c_cpp/lib/htslib/test/cross_validate.sh

   1 #!/bin/sh
   2
   3 #
   4 # -----------------------------------------------------------------------------
   5 # Author: James Bonfield.
   6 #
   7 # This cross validation script is designed to run the htslib test_view
   8 # and cramtools.jar CRAM implementations to test compatibility between
   9 # implementations.
  10 #
  11 # The test set may contain many dubious and ambiguous SAM cases, such as
  12 # single base reads (is that quality "*" really meaning Q9 or no quality?).
  13 # Some of these may fail one or the other implementations and be acceptable
  14 # in the short-term, so to spot more important regressions we can tag
  15 # specific cases as being known-pass or known-fail.
  16 # -----------------------------------------------------------------------------
  17 #
  18
  19 vers=3.0
  20
  21 cramtools_jar=$HOME/work/cram/cramtools/cramtools-$vers.jar
  22
  23 test_view="./test_view -o VERSION=$vers"
  24
  25 cramtools="/software/bin/java -Xmx4000m -jar $cramtools_jar"
  26 cramtools="/software/bin/java -Xmx4000m -jar $cramtools_jar"
  27
  28 run_out() {
  29     out=$1; shift
  30     echo "$@ > $out"
  31     $@ > $out
  32 }
  33
  34 run() {
  35     echo "$@"
  36     $@
  37 }
  38
  39
  40 sam_to_Ccram() {
  41     run_out _tmp.cram $test_view -C -t $1 $2
  42     #run_out _tmp.cram $HOME/io_lib/trunk/build.seq3/progs/scramble -r $1 -O CRAM $2
  43     if [ $? != 0 ]
  44     then
  45         crash=`expr $crash + 1`
  46         false
  47     fi
  48 }
  49
  50 Ccram_to_sam() {
  51     run_out _tmp.sam $test_view -i REFERENCE=$1 _tmp.cram
  52     #run_out _tmp.sam $HOME/io_lib/trunk/build.seq3/progs/scramble -r $1 _tmp.cram
  53
  54     if [ $? != 0 ]
  55     then
  56         crash=`expr $crash + 1`
  57         false
  58     fi
  59 }
  60
  61 sam_to_Jcram() {
  62     run $cramtools cram -R $1 -I $2 -O _tmp.cram -n -Q --capture-all-tags
  63     if [ $? != 0 ]
  64     then
  65         crash=`expr $crash + 1`
  66         false
  67     fi
  68 }
  69
  70 Jcram_to_sam() {
  71     run $cramtools bam -R $1 -I _tmp.cram -O _tmp.sam
  72
  73     if [ $? != 0 ]
  74     then
  75         crash=`expr $crash + 1`
  76         false
  77     fi
  78 }
  79
  80 compare_sam() {
  81     #run ./compare_sam.pl $i _tmp.sam -nomd -notemplate -unknownrg -Baux
  82     run ./compare_sam.pl $i _tmp.sam -nomd -Baux
  83     if [ $? != 0 ]
  84     then
  85         fails=`expr $fails + 1`
  86         false
  87     fi
  88 }
  89
  90 trials=0
  91 fails=0
  92 crash=0
  93
  94 files=`ls -1 *#*.sam`
  95
  96 # Restrict to known workers from SAM->CRAM->CRAM in cramtools
  97 #files="auxf#values.sam c1#bounds.sam c1#noseq.sam c1#pad1.sam c1#pad2.sam c1#pad3.sam c1#unknown.sam ce#1.sam ce#2.sam ce#5b.sam ce#large_seq.sam ce#tag_depadded.sam ce#tag_padded.sam ce#unmap.sam ce#unmap1.sam ce#unmap2.sam xx#large_aux.sam xx#large_aux2.sam xx#pair.sam xx#rg.sam xx#unsorted.sam"
  98
  99 for i in $files
 100 do
 101     r=`echo $i | sed 's/#.*/.fa/'`
 102     echo "=== $i"
 103
 104     # C to C
 105     trials=`expr $trials + 1`
 106     sam_to_Ccram $r $i && Ccram_to_sam $r && compare_sam $i _tmp.sam
 107
 108     # Java to Java
 109     trials=`expr $trials + 1`
 110     sam_to_Jcram $r $i && Jcram_to_sam $r && compare_sam $i _tmp.sam
 111
 112     # C to Java
 113     trials=`expr $trials + 1`
 114     sam_to_Ccram $r $i && Jcram_to_sam $r && compare_sam $i _tmp.sam
 115
 116     # Java to C
 117     trials=`expr $trials + 1`
 118     sam_to_Jcram $r $i && Ccram_to_sam $r && compare_sam $i _tmp.sam
 119 done
 120
 121 # Overcounts failures as an early fail can lead to 1 or 2 more fails.
 122 echo ""
 123 echo ============
 124 echo No. tests: $trials
 125 echo No. diffs: $fails
 126 echo No. crash: $crash