Merge branch 'release/5.1.0'

cancerit · Jun 20, 2020 · a2bcb9d · a2bcb9d
2 parents d0bb91b + 4aaf081
commit a2bcb9d
Show file tree

Hide file tree

Showing 18 changed files with 327 additions and 677 deletions.
diff --git a/.dockerignore b/.dockerignore
@@ -9,3 +9,4 @@
 /docs.tar.gz
 /setup.sh
 /prerelease.sh
+/blib
diff --git a/.gitignore b/.gitignore
@@ -15,6 +15,8 @@ examples/cgp_gnos_pull.ini
 /c/c_tests/tests_log
 /bin/diff_bams
 /bin/reheadSQ
+/bin/mismatchQc
+/bin/mmFlagModifier
 /c/c_tests/01_bam_stats_output_tests
 /c/c_tests/02_bam_access_tests
 /c/c_tests/03_bam_stats_calcs_tests
diff --git a/CHANGES.md b/CHANGES.md
@@ -1,5 +1,14 @@
 # CHANGES
 
+## 5.1.0
+
+* Base image updated to Focal (Ubuntu 20.04).
+* Majority of biobambam2 replaced with samtools functions.
+* Reads undergo full collate when mapping from BAM/CRAM (bwa-mem2 prep).
+* Duplicate marking `samtools markdup --mode` options exposed to `bwa_mem.pl`.
+  * Lanes mapped with earlier versions of PCAP-core cannot be merged without reporocessing to add "mate score tag" via `samtools fixmate`.
+* Scramble option for `bwa_mem.pl` deprecated, relevant option for fast CRAM random access exposed.
+
 ## 5.0.5
 
 * Add `noindex` commandline flag to `merge_or_mark.pl` for bammerge calls. Only permitted alongisde `qnamesort`

diff --git a/Dockerfile b/Dockerfile
@@ -1,38 +1,42 @@
-FROM quay.io/wtsicgp/cgpbigwig:1.1.0 as builder
+FROM quay.io/wtsicgp/cgpbigwig:1.3.0 as builder
 
 USER  root
 
-ARG BBB2_URL="https://gitlab.com/german.tischler/biobambam2/uploads/178774a8ece96d2201fcd0b5249884c7/biobambam2-2.0.146-release-20191030105216-x86_64-linux-gnu.tar.xz"
+# ALL tool versions used by opt-build.sh
+# need to keep in sync with setup.sh
+
+# newer gitlab versions do not work
+ARG BBB2_URL="https://github.com/gt1/biobambam2/releases/download/2.0.87-release-20180301132713/biobambam2-2.0.87-release-20180301132713-x86_64-etch-linux-gnu.tar.gz"
 ARG BWAMEM2_URL="https://github.com/bwa-mem2/bwa-mem2/releases/download/v2.0pre2/bwa-mem2-2.0pre2_x64-linux.tar.bz2"
 ARG STADEN="https://iweb.dl.sourceforge.net/project/staden/staden/2.0.0b11/staden-2.0.0b11-2016-linux-x86_64.tar.gz"
 ARG VER_BIODBHTS="3.01"
 ARG VER_BWA="v0.7.17"
-ARG VER_HTSLIB="1.9"
-ARG VER_SAMTOOLS="1.9"
+ARG VER_HTSLIB="1.10.2"
+ARG VER_SAMTOOLS="1.10"
 
