diffusion scATAC-seq (master) Browse

scATAC-seq/scripts/10xgenomics_PBMC_5k_motifs/analysis_ctcf_sp1_motif.sh
411e7adaffa4master

Last Change
Raw File

analysis_ctcf_sp1_motif.sh
View Options

# some paths
## directories
results_dir='data/10xgenomics_PBMC_5k_motifs'
read_dir="data/10xgenomics_PBMC_5k"
seq_dir="data/genomes"
## input
file_bed_ctcf=$read_dir'/ctcf_motifs_10e-6_rmsk.bed'
file_bed_sp1=$read_dir'/sp1_motifs_10e-7_rmsk.bed'
file_bed=$read_dir'/ctcf_motifs_10e-6_sp1_motifs_10e-7_rmsk.bed'
file_bam_open="$read_dir/atac_v1_pbmc_5k_possorted_filtered_30-84bp.bam"
file_bai_open="$read_dir/atac_v1_pbmc_5k_possorted_filtered_30-84bp.bam.bai"
file_bam_nucl="$read_dir/atac_v1_pbmc_5k_possorted_filtered_nucleosomes.bam"
file_bai_nucl="$read_dir/atac_v1_pbmc_5k_possorted_filtered_nucleosomes.bam.bai"
file_hg19="$seq_dir/hg19.fasta"
file_rmsk="$seq_dir/hg19_rmsk.bed"
mkdir -p $results_dir
# merge 5000 CTCF peaks and 5000 SP1 peaks
touch $file_bed
shuf $file_bed_ctcf | head -n 5000 >> $file_bed
shuf $file_bed_sp1 | head -n 5000 >> $file_bed
# matrix creation
## sequences and sequenced repeat masked
file_mat_seq="$results_dir/ctcf_motifs_10e-6_sp1_motifs_10e-7_sequences_rmsk.mat"
bin/SequenceMatrixCreator --bed $file_bed --fasta $file_hg19 --from -400 --to 400 > $file_mat_seq
## open chromatin around motifs
for method in 'read_atac'
do
file_mat_open_1="$results_dir/ctcf_motifs_10e-6_sp1_motifs_10e-7_open_bin1bp_"$method"_rmsk.mat"
bin/CorrelationMatrixCreator --bed $file_bed --bam $file_bam_open --bai $file_bai_open --from -400 --to 400 --binSize 1 --method $method > $file_mat_open_1
done
## all nucleosomes around motifs
for method in 'fragment_center'
do
### mono nucleosomes
file_mat_nucl_1="$results_dir/ctcf_motifs_10e-6_sp1_motifs_10e-7_nucleosomes_bin1bp_"$method"_rmsk.mat"
bin/CorrelationMatrixCreator --bed $file_bed --bam $file_bam_nucl --bai $file_bai_nucl --from -400 --to 400 --binSize 1 --method $method > $file_mat_nucl_1
done
c4science · Help