# 19 December 2010

JAVA=/usr/local/bin/java16


# Trying v4.0.16-dev.interval
FPDIR=/home/afejes/workspace/Vancouver_Package/jars/fp4
FP=$FPDIR/FindPeaks.jar
SEPR=$FPDIR/SeparateReads.jar
SRTF=$FPDIR/SortFiles.jar

WTBW=/home/grobertson/bin/x86_64/wigToBigWig

# have to run this on xhost06, can't find java16 on xhost08 or 09

#BEDDIR=/projects/remc_bigdata/Brad/3T3L1_t1_GSM535743_H3K4me1
#BED=3T3L1_t1_GSM535743_H3K4me1_ChIP-seq.mm9.bed
#NAME=3T3L1_t1_GSM535743_H3K4me1_ChIP-seq.mm9_ht3_q10
#FPWIG=3T3L1_t1_GSM535743_H3K4me1_ChIP-seq.mm9_ht3_q10_triangle_standard

#BEDDIR=/projects/remc_bigdata/Brad/3T3L1_t2_GSM535750_H3K4me1
#BED=3T3L1_t2_GSM535750_H3K4me1_ChIP-seq.mm9.bed
#NAME=3T3L1_t2_GSM535750_H3K4me1_ChIP-seq.mm9_ht3_q10
#WIGNAME=3T3L1_t2_GSM535750_H3K4me1_ChIP-seq.mm9_ht3_q10_triangle_standard

#BEDDIR=/projects/remc_bigdata/Brad/3T3L1_t3_GSM535757_H3K4me1
#BED=3T3L1_t3_GSM535757_H3K4me1_ChIP-seq.mm9.bed
#NAME=3T3L1_t3_GSM535757_H3K4me1_ChIP-seq.mm9_ht3_q10
#WIGNAME=3T3L1_t3_GSM535757_H3K4me1_ChIP-seq.mm9_ht3_q10_triangle_standard

#BEDDIR=/projects/remc_bigdata/Brad/3T3L1_t4_GSM535764_H3K4me1
#BED=3T3L1_t4_GSM535764_H3K4me1_ChIP-seq.mm9.bed
#NAME=3T3L1_t4_GSM535764_H3K4me1_ChIP-seq.mm9_ht3_q10
#WIGNAME=3T3L1_t4_GSM535764_H3K4me1_ChIP-seq.mm9_ht3_q10_triangle_standard

#BEDDIR=/projects/remc_bigdata/Brad/Bcell_GSM537991_H3K4me1
#BED=Bcell_GSM537991_H3K4me1_ChIP-seq_mm9.bed
#NAME=Bcell_GSM537991_H3K4me1_ChIP-seq.mm9_ht3_q10
#WIGNAME=Bcell_GSM537991_H3K4me1_ChIP-seq.mm9_ht3_q10_triangle_standard

#BEDDIR=/projects/remc_bigdata/Brad/BMDM_GSM538019_H3K4me1
#BED=BMDM_GSM538019_H3K4me1_ChIP-seq_mm9.bed
#NAME=BMDM_GSM538019_H3K4me1_ChIP-seq_mm9_ht3_q10
#WIGNAME=BMDM_GSM538019_H3K4me1_ChIP-seq_mm9_ht3_q10_triangle_standard

#BEDDIR=/projects/remc_bigdata/Brad/ES_129JAEC57bl6_H3K4me1
#BED=ES_129JAEC57bl6_H3K4me1_ChIP-seq_mm9.bed
#NAME=ES_129JAEC57bl6_H3K4me1_ChIP-seq_mm9_ht3_q10
#WIGNAME=ES_129JAEC57bl6_H3K4me1_ChIP-seq_mm9_ht3_q10_triangle_standard

#BEDDIR=/projects/remc_bigdata/Brad/ES_C57bl6_H3K4me1
#BED=ES_C57bl6_H3K4me1_ChIP-seq_mm9.bed
#NAME=ES_C57bl6_H3K4me1_ChIP-seq_mm9_ht3_q10
#WIGNAME=ES_C57bl6_H3K4me1_ChIP-seq_mm9_ht3_q10_triangle_standard

#Eland extended: problems

#BEDDIR=/projects/remc_bigdata/Brad/iPS_mef_H3K4me1
#BED=iPS_mef_H3K4me1_ChIP-seq_mm9.bed
#NAME=iPS_mef_H3K4me1_ChIP-seq_mm9_ht3_q10
#WIGNAME=iPS_mef_H3K4me1_ChIP-seq_mm9_ht3_q10_triangle_standard

#BEDDIR=/projects/remc_bigdata/Brad/Liverf_H3K4me1
#BED=Liverf_H3K4me1_ChIP-seq_mm9.bed
#NAME=Liverf_H3K4me1_mm9_ht3_q10
#WIGNAME=Liverf_H3K4me1_mm9_ht3_q10_triangle_standard

#BEDDIR=/projects/remc_bigdata/Brad/Mefs_H3K4me1
#BED=Mefs_H3K4me1_ChIP-seq_mm9.bed
#NAME=Mefs_H3K4me1_ChIP-seq_mm9_ht5_q10
#WIGNAME=Mefs_H3K4me1_ChIP-seq_mm9_ht5_q10_triangle_standard

