biology/canu: import canu-1.8

Canu is a fork of the Celera Assembler, designed for high-noise single-molecule
sequencing (such as the PacBio RS II/Sequel or Oxford Nanopore MinION).

Canu is a hierarchical assembly pipeline which runs in four steps:

    Detect overlaps in high-noise sequences using MHAP

    Generate corrected sequence consensus

    Trim corrected sequences

    Assemble trimmed corrected sequences

8 files changed, 356 insertions, 0 deletions

diff --git a/biology/canu/DESCR b/biology/canu/DESCR
new file mode 100644
index 00000000000..719f42cf95c
--- /dev/null
+++ b/biology/canu/DESCR
@@ -0,0 +1,12 @@
+Canu is a fork of the Celera Assembler, designed for high-noise single-molecule
+sequencing (such as the PacBio RS II/Sequel or Oxford Nanopore MinION).
+
+Canu is a hierarchical assembly pipeline which runs in four steps:
+
+    Detect overlaps in high-noise sequences using MHAP
+
+    Generate corrected sequence consensus
+
+    Trim corrected sequences
+
+    Assemble trimmed corrected sequences
diff --git a/biology/canu/Makefile b/biology/canu/Makefile
new file mode 100644
index 00000000000..db605ef37a5
--- /dev/null
+++ b/biology/canu/Makefile
@@ -0,0 +1,80 @@
+# $NetBSD: Makefile,v 1.1 2019/01/07 02:33:17 bacon Exp $
+
+DISTNAME=	canu-1.8
+CATEGORIES=	biology java
+MASTER_SITES=	${MASTER_SITE_GITHUB:=marbl/}
+GITHUB_TAG=	v${PKGVERSION_NOREV}
+
+MAINTAINER=	bacon@NetBSD.org
+HOMEPAGE=	http://canu.readthedocs.io/
+COMMENT=	Single molecule sequence assembler for genomes large and small
+LICENSE=	gnu-gpl-v2
+
+DEPENDS+=	gnuplot>0:../../graphics/gnuplot
+
+# Makefile compiles directly into ${DESTDIR}${PREFIX}
+SUBST_CLASSES+=		optimize
+SUBST_STAGE.optimize=	pre-configure
+SUBST_SED.optimize+=	-e 's|-g3||g'
+SUBST_SED.optimize+=	-e 's|-O3||g'
+SUBST_SED.optimize+=	-e 's|-O4||g'
+SUBST_SED.optimize+=	-e 's|-funroll-loops||g'
+SUBST_SED.optimize+=	-e 's|-fexpensive-optimizations||g'
+SUBST_SED.optimize+=	-e 's|-finline-functions||g'
+SUBST_FILES.optimize+=	${WRKSRC}/Makefile
+
+SUBST_CLASSES+=		jar
+SUBST_STAGE.jar=	pre-configure
+SUBST_SED.jar+=		-e 's|\\$$bin/mhap-|${PREFIX}/${JAVAJARDIR}/mhap-|g'
+SUBST_FILES.jar+=	${WRKSRC}/pipelines/canu/OverlapMhap.pm
+
+SUBST_CLASSES+=		perl
+SUBST_STAGE.perl=	pre-configure
+SUBST_SED.perl+=	-e 's|$$FindBin::RealBin/../lib/site_perl|${PREFIX}/${SITE_PERL_REL}|g'
+SUBST_FILES.perl+=	${WRKSRC}/pipelines/canu.pl
+
+REPLACE_PERL+=	bogart/*.pl
+REPLACE_PERL+=	pipelines/*.pl
+REPLACE_PERL+=	bogus/*.pl
+REPLACE_PERL+=	bogart-analysis/*.pl
+REPLACE_PERL+=	overlapBasedTrimming/*.pl
+REPLACE_PERL+=	overlapInCore-analysis/*.pl
+REPLACE_PERL+=	merTrim/*.pl
+REPLACE_PERL+=	erateEstimate/*.pl
+REPLACE_PERL+=	meryl/.pl
+REPLACE_PERL+=	fastq-utilities/*.pl
+REPLACE_PERL+=	*.pl
+
+# May work on other 64-bit processors, but untested
+ONLY_FOR_PLATFORM=	*-*-x86_64
+
+USE_LANGUAGES=	c c++
+USE_JAVA=	run
