biology/unanimity: Pacific Biosciences consensus library and applications

Reviewed by:    jrm
Differential Revision:  https://reviews.freebsd.org/D15172

11 files changed, 146 insertions, 0 deletions

diff --git a/biology/Makefile b/biology/Makefile
index 996d7f82218c..5cef59a1bcbe 100644
--- a/biology/Makefile
+++ b/biology/Makefile
@@ -128,6 +128,7 @@
     SUBDIR += trimadap
     SUBDIR += trimmomatic
     SUBDIR += ugene
+    SUBDIR += unanimity
     SUBDIR += vcftools
     SUBDIR += vcflib
     SUBDIR += velvet
diff --git a/biology/unanimity/Makefile b/biology/unanimity/Makefile
new file mode 100644
index 000000000000..e114b0e2495b
--- /dev/null
+++ b/biology/unanimity/Makefile
@@ -0,0 +1,40 @@
+# $FreeBSD$
+
+PORTNAME=	unanimity
+DISTVERSION=	g20180307
+CATEGORIES=	biology
+
+MAINTAINER=	jwb@FreeBSD.org
+COMMENT=	Pacific Biosciences consensus library and applications
+
+LICENSE=	BSD3CLAUSE
+LICENSE_FILE=	${WRKSRC}/LICENSE
+
+# Sensitive to dependency versions, use latest commit for everything
+BUILD_DEPENDS=	pbseqan==g20171002:biology/pbseqan \
+		pbbam==g20180416:biology/pbbam \
+		pbcopper==g20180413:biology/pbcopper
+LIB_DEPENDS=	libhts.so:biology/htslib \
+		libboost_regex.so:devel/boost-libs
+
+USES=		cmake pkgconfig python
+USE_GITHUB=	yes
+GH_ACCOUNT=	PacificBiosciences
+GH_TAGNAME=	6f11a13e1472b8c00337ba8c5e94bf83bdab31d6
+
+CMAKE_ARGS+=	-DHTSLIB_INCLUDE_DIRS:STRING=${LOCALBASE}/include/htslib \
+		-DHTSLIB_LIBRARIES:STRING="-L${LOCALBASE}/lib -lhts" \
+		-DPacBioBAM_INCLUDE_DIRS:STRING=${LOCALBASE}/include/pbbam \
+		-DPacBioBAM_LIBRARIES:STRING="-L${LOCALBASE}/lib -lpbbam" \
+		-DSEQAN_INCLUDE_DIRS:STRING=${LOCALBASE}/include/PacBio \
+		-Dpbcopper_INCLUDE_DIRS=${LOCALBASE}/include \
+		-Dpbcopper_LIBRARIES="-L${LOCALBASE}/lib -lpbcopper"
+CFLAGS_i386=	-msse2 -Wno-c++11-narrowing
+
+PLIST_FILES=	bin/arrow bin/ccs bin/gcpp lib/libcc2.a lib/libunanimity.a
+
+post-install:
+	${INSTALL_DATA} ${WRKSRC}/src/libcc2.a ${WRKSRC}/src/libunanimity.a \
+		${STAGEDIR}${PREFIX}/lib
+
+.include <bsd.port.mk>
diff --git a/biology/unanimity/distinfo b/biology/unanimity/distinfo
new file mode 100644
index 000000000000..b8204cecf16a
--- /dev/null
+++ b/biology/unanimity/distinfo
@@ -0,0 +1,3 @@
+TIMESTAMP = 1524324354
+SHA256 (PacificBiosciences-unanimity-g20180307-6f11a13e1472b8c00337ba8c5e94bf83bdab31d6_GH0.tar.gz) = 20ed61c46afcfe86403a564136b96f909149208c51fde16f608e924ad4c59e38
+SIZE (PacificBiosciences-unanimity-g20180307-6f11a13e1472b8c00337ba8c5e94bf83bdab31d6_GH0.tar.gz) = 6492231
diff --git a/biology/unanimity/files/patch-CMakeLists.txt b/biology/unanimity/files/patch-CMakeLists.txt
new file mode 100644
index 000000000000..77e1825a21f3
--- /dev/null
+++ b/biology/unanimity/files/patch-CMakeLists.txt
@@ -0,0 +1,11 @@
+--- CMakeLists.txt.orig	2018-03-07 23:10:34 UTC
++++ CMakeLists.txt