 RUN apt-get -yq update
-RUN apt-get install -yq --no-install-recommends\
-  apt-transport-https\
-  locales\
-  curl\
-  ca-certificates\
-  libperlio-gzip-perl\
-  make\
-  bzip2\
-  gcc\
-  psmisc\
-  time\
-  zlib1g-dev\
-  libbz2-dev\
-  liblzma-dev\
-  libcurl4-gnutls-dev\
-  libncurses5-dev\
-  nettle-dev\
-  libp11-kit-dev\
-  libtasn1-dev\
-  libgnutls-dev\
-  libgd-dev\
-  libdb-dev
+RUN apt-get install -yq --no-install-recommends apt-transport-https
+RUN apt-get install -yq --no-install-recommends locales
+RUN apt-get install -yq --no-install-recommends curl
+RUN apt-get install -yq --no-install-recommends ca-certificates
+RUN apt-get install -yq --no-install-recommends libperlio-gzip-perl
+RUN apt-get install -yq --no-install-recommends make
+RUN apt-get install -yq --no-install-recommends bzip2
+RUN apt-get install -yq --no-install-recommends gcc
+RUN apt-get install -yq --no-install-recommends psmisc
+RUN apt-get install -yq --no-install-recommends time
+RUN apt-get install -yq --no-install-recommends zlib1g-dev
+RUN apt-get install -yq --no-install-recommends libbz2-dev
+RUN apt-get install -yq --no-install-recommends liblzma-dev
+RUN apt-get install -yq --no-install-recommends libcurl4-gnutls-dev
+RUN apt-get install -yq --no-install-recommends libncurses5-dev
+RUN apt-get install -yq --no-install-recommends nettle-dev
+RUN apt-get install -yq --no-install-recommends libp11-kit-dev
+RUN apt-get install -yq --no-install-recommends libtasn1-dev
+RUN apt-get install -yq --no-install-recommends libdb-dev
+RUN apt-get install -yq --no-install-recommends libgnutls28-dev
+RUN apt-get install -yq --no-install-recommends xz-utils
+RUN apt-get install -yq --no-install-recommends libexpat1-dev
 
 RUN locale-gen en_US.UTF-8
 RUN update-locale LANG=en_US.UTF-8
@@ -54,11 +58,11 @@ RUN bash build/opt-build.sh $OPT
 COPY . .
 RUN bash build/opt-build-local.sh $OPT
 
-FROM  ubuntu:16.04
+FROM  ubuntu:20.04
 
 LABEL maintainer="cgphelp@sanger.ac.uk"\
       uk.ac.sanger.cgp="Cancer, Ageing and Somatic Mutation, Wellcome Sanger Institute" \
-      version="5.0.5" \
+      version="5.1.0" \
       description="pcap-core"
 
 ENV OPT /opt/wtsi-cgp
@@ -67,6 +71,7 @@ ENV PATH $OPT/bin:$PATH
 ENV PERL5LIB $OPT/lib/perl5
 ENV LD_LIBRARY_PATH $OPT/lib:$OPT/scramble/lib
 ENV LC_ALL C
+ENV GPERF_FOR_BWA /usr/lib/x86_64-linux-gnu/libtcmalloc_minimal.so.4
 
 RUN apt-get -yq update
 RUN apt-get install -yq --no-install-recommends \
@@ -82,8 +87,10 @@ zlib1g \
 liblzma5 \
 libncurses5 \
 p11-kit \
-libcurl3 \
+libcurl3-gnutls \
+libcurl4 \
 moreutils \
+google-perftools \
 unattended-upgrades && \
 unattended-upgrade -d -v && \
 apt-get remove -yq unattended-upgrades && \

diff --git a/README.md b/README.md
@@ -29,18 +29,12 @@ Available programs are described in the [wiki][wiki].
 
 ## Docker, Singularity and Dockstore
 
-There are docker and dockstore.org wrappers for this project at [dockstore-cgpmap][dockstore-cgpmap].
+There are dockstore.org CWL and wrappers for this project at [dockstore-cgpmap][dockstore-cgpmap].
 
-The docker image is held on [quay.io][quay-io-cgpmap].
-
-The CWL bindings of `dockstore-cgpmap` specifically target execution of the BWA mem mapping flow,
-however all tools are contained in the image and can be used if you construct the relevant docker
-commands.
+The docker image is held on [quay.io][quay-io-pcap-core].
 
 The docker image is know to work correctly after import into a singularity image.
 