#BEDDIR=/projects/remc_bigdata/Brad/NP_GSM281693_H3K4me1
#BED=NP_GSM281693_H3K4me1_ChIP-seq_mm9.bed
#NAME=NP_GSM281693_H3K4me1_ChIP-seq_mm9_ht3_q10
#WIGNAME=NP_GSM281693_H3K4me1_ChIP-seq_mm9_ht3_q10_triangle_standard

BEDDIR=/projects/remc_bigdata/Brad/Thiomacs_GSM537986_me1
BED=Thiomacs_GSM537986_H3K4me1_ChIP-seq_mm9.bed
NAME=Thiomacs_GSM537986_H3K4me1_ChIP-seq_mm9_ht3_q10
WIGNAME=Thiomacs_GSM537986_H3K4me1_ChIP-seq_mm9_ht3_q10_triangle_standard


WIGDIR=$BEDDIR/wig
# 'clean' has chrM, not chrMT; no header lines
CLEAN=$FPWIG.clean
CHRM=$FPWIG.chrM.wig
BWFILE=$WIGNAME.bigwig


# http://sourceforge.net/apps/mediawiki/vancouvershortr/index.php?title=WorkFlows#Work_Flow_-_BED
#you@machine:/path$ java -Xmx2G -jar /path/to/fp4/SeparateReads.jar bed ./input.bed out/
#you@machine:/path$ java -Xmx2G -jar /path/to/jars/fp4/SortFiles.jar bed out/ out/*.part.bed.gz
#you@machine:/path$ mkdir wig
#you@machine:/path$ java -Xmx2G -jar /home/fejes/workspace/Vancouver_Package/jars/fp4/FindPeaks.jar \
#-input out/*.part.bed.gz -aligner bed -output wig/ -dist_type 1 -name example -one_per

echo "separate reads"
echo "$JAVA -Xmx6G -jar $SEPR interval $BEDDIR/$BED out/"
$JAVA -Xmx6G -jar $SEPR interval $BEDDIR/$BED out/

echo "sort reads"
echo "$JAVA -Xmx6G -jar $SRTF interval out/ out/*.part.interval.gz"
$JAVA -Xmx6G -jar $SRTF interval out/ out/*.part.interval.gz
#mkdir $OUTDIR/wig

echo "findpeaks in 'interval' mode"
echo "$JAVA -Xmx6G -jar $FP -input out/*.part.interval.gz -aligner interval -output wig/ -dist_type 1 200 -name $NAME -qualityfilter 10 -duplicatefilter -minimum 3"
$JAVA -Xmx6G -jar $FP -input out/*.part.interval.gz -aligner interval -output wig/ -dist_type 1 200 -name $NAME -qualityfilter 10 -duplicatefilter -minimum 3


#=============================================================================
#echo "changing to 'chrMT' from 'chrM', removing header lines..."
#less $OUTDIR/$FPWIG.gz | sed 's/chrom=MT/chrom=chrM/' | sed '1,3d' > $OUTDIR/$CLEAN
#echo "less $OUTDIR/$FPWIG.gz | sed 's/chrom=MT/chrom=chrM/' > $OUTDIR/$CHRM"
#less $OUTDIR/$FPWIG.gz | sed 's/chrom=MT/chrom=chrM/' > $OUTDIR/$CHRM
#echo "gzipping"
#gzip $OUTDIR/$CHRM

#=============================================================================

#CHROMINFO=/projects/mapp_etc/Mappability/fastq/mm9/mm9.chromInfo.txt
CHROMINFO=/home/grobertson/bin/x86_64/mm9.chrom.sizes
#CHROMINFO=/home/grobertson/bin/x86_64/mm8.chrom.sizes
#GENOMEFA=/projects/mapp_etc/Mappability/fastq/mm9/36mer/bwa_ind/mm9_build37_mouse.fasta
# 'chr1' is '1'
#GENOMEFA=/projects/wtsspipeline/resources/Mus_musculus/bfa_NCBI-37/all_mouse_no_chr.fasta
GENOMEFA=/projects/wtsspipeline/resources/Mus_musculus/bfa_NCBI-37/all_mouse.fasta
echo "sed '1,3d'..."
echo "less $WIGDIR/$WIGNAME.wig.gz | sed '1,3d' > $WIGDIR/$WIGNAME.sed13.wig"
less $WIGDIR/$WIGNAME.wig.gz | sed '1,3d' > $WIGDIR/$WIGNAME.sed13.wig
echo "gzipping..."
gzip $WIGDIR/$WIGNAME.sed13.wig

echo "done"; 
echo "running wigToBigWig... (~/bin/x86_64/wigToBigWig)"
echo "$WTBW -clip $WIGDIR/$WIGNAME.sed13.wig.gz $CHROMINFO $WIGDIR/$BWFILE"
$WTBW -clip $WIGDIR/$WIGNAME.sed13.wig.gz $CHROMINFO $WIGDIR/$BWFILE