Merge pull request #54 from smithlabcode/fixing_test_output_comparisons

Fixing output validation in test suit

Makefile.am

@@ -72,6 +72,10 @@ EXTRA_DIST = \
              test_scripts/test_hmr.test \
 	     test_scripts/test_roi.test
 
+### ADS: the file md5sum.txt can be regenerated by running the tests
+### with "make check" and then doing `md5sum tests/* > md5sum.txt`
+### from the build directory.
+
 ACLOCAL_AMFLAGS = -I m4
 
 SUBDIRS := src/smithlab_cpp src/abismal
@@ -184,18 +188,18 @@ dnmtools_SOURCES += src/mlml/mlml.cpp
 ## grep "^outfile" test_scripts/*.test
 ##
 CLEANFILES = \
-	tests/tRex1.idx \
-	tests/reads.sam \
-	tests/reads.mstats \
 	tests/reads.bsrate \
-	tests/reads.counts.gz \
-	tests/reads.fmt.bam \
-	tests/reads.fmt.srt.bam \
+	tests/reads.counts \
+	tests/reads.counts.sym \
+	tests/reads.fmt.sam \
+	tests/reads.fmt.srt.sam \
+	tests/reads.fmt.srt.uniq.sam \
 	tests/reads.hmr \
 	tests/reads.levels \
-	tests/tRex1_promoters.roi.bed \
+	tests/reads.mstats \
+	tests/reads.sam \
+	tests/reads.ustats \
 	tests/simreads_1.fq \
 	tests/simreads_2.fq \
-	tests/reads.counts.sym \
-	tests/reads.fmt.srt.uniq.bam \
-	tests/reads.ustats
+	tests/tRex1.idx \
+	tests/tRex1_promoters.roi.bed

data/md5sum.txt

@@ -1,11 +1,11 @@
 cd862d6d57ea148375ddec9935574dfb  tests/reads_1.fq.gz
 43947b43b6888a17d9e8036692ce54c5  tests/reads_2.fq.gz
 96bfaea99753a9cf6193c5b7ced6d732  tests/reads.bsrate
-0b617e1d8dfbf4df1846ca502ebba2f9  tests/reads.counts.gz
+8e289a0d800c1165ca213812cbc75bdb  tests/reads.counts
 9af847370bbb0e78fdc0cd435a9f724f  tests/reads.counts.sym
-9d2826b98ae3642dbae75eef6f041cce  tests/reads.fmt.bam
-892cb87c55714055ff9dbd33132b0307  tests/reads.fmt.srt.bam
-7dcb23a0ce5478f9cc16d1f8992bd5a4  tests/reads.fmt.srt.uniq.bam
+4a27ffa9f858d8c32352d30c9623a338  tests/reads.fmt.sam
+e832cd0a2b0e1cfa43364610663590f3  tests/reads.fmt.srt.sam
+230fd48c3b9959a34078671770842fe6  tests/reads.fmt.srt.uniq.sam
 6dabc63bcab01f8c2b03708dad6bc0b7  tests/reads.hmr
 2610d1e63613e4e092f15aa132de3a20  tests/reads.levels
 8ee3daa36fad0e196d6d86696abed3dc  tests/reads.mstats

test_scripts/test_abismal.test

@@ -8,8 +8,8 @@ outfile2=tests/reads.mstats
 if [[ -e "${infile1}" && -e "${infile2}" && -e "${infile3}" ]]; then
     ./dnmtools abismal -s ${outfile2} -o ${outfile1} \
                -i ${infile3} ${infile1} ${infile2};
-    x1=$(md5sum -c tests/md5sum.txt | grep ${outfile1} | cut -d ' ' -f 2)
-    x2=$(md5sum -c tests/md5sum.txt | grep ${outfile2} | cut -d ' ' -f 2)
+    x1=$(md5sum -c tests/md5sum.txt | grep "${outfile1}:" | cut -d ' ' -f 2)
+    x2=$(md5sum -c tests/md5sum.txt | grep "${outfile2}:" | cut -d ' ' -f 2)
     if [[ "${x1}" != "OK" || "${x2}" != "OK" ]]; then
         exit 1;
     fi

