Merge branch 'hotfix/v3.2.2'

CHANGES.md

@@ -1,3 +1,11 @@
+### 3.2.2
+* Add bedtools2 to `setup.sh`
+* Added bedtools2 to `README.md`
+* Changes `Bio::DB::HTS`, `samtools` and `HTSlib` install methods.
+* Corrected condition indicating sort is required.
+* Fixes #23 Changed from vcf-sort to normal linux sort to ensure multiple indels with
+same start coord are sorted in a stable way.
+
 ### 3.2.0
 * Allows use of ensemblgenomes.org as a datasource
 * Handle genes without names, and give more useful error message

README.md

@@ -15,12 +15,14 @@ as well as Sequence Ontology terms to classify its consequences.
 
 ---
 
-###Dependencies/Install
+### Dependencies/Install
 Some of the code included in this package has dependencies on several packages:
 
  * [Samtools v1.3+](https://github.com/samtools/samtools)
  * [vcftools](https://vcftools.github.io/)
  * [Bio::DB::HTS](http://search.cpan.org/~rishidev/Bio-DB-HTS/)
+ * [bedtools2](http://bedtools.readthedocs.io/en/latest/index.html)
+   * Not >=2.24.0, no upgrades until [this ticket](https://github.com/arq5x/bedtools2/issues/319) is resolved (which may involve code changes)
 
 And various perl modules.
 
@@ -30,12 +32,12 @@ Please be aware that this expects basic C compilation libraries and tools to be
 
 ---
 
-##Creating a release
-####Preparation
+## Creating a release
+#### Preparation
 * Commit/push all relevant changes.
 * Pull a clean version of the repo and use this for the following steps.
 
-####Cutting the release
+#### Cutting the release
 1. Update `lib/Sanger/CGP/Vagrent.pm` to the correct version.
 2. Update `CHANGES.md` to show major items.
 3. Run `./prerelease.sh`
@@ -47,7 +49,7 @@ Please be aware that this expects basic C compilation libraries and tools to be
 LICENCE
 =======
 
-Copyright (c) 2014-2016 Genome Research Ltd.
+Copyright (c) 2014-2017 Genome Research Ltd.
 
 Author: Cancer Genome Project <[email protected]>
 

bin/AnnotateVcf.pl

@@ -1,7 +1,7 @@
 #!/usr/bin/perl
 
 ##########LICENCE##########
-# Copyright (c) 2014 Genome Research Ltd.
+# Copyright (c) 2014-2017 Genome Research Ltd.
 #
 # Author: Cancer Genome Project [email protected]
 #
@@ -34,6 +34,7 @@
 
 use List::Util qw(first);
 use File::Temp qw(tempfile);
+use File::Copy qw(copy);
 use Try::Tiny qw(try catch);
 
 use FindBin qw($Bin);
@@ -78,7 +79,7 @@
 const my $REPRE_BM => Sanger::CGP::Vagrent::Bookmarkers::RepresentativeTranscriptBookmarker->new();
 const my $WORST_BM => Sanger::CGP::Vagrent::Bookmarkers::MostDeleteriousBookmarker->new();
 
-const my $SORT_CMD => 'cat %s | vcf-sort > %s';
+const my $SORT_CMD => q{(grep -B 100000000 -m 1 '^#CHROM' %s ; grep -v '^#' %s | sort -k1,1 -k2,2n -k4,4 -k5,5) > %s};
 const my $BGZIP_CMD => 'bgzip %s';
 const my $TABIX_CMB => 'tabix -p vcf %s';
 
@@ -119,14 +120,17 @@
 sub compressAndIndex {
   my ($options, $tmpfile) = @_;
 
-  my $sort_cmd = sprintf $SORT_CMD, $tmpfile, $options->{'output'};
+  my $sort_cmd = sprintf $SORT_CMD, $tmpfile, $tmpfile, $options->{'output'};
   my $bgzip_cmd = sprintf $BGZIP_CMD, $options->{'output'};
   my $totabix = $options->{'output'} .'.gz';
   my $tabix_cmd = sprintf $TABIX_CMB, $totabix;
 
   try {
-    my $tabix_in = $options->{'input'}.'.tbx';
-    unless(-e $tabix_in){
+    my $tabix_in = $options->{'input'}.'.tbi';
+    if(-e $tabix_in) {
+      copy($tmpfile, $options->{'output'});
+    }
+    else {
       # If the input has a tabix index it must have already been sorted,
       # we haven't changed the order of the file so we can skip this sort
       system($sort_cmd);

lib/Sanger/CGP/Vagrent.pm

@@ -1,7 +1,7 @@
 package Sanger::CGP::Vagrent;
 
