makefiles/bwa.mk

   1 INPUTPATH := mda mlbac
   2 OUTPUTPREFIX := out
   3
   4 ALNARG:=aln -l 17 -q 10
   5 SAMPEARG:=sampe -a 800
   6 SAMSORTARG:=sort -l 9
   7 REF:=ref/hg19p10XYM
   8
   9 INF := $(patsubst %_1.fastq.gz,%,$(foreach ONEINPUTPATH,$(INPUTPATH),$(wildcard $(ONEINPUTPATH)/*_1.fastq.gz)))
  10
  11 OUTF := $(addprefix $(OUTPUTPREFIX)/,$(INF))
  12 SAI1 := $(addsuffix _1.sai,$(OUTF))
  13 SAI2 := $(addsuffix _2.sai,$(OUTF))
  14 SAMS := $(addsuffix .sam.gz,$(OUTF))
  15 BAMS := $(addsuffix .bam,$(OUTF))
  16 BAMSORT := $(addsuffix .sort.bam,$(OUTF))
  17 BAMRMDUP := $(addsuffix .rmdup.bam,$(OUTF))
  18
  19 PATHS := $(addprefix $(OUTPUTPREFIX)/,$(MDAPATH) $(MLBACPATH))
  20
  21 NUMPROC := $(shell grep -c ^processor /proc/cpuinfo)
  22 SAMCOUNT := $(words $(SAI1))
  23 SAITHREADS := $(shell echo 1.5*$(NUMPROC)/$(SAMCOUNT) |bc)
  24 ifneq ($(SAITHREADS),)
  25         ifneq ($(SAITHREADS),0)
  26                 ifneq ($(SAITHREADS),1)
  27                         ALNARG += -t $(SAITHREADS)
  28                  endif
  29         endif
  30 endif
  31 FREEMEM := $(shell free -m|grep -e '-/+ buffers/cache'|awk '{print $$NF}')
  32 SAMMEM := $(shell echo 0.8*$(FREEMEM)/$(SAMCOUNT)/$(SAITHREADS) |bc)
  33
  34 NEEDED_COMMANDS := bc bwa samtools grep mkdir free gzip
  35
  36 all: $(OUTF)
  37         @echo "[$(OUTF)]" "[$(INF)]" "[$@]" "[$(SAMMEM) $(FREEMEM)]"
  38         date > $(OUTPUTPREFIX)/_alldone.log
  39
  40 check:
  41         @for thecmd in $(NEEDED_COMMANDS); do \
  42         if ! command -v "$${thecmd%% *}" >/dev/null 2>&1; then \
  43                         checkok="0"; \
  44                         echo "[x]'$${thecmd%% *}' not found."; \
  45                 fi; \
  46         done; \
  47         if [ "$${checkok}" == "0" ]; then \
  48                 echo "[!]Please install missing cmd(s) above."; \
  49                 exit -1; \
  50         fi;
  51
  52 $(PATHS): check
  53         mkdir -p $@
  54
  55 $(OUTF): $(PATHS) $(BAMRMDUP)
  56         $(eval IN := $(patsubst $(OUTPUTPREFIX)/%,%,$@))
  57         echo "$@" "$<" "$(IN)"
  58         touch $@
  59
  60 $(SAI1) $(SAI2): | $(PATHS)
  61         $(eval IN := $(patsubst $(OUTPUTPREFIX)/%,%,$(@:%.sai=%)))
  62         bwa $(ALNARG) $(REF) $(IN).fastq.gz >$@ 2>$@.log && \
  63         date >>$@.log && \
  64         echo done. >>$@.log
  65
  66 $(SAMS): $(SAI1) $(SAI2)
  67         $(eval IN := $(@:%.sam.gz=%))
  68         $(eval FQ := $(patsubst $(OUTPUTPREFIX)/%,%,$(IN)))
  69         bwa $(SAMPEARG) $(REF) $(IN)_1.sai $(IN)_2.sai $(FQ)_1.fastq.gz $(FQ)_2.fastq.gz 2>$(IN).sam.log |gzip -9c >$@ && \
  70         date >>$(IN).sam.log && \
  71         echo done. >>$(IN).sam.log
  72
  73 $(BAMS): $(SAMS)
  74         samtools view -bS $< >$@ 2>$@.log && \
  75         date >>$@.log && \
  76         echo "done (sam.gz to bam)">>$@.log
  77
  78 $(BAMSORT): $(BAMS)
  79         if [ "$(SAMMEM)" -gt "768" ]; then \
  80                 SAMSORTMEM="-@ $(SAITHREADS) -m $(SAMMEM)M";\
  81         else \
  82                 SAMSORTMEM=;\
  83         fi;\
  84         samtools $(SAMSORTARG) $$SAMSORTMEM $< $(@:%.bam=%) >>$<.log && \
  85         date >>$<.log && \
  86         echo "done (bam sort)">>$<.log
  87
  88 $(BAMRMDUP): $(BAMSORT)
  89         $(eval LOG := $(@:%.rmdup.bam=%.bam.log))
  90         samtools rmdup $< $@ >>$(LOG) && \
  91         date >>$(LOG) && \
  92         echo "done (bam rmdup)">>$(LOG)
  93
  94 help: check
  95         @if [ "$(NUMPROC)" -gt "$(SAMCOUNT)"  ]; then \
  96                 JOBCNT="$(SAMCOUNT)";\
  97         fi;\
  98         echo -e "Usage: make -j $(SAMCOUNT) |tee make.log\n\n$(NUMPROC) core(s) & $(FREEMEM) mb free memory found now.\nBWA will run sai with [$(ALNARG)] for your $(SAMCOUNT) bam file(s)."
  99         @if [ "$(SAMMEM)" -gt "768" ]; then \
 100                 echo -e "SAMTOOLS will sort with [-@ $(SAITHREADS) -m $(SAMMEM)M], if free memory remains same when you type make."; \
 101         fi
 102
 103 clean:
 104         rm -fr out/*
 105
 106 .PHONY: help clean check