test_scripts/test_abismalidx.test

@@ -4,7 +4,7 @@ infile=tests/tRex1.fa
 outfile=tests/tRex1.idx
 if [[ -e "${infile}" ]]; then
     ./dnmtools abismalidx ${infile} ${outfile}
-    x=$(md5sum -c tests/md5sum.txt | grep ${outfile} | cut -d ' ' -f 2)
+    x=$(md5sum -c tests/md5sum.txt | grep "${outfile}:" | cut -d ' ' -f 2)
     if [[ "${x}" != "OK" ]]; then
         exit 1;
     fi

test_scripts/test_bsrate.test

@@ -1,11 +1,11 @@
 #!/usr/bin/env bash
 
-infile1=tests/reads.fmt.srt.uniq.bam
+infile1=tests/reads.fmt.srt.uniq.sam
 infile2=tests/tRex1.fa
 outfile=tests/reads.bsrate
 if [[ -e "${infile1}" && -e "${infile2}" ]]; then
     ./dnmtools bsrate -c ${infile2} -o ${outfile} ${infile1}
-    x=$(md5sum -c tests/md5sum.txt | grep ${outfile} | cut -d ' ' -f 2)
+    x=$(md5sum -c tests/md5sum.txt | grep "${outfile}:" | cut -d ' ' -f 2)
     if [[ "${x}" != "OK" ]]; then
         exit 1;
     fi

test_scripts/test_counts.test

@@ -1,11 +1,11 @@
 #!/usr/bin/env bash
 
-infile1=tests/reads.fmt.srt.uniq.bam
+infile1=tests/reads.fmt.srt.uniq.sam
 infile2=tests/tRex1.fa
-outfile=tests/reads.counts.gz
+outfile=tests/reads.counts
 if [[ -e "${infile1}" && -e "${infile2}" ]]; then
-    ./dnmtools counts -v -z -o ${outfile} -c ${infile2} ${infile1}
-    x=$(md5sum -c tests/md5sum.txt | grep ${outfile} | cut -d ' ' -f 2)
+    ./dnmtools counts -v -o ${outfile} -c ${infile2} ${infile1}
+    x=$(md5sum -c tests/md5sum.txt | grep "${outfile}:" | cut -d ' ' -f 2)
     if [[ "${x}" != "OK" ]]; then
         exit 1;
     fi

test_scripts/test_format.test

@@ -1,12 +1,12 @@
 #!/usr/bin/env bash
 
 infile=tests/reads.sam
-outfile1=tests/reads.fmt.bam
-outfile2=tests/reads.fmt.srt.bam
+outfile1=tests/reads.fmt.sam
+outfile2=tests/reads.fmt.srt.sam
 cmd=samtools
 if [[ -e "${infile}" ]]; then
     ./dnmtools format -f abismal ${infile} ${outfile1}
-    x=$(md5sum -c tests/md5sum.txt | grep ${outfile1} | cut -d ' ' -f 2)
+    x=$(md5sum -c tests/md5sum.txt | grep "${outfile1}:" | cut -d ' ' -f 2)
     ### ADS: only want to check the first output here; any failure
     ### later will result in a skip for subsequent tests.
     if [[ "${x}" != "OK" ]]; then
@@ -18,7 +18,7 @@ else
 fi
 
 if [[ -e $(type -P "${cmd}") ]]; then
-    samtools sort -o ${outfile2} ${outfile1};
+    samtools sort --no-PG -O SAM -o ${outfile2} ${outfile1};
 else
     echo "${cmd} not found"
 fi

test_scripts/test_hmr.test

@@ -4,7 +4,7 @@ infile=tests/reads.counts.sym
 outfile=tests/reads.hmr
 if [[ -e "${infile}" ]]; then
     ./dnmtools hmr -v -o ${outfile} ${infile}
-    x=$(md5sum -c tests/md5sum.txt | grep ${outfile} | cut -d ' ' -f 2)
+    x=$(md5sum -c tests/md5sum.txt | grep "${outfile}:" | cut -d ' ' -f 2)
     if [[ "${x}" != "OK" ]]; then
         exit 1;
     fi

