biostars9571322 Mafile based ngs worfklow for tutorial

Makefile

# input is
#  DATA/father.R1.fq.gz
#  DATA/father.R2.fq.gz
#  DATA/mother.R1.fq.gz
#  DATA/mother.R2.fq.gz
#  DATA/child.R1.fq.gz
#  DATA/child.R2.fq.gz
#
define mapreads

$(1).bam.bai : $(1).bam 
	samtools index $$<

$(1).bam : $(1).unsorted.bam 
	samtools sort -o $$@ -T $$(basename $$@) $$<

$(1).unsorted.bam :$(1).sam
	samtools view -Sb -o $$@ $$< 

$(1).sam : DATA/$(1).R1.fq.gz DATA/$(1).R2.fq.gz ref.fa.bwt
	bwa mem   -R '@RG\tID:$$(basename $$@)\tSM:$(1)' ref.fa $$(filter %.fq.gz,$$^) > $$@



endef
all: result.annot.vcf.gz.tbi 

result.annot.vcf.gz.tbi: result.annot.vcf.gz
	bcftools index $<

result.annot.vcf.gz: result.vcf.gz annot.gff.gz ref.fa.fai
	bcftools csq --phase a --fasta-ref ref.fa --gff-annot annot.gff.gz -O b -o $@ $< 

annot.gff.gz:
	wget -O - "ftp://ftp.ensembl.org/pub/grch37/release-99/gff3/homo_sapiens/Homo_sapiens.GRCh37.87.chromosome.22.gff3.gz" | gunzip -c | sed 's/^22/chr22/' | gzip > $@


result.vcf.gz.tbi: result.vcf.gz
	tabix --force --preset vcf $<

result.vcf.gz: result.bcf
	bcftools call  --multiallelic-caller --variants-only --output-type z -o $@ --format-fields GQ,GP $< 

result.bcf : $(addsuffix .bam.bai,child father mother) $(addsuffix .bam,child father mother) ref.fa.fai
	bcftools mpileup --fasta-ref ref.fa --output-type b --output $@ -a 'FORMAT/AD' -a 'FORMAT/DP' $(filter %.bam,$^)
	

$(eval $(call mapreads,father))
$(eval $(call mapreads,child))
$(eval $(call mapreads,mother))

ref.fa.bwt: ref.fa
	bwa index $<

ref.fa.fai: ref.fa
	samtools faidx $< 
ref.fa:
	wget -O -  "http://hgdownload.cse.ucsc.edu/goldenPath/hg19/chromosomes/chr22.fa.gz" | gunzip -c > $@
	wget -O -  "http://hgdownload.cse.ucsc.edu/goldenPath/hg19/chromosomes/chrM.fa.gz" | gunzip -c >> $@

clean:
	rm -f ref.* child.* father.* mother.* result.*