+USE_JAVA2=	8
+USE_TOOLS+=	gmake pax perl
+GCC_REQD+=	4.8
+
+WRKSRC=		${WRKDIR}/canu-${PKGVERSION_NOREV}/src
+MAKE_ENV+=	DESTDIR=${WRKSRC}
+
+TMP_INST=		${WRKSRC}${PREFIX}/${OPSYS}-${MACHINE_ARCH:S/x86_64/amd64/}
+SITE_PERL_REL=		lib/perl5/site_perl
+JAVAJARDIR=		share/java/classes
+INSTALLATION_DIRS=	bin ${JAVAJARDIR}
+
+post-extract:
+	${CHMOD} -R g-w ${WRKDIR}
+
+post-build:
+	${MKDIR} ${TMP_INST}/lib/perl5
+	${MV} ${TMP_INST}/lib/site_perl ${TMP_INST}/lib/perl5
+	${RM} -f ${TMP_INST}/bin/canu.defaults
+
+do-install:
+	cd ${TMP_INST}/bin && ${PAX} -wr * ${DESTDIR}${PREFIX}/bin
+	cd ${TMP_INST}/lib/perl5 && ${PAX} -wr * ${DESTDIR}${PREFIX}/lib/perl5
+	cd ${TMP_INST}/share && ${PAX} -wr * ${DESTDIR}${PREFIX}/share
+
+.include "../../devel/boost-libs/buildlink3.mk"
+.include "../../lang/perl5/module.mk"
+.include "../../mk/bsd.pkg.mk"
diff --git a/biology/canu/PLIST b/biology/canu/PLIST
new file mode 100644
index 00000000000..1f88345abe0
--- /dev/null
+++ b/biology/canu/PLIST
@@ -0,0 +1,80 @@
+@comment $NetBSD: PLIST,v 1.1 2019/01/07 02:33:17 bacon Exp $
+bin/alignGFA
+bin/bogart
+bin/bogus
+bin/canu
+bin/correctOverlaps
+bin/dumpBlob
+bin/edalign
+bin/errorEstimate
+bin/falconsense
+bin/fastqAnalyze
+bin/fastqSample
+bin/fastqSimulate
+bin/fastqSimulate-sort
+bin/filterCorrectionLayouts
+bin/filterCorrectionOverlaps
+bin/findErrors
+bin/findErrors-Dump
+bin/generateCorrectionLayouts
+bin/loadCorrectedReads
+bin/loadErates
+bin/loadTrimmedReads
+bin/meryl
+bin/mhapConvert
+bin/mmapConvert
+bin/ovStoreBucketizer
+bin/ovStoreBuild
+bin/ovStoreConfig
+bin/ovStoreDump
+bin/ovStoreIndexer
+bin/ovStoreSorter
+bin/ovStoreStats
+bin/overlapConvert
+bin/overlapImport
+bin/overlapInCore
+bin/overlapInCorePartition
+bin/overlapPair
+bin/prefixEditDistance-matchLimitGenerate
+bin/readConsensus
+bin/sequence
+bin/splitHaplotype
+bin/splitReads
+bin/sqStoreCreate
+bin/sqStoreCreatePartition
+bin/sqStoreDumpFASTQ
+bin/sqStoreDumpMetaData
+bin/tgStoreCompress
+bin/tgStoreCoverageStat
+bin/tgStoreDump
+bin/tgStoreFilter
+bin/tgStoreLoad
+bin/tgTigDisplay
+bin/trimReads
+bin/utgcns
+bin/wtdbgConvert
+lib/perl5/site_perl/canu/Configure.pm
+lib/perl5/site_perl/canu/Consensus.pm
+lib/perl5/site_perl/canu/CorrectReads.pm
+lib/perl5/site_perl/canu/Defaults.pm
+lib/perl5/site_perl/canu/Execution.pm
+lib/perl5/site_perl/canu/Grid.pm
+lib/perl5/site_perl/canu/Grid_Cloud.pm
+lib/perl5/site_perl/canu/Grid_DNANexus.pm
+lib/perl5/site_perl/canu/Grid_LSF.pm
+lib/perl5/site_perl/canu/Grid_PBSTorque.pm
+lib/perl5/site_perl/canu/Grid_SGE.pm
+lib/perl5/site_perl/canu/Grid_Slurm.pm
+lib/perl5/site_perl/canu/HaplotypeReads.pm
+lib/perl5/site_perl/canu/Meryl.pm
+lib/perl5/site_perl/canu/Output.pm
+lib/perl5/site_perl/canu/OverlapBasedTrimming.pm
+lib/perl5/site_perl/canu/OverlapErrorAdjustment.pm
+lib/perl5/site_perl/canu/OverlapInCore.pm
+lib/perl5/site_perl/canu/OverlapMMap.pm