test_scripts/test_levels.test

@@ -1,10 +1,10 @@
 #!/usr/bin/env bash
 
-infile=tests/reads.counts.gz
+infile=tests/reads.counts
 outfile=tests/reads.levels
 if [[ -e "${infile}" ]]; then
     ./dnmtools levels -v -o ${outfile} ${infile}
-    x=$(md5sum -c tests/md5sum.txt | grep ${outfile} | cut -d ' ' -f 2)
+    x=$(md5sum -c tests/md5sum.txt | grep "${outfile}:" | cut -d ' ' -f 2)
     if [[ "${x}" != "OK" ]]; then
         exit 1;
     fi

test_scripts/test_roi.test

@@ -5,7 +5,7 @@ infile2=tests/tRex1_promoters.bed
 outfile=tests/tRex1_promoters.roi.bed
 if [[ -e "${infile1}" && -e "${infile2}" ]]; then
     ./dnmtools roi -v -M -o ${outfile} ${infile2} ${infile1}
-    x=$(md5sum -c tests/md5sum.txt | grep ${outfile} | cut -d ' ' -f 2)
+    x=$(md5sum -c tests/md5sum.txt | grep "${outfile}:" | cut -d ' ' -f 2)
     if [[ "${x}" != "OK" ]]; then
         exit 1;
     fi

test_scripts/test_simreads.test

@@ -7,8 +7,8 @@ outprefix=tests/simreads
 if [[ -e "${infile}" ]]; then
     ./dnmtools simreads -seed 1 \
                -o ${outprefix} -n 10000 -m 0.01 -b 0.98 ${infile};
-    x1=$(md5sum -c tests/md5sum.txt | grep ${outfile1} | cut -d ' ' -f 2)
-    x2=$(md5sum -c tests/md5sum.txt | grep ${outfile2} | cut -d ' ' -f 2)
+    x1=$(md5sum -c tests/md5sum.txt | grep "${outfile1}:" | cut -d ' ' -f 2)
+    x2=$(md5sum -c tests/md5sum.txt | grep "${outfile2}:" | cut -d ' ' -f 2)
     if [[ "${x1}" != "OK" || "${x2}" != "OK" ]]; then
         exit 1;
     fi

test_scripts/test_sym.test

@@ -1,10 +1,10 @@
 #!/usr/bin/env bash
 
-infile=tests/reads.counts.gz
+infile=tests/reads.counts
 outfile=tests/reads.counts.sym
 if [[ -e "${infile}" ]]; then
     ./dnmtools sym -o ${outfile} ${infile}
-    x=$(md5sum -c tests/md5sum.txt | grep ${outfile} | cut -d ' ' -f 2)
+    x=$(md5sum -c tests/md5sum.txt | grep "${outfile}:" | cut -d ' ' -f 2)
     if [[ "${x}" != "OK" ]]; then
         exit 1;
     fi

test_scripts/test_uniq.test

@@ -1,12 +1,12 @@
 #!/usr/bin/env bash
 
-infile=tests/reads.fmt.srt.bam
-outfile1=tests/reads.fmt.srt.uniq.bam
+infile=tests/reads.fmt.srt.sam
+outfile1=tests/reads.fmt.srt.uniq.sam
 outfile2=tests/reads.ustats
 if [[ -e "${infile}" ]]; then
-    ./dnmtools uniq -v -B -S ${outfile2} ${infile} ${outfile1}
-    x1=$(md5sum -c tests/md5sum.txt | grep ${outfile1} | cut -d ' ' -f 2)
-    x2=$(md5sum -c tests/md5sum.txt | grep ${outfile2} | cut -d ' ' -f 2)
+    ./dnmtools uniq -v -S ${outfile2} ${infile} ${outfile1}
+    x1=$(md5sum -c tests/md5sum.txt | grep "${outfile1}:" | cut -d ' ' -f 2)
+    x2=$(md5sum -c tests/md5sum.txt | grep "${outfile2}:" | cut -d ' ' -f 2)
     if [[ "${x1}" != "OK" || "${x2}" != "OK" ]]; then
         exit 1;
     fi