 ##########LICENCE##########
-# Copyright (c) 2014-2016 Genome Research Ltd.
+# Copyright (c) 2014-2017 Genome Research Ltd.
 #
 # Author: Cancer Genome Project [email protected]
 #
@@ -26,7 +26,7 @@ use strict;
 use Const::Fast qw(const);
 
 use base 'Exporter';
-our $VERSION = '3.2.1';
+our $VERSION = '3.2.2';
 our @EXPORT = qw($VERSION);
 
 1;

setup.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 
 ##########LICENCE##########
-# Copyright (c) 2014-2016 Genome Research Ltd.
+# Copyright (c) 2014-2017 Genome Research Ltd.
 #
 # Author: Cancer Genome Project [email protected]
 #
@@ -21,10 +21,13 @@
 # along with this program. If not, see <http://www.gnu.org/licenses/>.
 ##########LICENCE##########
 
-
 SOURCE_SAMTOOLS="https://github.com/samtools/samtools/releases/download/1.3.1/samtools-1.3.1.tar.bz2"
-BIODBHTS_INSTALL="https://raw.githubusercontent.com/Ensembl/Bio-HTS/master/INSTALL.pl"
+SOURCE_HTSLIB="https://github.com/samtools/htslib/releases/download/1.3.2/htslib-1.3.2.tar.bz2"
+SOURCE_BIOBDHTS="https://github.com/Ensembl/Bio-HTS/archive/2.3.tar.gz"
 SOURCE_VCFTOOLS="https://github.com/vcftools/vcftools/releases/download/v0.1.14/vcftools-0.1.14.tar.gz"
+# Warning bedtools 2.24.0 and 2.25.0 have a swapped usage in coverageBed
+# No upgrades until [this ticket](https://github.com/arq5x/bedtools2/issues/319) is resolved
+SOURCE_BEDTOOLS="https://github.com/arq5x/bedtools2/releases/download/v2.21.0/bedtools-2.21.0.tar.gz"
 
 get_distro () {
   EXT=""
@@ -62,7 +65,7 @@ if [[ ($# -ne 1 && $# -ne 2) ]] ; then
   echo "Please provide an installation path and optionally perl lib paths to allow, e.g."
   echo "  ./setup.sh /opt/myBundle"
   echo "OR all elements versioned:"
-  echo "  ./setup.sh /opt/cgpVcf-X.X.X /opt/PCAP-X.X.X/lib/perl"
+  echo "  ./setup.sh /opt/myBundle /opt/cgpVcf-X.X.X/lib/perl5:/opt/PCAP-X.X.X/lib/perl5"
   exit 0
 fi
 
@@ -120,18 +123,35 @@ perl $SETUP_DIR/cpanm -l $INST_PATH App::cpanminus
 CPANM=`which cpanm`
 echo $CPANM
 
-perlmods=( "File::ShareDir" "File::ShareDir::Install" "Bio::Root::[email protected]")
+perlmods=( "File::ShareDir" "File::ShareDir::Install" "Module::Build" "Bio::Root::[email protected]")
 
 for i in "${perlmods[@]}" ; do
   echo -n "Installing build prerequisite $i..."
   $CPANM --notest --mirror http://cpan.metacpan.org -l $INST_PATH $i
 done
 