+@@ -37,7 +37,7 @@ include(uny-ccache)
+ include(uny-releasewithassert)
+ include(uny-dependencies)
+ include(uny-compilerflags)
+-include(uny-gitsha1)
++# include(uny-gitsha1)
+ include(uny-config)
+ 
+ # Build library
diff --git a/biology/unanimity/files/patch-cmake_uny-config.cmake b/biology/unanimity/files/patch-cmake_uny-config.cmake
new file mode 100644
index 000000000000..33ec308164a4
--- /dev/null
+++ b/biology/unanimity/files/patch-cmake_uny-config.cmake
@@ -0,0 +1,10 @@
+--- cmake/uny-config.cmake.orig	2016-09-12 19:55:43 UTC
++++ cmake/uny-config.cmake
+@@ -1,6 +1,6 @@
+ # Config generation
+ 
+-find_git_sha1(UNANIMITY_GIT_SHA1)
++# find_git_sha1(UNANIMITY_GIT_SHA1)
+ 
+ file (STRINGS "${UNY_RootDir}/CHANGELOG.md" UNANIMITY_CHANGELOG)
+ 
diff --git a/biology/unanimity/files/patch-cmake_uny-gitsha1.cmake b/biology/unanimity/files/patch-cmake_uny-gitsha1.cmake
new file mode 100644
index 000000000000..0ebc1a0afcf2
--- /dev/null
+++ b/biology/unanimity/files/patch-cmake_uny-gitsha1.cmake
@@ -0,0 +1,11 @@
+--- cmake/uny-gitsha1.cmake.orig	2018-03-07 23:10:34 UTC
++++ cmake/uny-gitsha1.cmake
+@@ -7,7 +7,7 @@ set(__find_git_sha1 YES)
+ function(find_git_sha1 _GIT_SHA1)
+     find_package(Git QUIET REQUIRED)
+     execute_process(COMMAND
+-        "${GIT_EXECUTABLE}" "describe" "--always" "--dirty=-dirty"
++        echo 6f11a13e1472b8c00337ba8c5e94bf83bdab31d6
+         WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}"
+         RESULT_VARIABLE res
+         OUTPUT_VARIABLE out
diff --git a/biology/unanimity/files/patch-include_pacbio_genomicconsensus_Input.h b/biology/unanimity/files/patch-include_pacbio_genomicconsensus_Input.h
new file mode 100644
index 000000000000..89e43e082aff
--- /dev/null
+++ b/biology/unanimity/files/patch-include_pacbio_genomicconsensus_Input.h
@@ -0,0 +1,16 @@
+--- include/pacbio/genomicconsensus/Input.h.orig	2018-03-07 23:10:34 UTC
++++ include/pacbio/genomicconsensus/Input.h
+@@ -25,7 +25,12 @@ namespace GenomicConsensus {
+ class Input
+ {
+ public:
+-    explicit Input(const Settings& settings) : settings_{settings} {}
++    // Original
++    //explicit Input(const Settings& settings) : settings_{settings} {}
++    // Probably OK
++    //explicit Input(const Settings& settings) { settings_ = settings; }
++    // https://forums.freebsd.org/threads/compile-error-with-clang-3-4.65595/#post-385293
++    explicit Input(const Settings& settings) : settings_(settings) {}
+ 
+     Input() = delete;
+     Input(const Input&) = delete;
diff --git a/biology/unanimity/files/patch-include_pacbio_genomicconsensus_Output.h b/biology/unanimity/files/patch-include_pacbio_genomicconsensus_Output.h
new file mode 100644
index 000000000000..b17a496397aa
--- /dev/null
+++ b/biology/unanimity/files/patch-include_pacbio_genomicconsensus_Output.h
@@ -0,0 +1,19 @@