-See the [dockstore-cgpmap][dockstore-cgpmap] documentation for more detail.
-
 ## Dependencies/Install
 
 Please be aware that this expects basic C compilation libraries and tools to be available, most are listed in `INSTALL`.
@@ -69,13 +63,10 @@ Please see the respective licence for each before use.
 ### Cutting the release
 
 1. Update `lib/PCAP.pm` to the correct version.
-2. Ensure upgrade path for new version number is added to `lib/PCAP.pm`.
+2. Update `Dockerfile` to the correct version.
 3. Update `CHANGES.md` to show major items.
-4. Run `./prerelease.sh`
-5. Check all tests and coverage reports are acceptable.
-6. Commit the updated docs tree and updated module/version.
-7. Push commits.
-8. Use the GitHub tools to draft a release.
+4. Push commits and verify with Sanger internal CI.
+5. Use the GitHub tools to draft a release.
 
 <!-- References -->
 
@@ -87,7 +78,7 @@ Please see the respective licence for each before use.
 [cancerit_github]: https://github.com/cancerit
 [old_repo]: https://github.com/ICGC-TCGA-PanCancer/PCAP-core
 [dockstore-cgpmap]: https://github.com/cancerit/dockstore-cgpmap
-[quay-io-cgpmap]: https://quay.io/repository/wtsicgp/dockstore-cgpmap
+[quay-io-pcap-core]: https://quay.io/repository/wtsicgp/pcap-core
 
 <!-- Travis -->
 [travis-base]: https://travis-ci.org/cancerit/PCAP-core

diff --git a/bin/bwa_mem.pl b/bin/bwa_mem.pl
@@ -52,15 +52,14 @@
   my $options = setup();
 
  	my $threads = PCAP::Threaded->new($options->{'threads'});
-  &PCAP::Threaded::disable_out_err if(exists $options->{'index'});
 
   # register processes
-	$threads->add_function('split', \&PCAP::Bwa::split_in);
-	$threads->add_function('bwamem', \&PCAP::Bwa::bwa_mem, exists $options->{'index'} ? 1 : $options->{'map_threads'});
+  $threads->add_function('split', \&PCAP::Bwa::split_in, split_threads($options));
+  $threads->add_function('bwamem', \&PCAP::Bwa::bwa_mem, exists $options->{'index'} ? 1 : $options->{'map_threads'});
 
   PCAP::Bwa::mem_setup($options) if(!exists $options->{'process'} || $options->{'process'} eq 'setup');
 
-	$threads->run($options->{'max_split'}, 'split', $options) if(!exists $options->{'process'} || $options->{'process'} eq 'split');
+  $threads->run($options->{'max_split'}, 'split', $options) if(!exists $options->{'process'} || $options->{'process'} eq 'split');
 
   if(!exists $options->{'process'} || $options->{'process'} eq 'bwamem') {
     $options->{'max_index'} = PCAP::Bwa::mem_mapmax($options);
@@ -78,6 +77,24 @@
   }
 }
 
+sub split_threads {
+  my $options = shift;
+  my $div = 1;
+  my $threads_per_split = 1;
+  if(exists $options->{index}) {
+    $div = 1;
+    $threads_per_split = $options->{threads};
+  }
+  elsif($options->{raw_files}->[0] =~ m/(bam|cram)$/) {
+    my $inputs = scalar @{$options->{raw_files}};
+    $threads_per_split = int ($options->{threads} / $inputs);
+    $threads_per_split = 1 if($threads_per_split < 1);
+    $div = $threads_per_split;
+  }
+  $options->{threads_per_split} = $threads_per_split; # so can be used later
+  return $div; # so can be used as return
+}
+
 sub cleanup {
   my $options = shift;
   my $tmpdir = $options->{'tmp'};
@@ -91,6 +108,8 @@ sub setup {
               'mmqcfrac' => 0.05,
               'threads' => 1,
               'fragment' => 10,
+              'dupmode' => 't',
+              'seqslice' => 10000,
               'csi' => undef,
              );
 
