Merge pull request #81 from ARUP-NGS/dev

Zlib issues
ARUP-NGS · May 12, 2016 · 73a8c4e · 73a8c4e
2 parents 4ccdd60 + 6a1abf7
commit 73a8c4e
Show file tree

Hide file tree

Showing 6 changed files with 18 additions and 9 deletions.
diff --git a/sample_test/compile.sh b/sample_test/compile.sh
diff --git a/scripts/secondary_pipeline.sh b/scripts/secondary_pipeline.sh
@@ -30,6 +30,7 @@ TMPFQ=${tmpstr%.fastq*}.tmp.fq
 TMPBAM=${tmpstr%.fastq*}.tmp.bam
 PRERSQBAM=${tmpstr%.fastq*}.prersq.bam
 FINALBAM=${tmpstr%.fastq*}.rsq.bam
+FASTQC="/uufs/chpc.utah.edu/common/home/arup-storage1/tools/FastQC/"
 
 # Perform inline barcode demultiplexing.
 echo time bmftools sdmp -zdp${THREADS} -s${SALT} -n${PREFIX_LEN} -i $rindex -o${TMP_PREF} $r1 $r2 -f${FINAL_FQ_PREFIX}
@@ -41,14 +42,13 @@ echo Number of collapsed observations after dmp: $(zgrep -c '^+$' $R1) >> $LOG
 
 mkdir -p _FASTQC_${tmpstr}
 mkdir -p _FASTQC_DMP_${tmpstr}
-fastqc -t ${THREADS} --nogroup -o _FASTQC_${tmpstr} $1 $2
-fastqc -t ${THREADS} --nogroup -o _FASTQC_DMP_${tmpstr} $R1 $R2
+$FASTQC -t ${THREADS} --nogroup -o _FASTQC_${tmpstr} $1 $2
+$FASTQC -t ${THREADS} --nogroup -o _FASTQC_DMP_${tmpstr} $R1 $R2
 
 # There are a lot of processes here. We save a lot of time by avoiding I/O by piping.
 bwa mem -CYT0 -t${THREADS} $REF $R1 $R2 | samtools view -Sbh - | bmftools mark -l 0 | \
     bmftools sort -o${PRERSQBAM} -l6 -m 6G -@ 4 -k ucs -T tmpfileswtf -
 
-getsums.py $PRERSQBAM >>$LOG &
 echo Post-BMF-sort read count -cF2816: $(samtools view -cF2816 $PRERSQBAM) >> $LOG &
 echo Post-BMF-sort read count -cF2304: $(samtools view -cF2304 $PRERSQBAM) >> $LOG &
 
@@ -63,12 +63,9 @@ echo Post-rescue, before merge read count -cF2304: $(samtools view -cF2304 $TMPB
 bwa mem -pCYT0 -t${THREADS} $REF $TMPFQ | bmftools mark -l 0 | \
     samtools sort -l 0 -O bam -T tmprsqsort -O bam -@ $SORT_THREADS2 -m $SORTMEM - | \
     samtools merge -fh $TMPBAM $FINALBAM $TMPBAM -
-getsums.py $TMPBAM >>$LOG
 echo Post-rescue, merged-reads count -cF2816: $(samtools view -cF2816 $FINALBAM) >> $LOG &
 echo Post-rescue, merged-reads count -cF2304: $(samtools view -cF2304 $FINALBAM) >> $LOG &
 
-getsums.py $FINALBAM >>$LOG
-
 # QC
 
 samtools index $FINALBAM

diff --git a/src/bmf_depth.cpp b/src/bmf_depth.cpp
@@ -342,7 +342,7 @@ namespace BMF {
                 }
             }
             // Only print the first 3 columns plus the name column.
-            for(p = str.s, i = 0; i < 2;*p++ == '\t' ? ++i: 0);
+            for(p = str.s, i = 0; i < 3;*p++ == '\t' ? ++i: 0);
             str.l = p - str.s;
             for(i = 0; i < n; ++i) {
                 kputc('\t', &str);

diff --git a/src/bmf_dmp.cpp b/src/bmf_dmp.cpp
@@ -2,6 +2,7 @@
 
 #include <getopt.h>
 #include <omp.h>
+#include <zlib.h>
 #include "dlib/nix_util.h"
 #include "lib/binner.h"
 #include "lib/mseq.h"
@@ -432,7 +433,13 @@ namespace BMF {
         settings.max_blen = -1;
         settings.gzip_compression = 1;
         settings.cleanup = 1;
+#if ZLIB_VER_MAJOR <= 1 && ZLIB_VER_MINOR <= 2 && ZLIB_VER_REVISION < 5
+#pragma message("Note: zlib version < 1.2.5 doesn't support transparent file writing. Writing uncompressed temporary gzip files by default.")
+        // If not set, zlib compresses all our files enormously.
+        sprintf(settings.mode, "wb0");
+#else
         sprintf(settings.mode, "wT");
+#endif
 
         //omp_set_dynamic(0); // Tell omp that I want to set my number of threads 4realz
         int c;

diff --git a/src/bmf_sdmp.cpp b/src/bmf_sdmp.cpp
@@ -169,7 +169,13 @@ namespace BMF {
         settings.threads = DEFAULT_N_THREADS;
         settings.gzip_compression = 1;
         settings.cleanup = 1;
+#if ZLIB_VER_MAJOR <= 1 && ZLIB_VER_MINOR <= 2 && ZLIB_VER_REVISION < 5
+#pragma message("Note: zlib version < 1.2.5 doesn't support transparent file writing. Writing uncompressed temporary gzip files by default.")
+        // If not set, zlib compresses all our files enormously.
+        sprintf(settings.mode, "wb0");
+#else
         sprintf(settings.mode, "wT");
+#endif
 
         int c;
         while ((c = getopt(argc, argv, "t:o:i:n:m:s:f:u:p:g:v:r:T:hdczw?S&")) > -1) {

diff --git a/test/dmp/hashdmp_test.py b/test/dmp/hashdmp_test.py
@@ -27,7 +27,7 @@ def main():
     for ex in ["bmftools_db", "bmftools", "bmftools_p"]:
         cstr = "../../%s hashdmp -o hashdmp_test.out hashdmp_test.fq" % ex
         subprocess.check_call(shlex.split(cstr))
-        fqh  = pysam.FastqFile("hashdmp_test.out")
+        fqh = pysam.FastqFile("hashdmp_test.out")
         r1 = fqh.next()
         tags = get_tags(r1)
         assert tags["FM"] == 7