+--- include/pacbio/genomicconsensus/Output.h.orig	2018-03-07 23:10:34 UTC
++++ include/pacbio/genomicconsensus/Output.h
+@@ -80,8 +80,15 @@ inline void PrintVcfLite(const Variant& 
+         << "\tPASS\n";
+ }
+ 
+-inline Output::Output(const Settings& settings) : settings_{settings}
++// Original
++//inline Output::Output(const Settings& settings) : settings_{settings}
++// Probably OK patch
++//inline Output::Output(const Settings& settings)
++// https://forums.freebsd.org/threads/compile-error-with-clang-3-4.65595/#post-385293
++inline Output::Output(const Settings& settings) : settings_(settings)
+ {
++    // Probably OK patch
++    //settings_ = settings;
+     // initialize maps
+     PacBio::BAM::FastaSequenceQuery fasta{settings_.referenceFilename};
+     for (const auto& seq : fasta) {
diff --git a/biology/unanimity/files/patch-src_CMakeLists.txt b/biology/unanimity/files/patch-src_CMakeLists.txt
new file mode 100644
index 000000000000..b15b0949b992
--- /dev/null
+++ b/biology/unanimity/files/patch-src_CMakeLists.txt
@@ -0,0 +1,15 @@
+--- src/CMakeLists.txt.orig	2018-03-07 23:10:34 UTC
++++ src/CMakeLists.txt
+@@ -53,11 +53,11 @@ configure_file(
+ 
+ # includes
+ set(UNY_INCLUDE_DIRS
++    ${SEQAN_INCLUDE_DIRS}
+     ${UNY_IncludeDir}
+     ${CMAKE_BINARY_DIR}/generated
+     ${Boost_INCLUDE_DIRS}
+     ${HTSLIB_INCLUDE_DIRS}
+-    ${SEQAN_INCLUDE_DIRS}
+     ${CPPOPTPARSE_IncludeDir}
+     ${PacBioBAM_INCLUDE_DIRS}
+     ${pbcopper_INCLUDE_DIRS}
diff --git a/biology/unanimity/files/patch-src_Recursor.h b/biology/unanimity/files/patch-src_Recursor.h
new file mode 100644
index 000000000000..4c880356c882
--- /dev/null
+++ b/biology/unanimity/files/patch-src_Recursor.h
@@ -0,0 +1,12 @@
+# Type mismatch on 11.0 i386
+--- src/Recursor.h.orig	2018-03-07 23:10:34 UTC
++++ src/Recursor.h
+@@ -708,7 +708,7 @@ size_t Recursor<Derived>::FillAlphaBeta(
+     size_t J = tpl.Length();
+     int flipflops = 0;
+     size_t maxSize =
+-        std::max(100ul, static_cast<size_t>(0.5 + REBANDING_THRESHOLD * (I + 1) * (J + 1)));
++        std::max(100ul, static_cast<unsigned long>(0.5 + REBANDING_THRESHOLD * (I + 1) * (J + 1)));
+ 
+     // if we use too much space, do at least one more round
+     // to take advantage of rebanding
diff --git a/biology/unanimity/pkg-descr b/biology/unanimity/pkg-descr
new file mode 100644
index 000000000000..fa2abdd88c6f
--- /dev/null
+++ b/biology/unanimity/pkg-descr
@@ -0,0 +1,8 @@
+Unanimity is a C++ library + applications to generate and process accurate
+consensus sequences from Pacific Biosciences sequence data.
+
+The core binary, CCS (circular consensus calling) takes multiple reads of the
+same SMRTbell sequence and combines them, employing a statistical model, to
+produce one high quality consensus sequence.
+
+WWW: https://github.com/PacificBiosciences/unanimity