@@ -116,6 +135,8 @@ sub setup {
               'q|mmqc' => \$opts{'mmqc'},
               'qf|mmqcfrac:f' => \$opts{'mmqcfrac'},
               'bm2|bwamem2' => \$opts{'bwamem2'},
+              'd|dupmode:s' => \$opts{'dupmode'},
+              'ss|seqslice:i' => $opts{'seqslice'},
   ) or pod2usage(2);
 
   pod2usage(-verbose => 1, -exitval => 0) if(defined $opts{'h'});
@@ -145,10 +166,14 @@ sub setup {
     die "ERROR: Please generate $opts{dict}, e.g.\n\t\$ samtools dict -a \$ASSEMBLY -s \$SPECIES $opts{reference} > $opts{dict}\n";
   }
 
+  if(defined $opts{'scramble'}) {
+    die "ERROR: -scramble option is deprecated, please see -seqslice\n";
+  }
+
   delete $opts{'process'} unless(defined $opts{'process'});
   delete $opts{'index'} unless(defined $opts{'index'});
   delete $opts{'bwa'} unless(defined $opts{'bwa'});
-  delete $opts{'scramble'} unless(defined $opts{'scramble'});
+  delete $opts{'scramble'};
   delete $opts{'bwa_pl'} unless(defined $opts{'bwa_pl'});
   delete $opts{'mmqc'} unless(defined $opts{'mmqc'});
   delete $opts{'csi'} unless(defined $opts{'csi'});
@@ -220,11 +245,12 @@ =head1 SYNOPSIS
   Optional parameters:
     -bwamem2     -bm2  Use bwa-mem2 instead of bwa.
     -fragment    -f    Split input into fragments of X million repairs [10]
+                        - only applies to fastq[.gz] input
     -nomarkdup   -n    Don't mark duplicates [flag]
     -csi               Use CSI index instead of BAI for BAM files [flag].
     -cram        -c    Output cram, see '-sc' [flag]
-    -scramble    -sc   Single quoted string of parameters to pass to Scramble when '-c' used
-                       - '-I,-O' are used internally and should not be provided
+    -seqslice    -ss   seqs_per_slice for CRAM compression [samtools default: 10000]
+    -scramble    -sc   DEPRECATED
     -bwa         -b    Single quoted string of additional parameters to pass to BWA
                         - '-t,-p,-R' are used internally and should not be provided.
                         - '-v' is set to 1 unless '-bwa' is set.
@@ -234,12 +260,15 @@ =head1 SYNOPSIS
     -mmqc        -q    Mark reads as QCFAIL (0x200, 512) if mismatch rate exceeded [flag]
                         - Please see 'bwa_mem.pl -m'
     -mmqcfrac    -qf   Mismatch fraction for -mmqc [0.05]
+    -dupmode     -d    see "samtools markdup -m" [t]
 
   Targeted processing:
     -process     -p    Only process this step then exit, optionally set -index
+                         setup - checks and configure workspace (-index N/A)
+                         split - split data by readgroup and chunk size (if applicable)
                         bwamem - only applicable if input is bam
                           mark - Run duplicate marking (-index N/A)
-                         stats - Generates the *.bas file for the final BAM.
+                         stats - Generates the *.bas file for the final BAM  (-index N/A)
 
     -index       -i    Optionally restrict '-p' to single job
                         bwamem - 1..<lane_count>
@@ -249,6 +278,7 @@ =head1 SYNOPSIS
                        https://github.com/gperftools/ (assuming number of cores not exceeded)
                        If available specify the path to 'gperftools/lib/libtcmalloc_minimal.so'.
                        - NOT APPLIED TO bwa-mem2
+                       Falls back to environment variable GPERF_FOR_BWA when not set, or nothing.
 