+lib/perl5/site_perl/canu/OverlapMhap.pm
+lib/perl5/site_perl/canu/OverlapStore.pm
+lib/perl5/site_perl/canu/Report.pm
+lib/perl5/site_perl/canu/SequenceStore.pm
+lib/perl5/site_perl/canu/Unitig.pm
+share/java/classes/mhap-2.1.3.jar
diff --git a/biology/canu/distinfo b/biology/canu/distinfo
new file mode 100644
index 00000000000..aa0c188c636
--- /dev/null
+++ b/biology/canu/distinfo
@@ -0,0 +1,10 @@
+$NetBSD: distinfo,v 1.1 2019/01/07 02:33:17 bacon Exp $
+
+SHA1 (canu-1.8.tar.gz) = 7dd79415aa5ecb95f05109f0d8d58f7cbfc336e9
+RMD160 (canu-1.8.tar.gz) = 78d4872b4034f526037ce225c699debd910bd586
+SHA512 (canu-1.8.tar.gz) = 650bc96675f371596f8e7748d4ab2d229f0262bf84cee8fed59af43d534d76095a72e4ba0b4a5ce9f561992268c317964cda2f6c89ee514f4920e2ba47fbc86c
+Size (canu-1.8.tar.gz) = 2465314 bytes
+SHA1 (patch-Makefile) = aa83003677cbb12558e438c776402ec48df0598d
+SHA1 (patch-pipelines_canu_Defaults.pm) = 55a4631d86abb1881b0cc997514d44c536209ea6
+SHA1 (patch-pipelines_canu_Execution.pm) = fbb080c06ea5d2393d1835a61771715d2aef0274
+SHA1 (patch-utility_objectStore.C) = c8407de79abbaf296f027f704080cc7e878b85f4
diff --git a/biology/canu/patches/patch-Makefile b/biology/canu/patches/patch-Makefile
new file mode 100644
index 00000000000..d3aa3323d49
--- /dev/null
+++ b/biology/canu/patches/patch-Makefile
@@ -0,0 +1,48 @@
+$NetBSD: patch-Makefile,v 1.1 2019/01/07 02:33:17 bacon Exp $
+
+# Template Makefile requires platform-specific defaults for each OS
+# To be sent upstream following commit
+
+--- Makefile.orig	2018-10-22 16:47:31.000000000 +0000
++++ Makefile
+@@ -545,6 +545,40 @@ endif
+ endif
+ 
+ 
++ifeq (${OSTYPE}, NetBSD)
++  CC       ?= gcc
++  CXX      ?= g++
++
++  #  GCC
++  CXXFLAGS  += -pthread -fopenmp -fPIC
++  LDFLAGS   += -pthread -fopenmp -lm -lexecinfo
++
++  #CXXFLAGS  += -Wall -Wextra -Wformat  -Wno-unused   -Wno-parentheses
++  CXXFLAGS  += -Wall -Wextra -Wformat -Wno-unused-function -Wno-unused-parameter -Wno-unused-variable -Wno-char-subscripts -Wno-write-strings -Wno-sign-compare -Wno-format-truncation
++
++  #  Google Performance Tools malloc and heapchecker (HEAPCHECK=normal)
++  #CXXFLAGS  +=
++  #LDFLAGS   += -ltcmalloc
++
++  #  Google Performance Tools cpu profiler (CPUPROFILE=/path)
++  #CXXFLAGS  +=
++  #LDFLAGS   += -lprofiler
++
++  #  callgrind
++  #CXXFLAGS  += -g3 -Wa,--gstabs -save-temps
++
++  ifeq ($(BUILDOPTIMIZED), 1)
++  else
++    CXXFLAGS += -g3
++  endif
++
++  ifeq ($(BUILDDEBUG), 1)
++  else
++    CXXFLAGS += -O3   -finline-functions -fomit-frame-pointer
++  endif
++endif
++
++
+ ifneq (,$(findstring CYGWIN, ${OSTYPE}))
+   CC        ?= gcc
+   CXX       ?= g++
diff --git a/biology/canu/patches/patch-pipelines_canu_Defaults.pm b/biology/canu/patches/patch-pipelines_canu_Defaults.pm
new file mode 100644
index 00000000000..aabd51f3ca0
--- /dev/null
+++ b/biology/canu/patches/patch-pipelines_canu_Defaults.pm
@@ -0,0 +1,24 @@
