help:
	cat Makefile
	
########################################## CNPs genomes ##########################################

af_genomes_imgag_cnp_detetion:
	cp /mnt/storage3/users/ahsturm1/Sandbox/2022_03_25_clincnv_cnps/cn_polymorphisms_3500genomes_imgag.igv af_genomes_imgag_cnp_detetion.igv

af_genomes_imgag_ngsd_export:
	/mnt/storage2/megSAP/tools/ngs-bits-current/NGSDExportCnvTrack -system TruSeqPCRfree -min_af 0.02 -min_dp 30 -max_cnvs 2000 -out af_genomes_imgag_ngsd_export.igv -stats af_genomes_imgag_ngsd_export.log
	
af_genomes_imgag:
	php merge_cnps.php af_genomes_imgag_ngsd_export.igv af_genomes_imgag_cnp_detetion.igv > af_genomes_imgag.igv
	cut -f1,2,3,5 af_genomes_imgag.igv | grep -v "Chromosome" > af_genomes_imgag.bed


########################################## CNPs exomes ##########################################

af_ssHAEv7_imgag:
	/mnt/storage2/megSAP/tools/ngs-bits-current/NGSDExportCnvTrack -system ssHAEv7 -min_af 0.02 -min_dp 80 -max_cnvs 200 -out af_ssHAEv7_imgag.igv -stats af_ssHAEv7_imgag.log
	cut -f1,2,3,5 af_ssHAEv7_imgag.igv | grep -v "Chromosome" > af_ssHAEv7_imgag.bed


########################################## misc ##########################################

cytoBand:
	wget -O - https://hgdownload.soe.ucsc.edu/goldenPath/hg38/database/cytoBand.txt.gz | gunzip > cytoBand.txt

nobase_regions:
	php extract_nobase.php GRCh38 > nobase_regions.bed

pfam_description:
	wget -O - ftp://ftp.ebi.ac.uk/pub/databases/Pfam/current_release/Pfam-A.seed.gz | zcat | grep -e '^#=GF AC' -e '^#=GF DE' | tr -d '\n' | sed 's/#=GF AC   /\n/g' | sed 's/\.[0-9]\+#=GF DE   /\t/g' | sed '/^$$/d' | sed '1 i\#PFAM_ID\tDESCRIPTION' | tr '\n\n' '\n'> pfam_description.tsv
	wget -O - ftp://ftp.ebi.ac.uk/pub/databases/Pfam/current_release/Pfam-A.dead.gz | zcat | grep -e '^#=GF AC' -e '^#=GF FW' | tr -d '\n' | sed 's/#=GF AC   /\n/g' | sed 's/#=GF FW   /\t/g' | sed '/^$$/d' | sed '1 i\#OLD_PFAM_ID\tNEW_PFAM_ID' | tr '\n\n' '\n'> pfam_replacements.tsv
	
telomere_centromere:
	# download Gap (filter: gap.type = 'telomere') and Centromeres table from https://genome.ucsc.edu/cgi-bin/hgTables
	egrep -v "^#" telomere.tsv | cut -f2-4,8 > telomere.bed
	egrep -v "^#" centromere.tsv | cut -f2-5 | BedExtend -n 500000 | BedMerge | BedShrink -n 500000 | awk 'BEGIN {OFS="\t"}; {print $$0, "centromere"}' > centromere.bed 
	cat telomere.bed centromere.bed | BedSort -out centromer_telomer.bed
	rm telomere.bed centromere.bed

circos_karyotype_files:
	echo "These files where created manually by merging the chromosome regions and the centromer_telomer.bed (extended by 3 columns and replaced names by color)"

hemoglobin_transcripts:
	wget -O - 'https://ftp.ensembl.org/pub/release-109/fasta/homo_sapiens/cdna/Homo_sapiens.GRCh38.cdna.all.fa.gz' | \
		zcat | \
		awk -v RS=">" -v FS="\n" '$$1 ~ / gene_symbol:(HBA1|HBA2|HBB) / { print ">"$$1; {for (i=2; i<=NF; i++) printf("%s", $$i)}; printf("\n") }' | \
		sed '/^>/s/ /|kraken:taxid|9606 /' \
		> human_hemoglobin_tx.fa