   Other:
     -jobs        -j   For a parallel step report the number of jobs required

diff --git a/bin/merge_or_mark.pl b/bin/merge_or_mark.pl
@@ -39,7 +39,6 @@
 use PCAP::Bwa;
 use version;
 
-const my $COORD_SORT_ORDER => 'coordinate';
 const my $QUERYNAME_SORT_ORDER => 'queryname';
 const my @VALID_PROCESS => qw(setup mark stats);
 const my %INDEX_FACTOR => ( 'setup' => 1,
@@ -72,7 +71,8 @@ sub setup {
   my %opts = (
               'threads' => 1,
               'csi' => undef,
-              'sortorder' => $COORD_SORT_ORDER,
+              'dupmode' => 't',
+              'seqslice' => 10000,
              );
 
   GetOptions( 'h|help' => \$opts{'h'},
@@ -84,11 +84,13 @@ sub setup {
               's|sample=s' => \$opts{'sample'},
               'n|nomarkdup' => \$opts{'nomarkdup'},
               'p|process=s' => \$opts{'process'},
-              'q|querynamesort' => \$opts{'qnamesort'},
+              'q|qnamesort' => \$opts{'qnamesort'},
               'i|noindex' => \$opts{'noindex'},
               'csi' => \$opts{'csi'},
               'c|cram' => \$opts{'cram'},
               'sc|scramble=s' => \$opts{'scramble'},
+              'd|dupmode:s' => \$opts{'dupmode'},
+              'ss|seqslice:i' => $opts{'seqslice'},
   ) or pod2usage(2);
 
   pod2usage(-verbose => 1, -exitval => 0) if(defined $opts{'h'});
@@ -113,17 +115,20 @@ sub setup {
     die "ERROR: Please generate $opts{dict}, e.g.\n\t\$ samtools dict -a \$ASSEMBLY -s \$SPECIES $opts{reference} > $opts{dict}\n";
   }
 
+  if(defined $opts{'scramble'}) {
+    die "ERROR: -scramble option is deprecated, please see -seqslice\n";
+  }
+
   delete $opts{'process'} unless(defined $opts{'process'});
   delete $opts{'index'} unless(defined $opts{'index'});
-  delete $opts{'scramble'} unless(defined $opts{'scramble'});
+  delete $opts{'scramble'};
   delete $opts{'csi'} unless(defined $opts{'csi'});
   if($opts{'qnamesort'} && !$opts{'nomarkdup'}){
       die "ERROR: -qnamesort can only be used in conjunction with -nomarkdups\n";
   }
   if($opts{'noindex'} && !$opts{'qnamesort'}){
       die "ERROR: -noindex can only be used in conjunction with -qnamesort\n";
   }
-  $opts{'sortorder'} = $QUERYNAME_SORT_ORDER if($opts{'qnamesort'});
 
   if($opts{'threads'} > 4) {
     warn "Setting 'threads' to 4 as higher values are of limited value\n";
@@ -172,18 +177,19 @@ =head1 SYNOPSIS
     -nomarkdup   -n   Don't mark duplicates [flag]
     -qnamesort   -q   Use queryname sorting flag in bammerge rather than coordinate. [flag].
                       To be used in conjunction with -nomarkdup only
-    -noindex     -i   Don't attempt to index the merged file. Only available in conjunction with 
+    -noindex     -i   Don't attempt to index the merged file. Only available in conjunction with
                       -qnamesort.
     -csi              Use CSI index instead of BAI for BAM files [flag].
     -cram        -c   Output cram, see '-sc' [flag]
-    -scramble    -sc  Single quoted string of parameters to pass to Scramble when '-c' used
-                      - '-I,-O' are used internally and should not be provided
+    -seqslice    -ss   seqs_per_slice for CRAM compression [samtools default: 10000]
+    -scramble    -sc   DEPRECATED
+    -dupmode     -d    see "samtools markdup -m" [t]
 
   Targeted processing:
-    -process     -p   Only process this step then exit, optionally set -index
-                        bwamem - only applicable if input is bam
-                          mark - Run duplicate marking (-index N/A)
-                         stats - Generates the *.bas file for the final BAM.
+    -process     -p   Only process this step then exit
+                         setup - only applicable if input is bam
+                          mark - Run duplicate marking
+                         stats - Generates the *.bas file for the final BAM
 
   Other:
     -help        -h   Brief help message.
@@ -261,6 +267,8 @@ =head2 OPTIONAL parameters
 
 =item B<-scramble>
 
+DEPRECATED - see -seqslice
+
 Single quoted string of parameters to pass to Scramble when '-c' used.  Please see the Scramble
 documentation for details.