+$NetBSD: patch-pipelines_canu_Defaults.pm,v 1.1 2019/01/07 02:33:17 bacon Exp $
+
+# Add resource limits for SLURM
+# Upstream is considering a scheduler-independent approach to this feature
+
+--- pipelines/canu/Defaults.pm.orig	2018-06-22 08:20:52.000000000 +0000
++++ pipelines/canu/Defaults.pm
+@@ -812,6 +812,16 @@ sub setDefaults () {
+     setDefault("gridEngineArraySubmitID",             undef, "Grid engine configuration, not documented");
+     setDefault("gridEngineJobID",                     undef, "Grid engine configuration, not documented");
+ 
++    #####  Slurm-specific parameters for controlling the number of
++    #####  cores / tasks dispatched per step or globally (WIP)
++
++    setDefault( 'slurmCormhapCoreLimit', undef, 'Maximum number of cores allocated for MHAP pre-computing and alignment within the correction phase' );
++    setDefault( 'slurmOvbCoreLimit', undef, 'Maximum number of single-core tasks dispatched for the ovlStore bucketizing step within the trimming phase' );
++    setDefault( 'slurmOvsCoreLimit', undef, 'Maximum number of single-core tasks dispatched for the ovlStore sorting step within the trimming phase' );
++    setDefault( 'slurmRedCoreLimit', undef, 'Maximum number of cores allocated for read error detection within the unitigging phase' );
++    setDefault( 'slurmArrayTaskLimit', undef, 'Maximum number of tasks permitted for each step throughout assembly' );
++    setDefault( 'slurmArrayCoreLimit', undef, 'Maximum number of cores allocated for each step throughout assembly' );
++
+     #####  Grid Engine Pipeline
+ 
+     setDefault("useGrid", 1, "If 'true', enable grid-based execution; if 'false', run all jobs on the local machine; if 'remote', create jobs for grid execution but do not submit; default 'true'");
diff --git a/biology/canu/patches/patch-pipelines_canu_Execution.pm b/biology/canu/patches/patch-pipelines_canu_Execution.pm
new file mode 100644
index 00000000000..30f37359990
--- /dev/null
+++ b/biology/canu/patches/patch-pipelines_canu_Execution.pm
@@ -0,0 +1,83 @@
+$NetBSD: patch-pipelines_canu_Execution.pm,v 1.1 2019/01/07 02:33:17 bacon Exp $
+
+# Add resource limits for SLURM
+# Upstream is considering a scheduler-independent approach to this feature
+
+--- pipelines/canu/Execution.pm.orig	2018-06-22 08:20:52.000000000 +0000
++++ pipelines/canu/Execution.pm
+@@ -303,10 +303,6 @@ sub skipStage ($$@) {
+ sub getInstallDirectory () {
+     my $installDir = $FindBin::RealBin;
+ 
+-    if ($installDir =~ m!^(.*)/\w+-\w+/bin$!) {
+-        $installDir = $1;
+-    }
+-
+     return($installDir);
+ }
+ 
+@@ -694,8 +690,8 @@ sub submitScript ($$) {
+ 
+ 
+ 
+-sub buildGridArray ($$$$) {
+-    my ($name, $bgn, $end, $opt) = @_;
++sub buildGridArray (@) {
++    my ( $name, $bgn, $end, $opt, $thr ) = @_;
+     my  $off = 0;
+ 
+     #  In some grids (SGE)   this is the maximum size of an array job.