+echo -n "Building bedtools2 ..."
+if [ -e $SETUP_DIR/bedtools.success ]; then
+  echo -n " previously installed (resumed)...";
+elif [ -e $INST_PATH/bin/bedtools ]; then
+  echo -n " previously installed ...";
+else
+  cd $SETUP_DIR
+  get_distro "bedtools2" $SOURCE_BEDTOOLS
+  mkdir -p bedtools2
+  tar --strip-components 1 -C bedtools2 -zxf bedtools2.tar.gz
+  make -C bedtools2 -j$CPU
+  cp bedtools2/bin/* $INST_PATH/bin/.
+  touch $SETUP_DIR/bedtools.success
+fi
+
 CURR_TOOL="vcftools"
 CURR_SOURCE=$SOURCE_VCFTOOLS
 echo -n "Building $CURR_TOOL ..."
 if [ -e $SETUP_DIR/$CURR_TOOL.success ]; then
-  echo -n " previously installed ..."
+  echo -n " previously installed (resumed) ..."
+elif [ -e $INST_PATH/bin/$CURR_TOOL ]; then
+  echo -n " previously installed ...";
 else
   get_distro $CURR_TOOL $CURR_SOURCE
   cd $SETUP_DIR/$CURR_TOOL
@@ -142,29 +162,71 @@ else
   touch $SETUP_DIR/$CURR_TOOL.success
 fi
 
-echo -n "Building samtools ..."
-if [ -e "$SETUP_DIR/samtools.success" ]; then
-  echo -n " previously installed ...";
+if [ -e $SETUP_DIR/htslibGet.success ]; then
+  echo " already staged ...";
 else
+  echo
   cd $SETUP_DIR
-  get_distro "samtools" $SOURCE_SAMTOOLS
-  cd samtools
+  get_distro "htslib" $SOURCE_HTSLIB
+  touch $SETUP_DIR/htslibGet.success
+fi
+
+echo -n "Building htslib ..."
+if [ -e $SETUP_DIR/htslib.success ]; then
+  echo " previously installed ...";
+else
+  echo
+  mkdir -p htslib
+  tar --strip-components 1 -C htslib -jxf htslib.tar.bz2
+  cd htslib
   ./configure --enable-plugins --enable-libcurl --prefix=$INST_PATH
-  make all all-htslib
-  make install install-htslib
-  touch $SETUP_DIR/samtools.success
+  make -j$CPU
+  make install
+  cd $SETUP_DIR
+  touch $SETUP_DIR/htslib.success
 fi
 
+export HTSLIB=$INST_PATH
+
 CHK=`perl -le 'eval "require $ARGV[0]" and print $ARGV[0]->VERSION' Bio::DB::HTS`
 if [[ "x$CHK" == "x" ]] ; then
   echo -n "Building Bio::DB::HTS ..."
-  cd $SETUP_DIR
-  # now Bio::DB::HTS
-  get_file "INSTALL.pl" $BIODBHTS_INSTALL
-  perl -I $PERL5LIB INSTALL.pl --prefix $INST_PATH --static
-  rm -f BioDbHTS_INSTALL.pl
+  if [ -e $SETUP_DIR/biohts.success ]; then
+    echo " previously installed ...";
+  else
+    echo
+    cd $SETUP_DIR
+    rm -rf bioDbHts
+    get_distro "bioDbHts" $SOURCE_BIOBDHTS
+    tar --strip-components 1 -C bioDbHts -zxf bioDbHts.tar.gz
+    cd bioDbHts
+    perl Build.PL --install_base=$INST_PATH --htslib=$INST_PATH
+    ./Build test
+    ./Build install
+    cd $SETUP_DIR
+    rm -f bioDbHts.tar.gz
+    touch $SETUP_DIR/biohts.success
+  fi
+else
+  echo "Bio::DB::HTS already installed ..."
+fi
+
+if [ -e $SETUP_DIR/samtools.success ]; then
+  echo " previously installed ...";
 else
-  echo "Bio::DB::HTS already installed"
+echo
+  cd $SETUP_DIR
+  rm -rf samtools
+  get_distro "samtools" $SOURCE_SAMTOOLS
+  mkdir -p samtools
+  tar --strip-components 1 -C samtools -xjf samtools.tar.bz2
+  cd samtools
+  ./configure --enable-plugins --enable-libcurl --with-htslib=$HTSLIB --prefix=$INST_PATH
+  make -j$CPU all
+  make install
+  cd $SETUP_DIR
+  rm -f samtools.tar.bz2
+  touch $SETUP_DIR/samtools.success
 fi
 
 cd $INIT_DIR