+@@ -725,8 +721,42 @@ sub buildGridArray ($$$$) {
+         $off = "-F \"$off\"";
+     }
+ 
+-    $opt =~ s/ARRAY_NAME/$name/g;        #  Replace ARRAY_NAME with 'job name'
+-    $opt =~ s/ARRAY_JOBS/$bgn-$end/g;    #  Replace ARRAY_JOBS with 'bgn-end'
++    if( $opt =~ m/(ARRAY_NAME)/ )
++    {
++	$opt =~ s/$1/$name/; # Replace ARRAY_NAME with 'job name'
++    }
++    elsif( $opt =~ m/(ARRAY_JOBS)/ )
++    {
++	$opt =~ s/$1/$bgn-$end/; # Replace ARRAY_JOBS with 'bgn-end'
++
++	if( lc( getGlobal( 'gridEngine' ) ) eq 'slurm' && $end > 1 )
++	{
++	    if( $name =~ m/^cormhap_/i && defined getGlobal( 'slurmCormhapCoreLimit' ) )
++	    {
++		$opt .= '%' . int( getGlobal( 'slurmCormhapCoreLimit' ) / $thr );
++	    }
++	    elsif( $name =~ m/^ovb_/i && defined getGlobal( 'slurmOvbCoreLimit' ) )
++	    {
++		$opt .= '%' . getGlobal( 'slurmOvbCoreLimit' );
++	    }
++	    elsif( $name =~ m/^ovs_/i && defined getGlobal( 'slurmOvsCoreLimit' ) )
++	    {
++		$opt .= '%' . getGlobal( 'slurmOvsCoreLimit' );
++	    }
++	    elsif( $name =~ m/^red_/i && defined getGlobal( 'slurmRedCoreLimit' ) )
++	    {
++		$opt .= '%' . int( getGlobal( 'slurmRedCoreLimit' ) / $thr );
++	    }
++	    elsif( defined getGlobal( 'slurmArrayTaskLimit' ) )
++	    {
++		$opt .= '%' . getGlobal( 'slurmArrayTaskLimit' );
++	    }
++	    elsif( defined getGlobal( 'slurmArrayCoreLimit' ) )
++	    {
++		$opt .= '%' . int( getGlobal( 'slurmArrayCoreLimit' ) / $thr );
++	    }
++	}
++    }
+ 
+     return($opt, $off);
+ }
+@@ -870,7 +900,7 @@ sub buildGridJob ($$$$$$$$$) {
+     my $jobNameT               = makeUniqueJobName($jobType, $asm);
+ 
+     my ($jobName,  $jobOff)    = buildGridArray($jobNameT, $bgnJob, $endJob, getGlobal("gridEngineArrayName"));
+-    my ($arrayOpt, $arrayOff)  = buildGridArray($jobNameT, $bgnJob, $endJob, getGlobal("gridEngineArrayOption"));
++    my ( $arrayOpt, $arrayOff ) = buildGridArray( $jobNameT, $bgnJob, $endJob, getGlobal( "gridEngineArrayOption" ), $thr );
+ 
+     my $outputOption           = buildOutputOption($path, $script);
+ 
diff --git a/biology/canu/patches/patch-utility_objectStore.C b/biology/canu/patches/patch-utility_objectStore.C
new file mode 100644
index 00000000000..8c4af7f9091
--- /dev/null
+++ b/biology/canu/patches/patch-utility_objectStore.C
@@ -0,0 +1,19 @@
+$NetBSD: patch-utility_objectStore.C,v 1.1 2019/01/07 02:33:17 bacon Exp $
+
+# NetBSD does not provide WEXITED
+# Empty bitmask should suffice in this case
+
+--- utility/objectStore.C.orig	2018-12-23 01:57:06.000000000 +0000
++++ utility/objectStore.C
+@@ -286,6 +286,11 @@ fetchFromObjectStore(char *requested) {
+   //  Otherwise, we're still the parent, so wait for the (-1 == any) child
+   //  process to terminate.
+ 
++// NetBSD does not provide WEXITED so send empty bitmask
++#ifdef __NetBSD__
++#define	WEXITED	0
++#endif
++
+   waitpid(-1, &err, WEXITED);
+ 
+   if ((WIFEXITED(err)) &&