Merge pull request #75900 from jbedo/tebreak

Tebreak, minia, last, bx-python, python-lzo

pkgs/applications/science/biology/last/default.nix

@@ -0,0 +1,30 @@
+{ stdenv, fetchurl, unzip, zlib, python3, parallel }:
+
+stdenv.mkDerivation rec {
+  pname = "last";
+  version = "1042";
+
+  src = fetchurl {
+    url = "http://last.cbrc.jp/last-${version}.zip";
+    sha256 = "0mgbhd01m9riqza2gx56qk6x5682kg71zflhisr6d3q05wrv103f";
+  };
+
+  nativeBuildInputs = [ unzip ];
+  buildInputs = [ zlib python3 ];
+
+  makeFlags = [ "prefix=${placeholder "out"}" ];
+
+  postFixup = ''
+    for f in $out/bin/parallel-* ; do
+      sed -i 's|parallel |${parallel}/bin/parallel |' $f
+    done
+  '';
+
+  meta = with stdenv.lib; {
+    description = "Genomic sequence aligner";
+    homepage = "http://last.cbrc.jp/";
+    license = licenses.gpl3;
+    maintainers = with maintainers; [ jbedo ];
+    platforms = platforms.x86_64;
+  };
+}

pkgs/applications/science/biology/minia/default.nix

@@ -0,0 +1,33 @@
+{ stdenv, fetchFromGitHub, cmake, hdf5, boost }:
+
+stdenv.mkDerivation rec {
+  pname = "minia";
+  version = "3.2.1";
+
+  src = fetchFromGitHub {
+    owner = "GATB";
+    repo = "minia";
+    rev = "v${version}";
+    sha256 = "0bmfrywixaaql898l0ixsfkhxjf2hb08ssnqzlzacfizxdp46siq";
+    fetchSubmodules = true;
+  };
+
+  patches = [ ./no-bundle.patch ];
+
+  NIX_CFLAGS_COMPILE = [ "-Wformat" ];
+
+  nativeBuildInputs = [ cmake ];
+  buildInputs = [ hdf5 boost ];
+
+  prePatch = ''
+    rm -rf thirdparty/gatb-core/gatb-core/thirdparty/{hdf5,boost}
+  '';
+
+  meta = with stdenv.lib; {
+    description = "Short read genome assembler";
+    homepage = "https://github.com/GATB/minia";
+    license = licenses.agpl3;
+    maintainers = with maintainers; [ jbedo ];
+    platforms = [ "x86_64-linux" ];
+  };
+}

pkgs/applications/science/biology/minia/no-bundle.patch

@@ -0,0 +1,222 @@
+diff --git a/thirdparty/gatb-core/gatb-core/CMakeLists.txt b/thirdparty/gatb-core/gatb-core/CMakeLists.txt
+index f48a70b..0e11ece 100644
+--- a/thirdparty/gatb-core/gatb-core/CMakeLists.txt
++++ b/thirdparty/gatb-core/gatb-core/CMakeLists.txt
+@@ -257,7 +257,6 @@ ADD_SUBDIRECTORY(thirdparty)
+ #  DEPENDENCIES 
+ ################################################################################
+ # we must be sure that hdf5 is built and installed before building gatb-core
+-ADD_DEPENDENCIES (gatbcore-static hdf5 hdf5_postbuild)
+ 
+ ################################################################################
+ #  DOCUMENTATION GENERATION 
+@@ -288,7 +287,6 @@ IF (NOT DEFINED GATB_CORE_INSTALL_EXCLUDE)
+     INSTALL (FILES ${PROJECT_SOURCE_DIR}/doc/misc/README.txt  DESTINATION . OPTIONAL)
+     INSTALL (FILES ${PROJECT_SOURCE_DIR}/LICENCE              DESTINATION . OPTIONAL)
+     INSTALL (FILES ${PROJECT_SOURCE_DIR}/THIRDPARTIES.md      DESTINATION . OPTIONAL)
+-    INSTALL (DIRECTORY ${PROJECT_SOURCE_DIR}/thirdparty/boost DESTINATION ./include)
+ ENDIF()
+ 
+ ################################################################################
+diff --git a/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/LargeInt.hpp b/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/LargeInt.hpp
+index dfeee1c..d5553a2 100644
+--- a/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/LargeInt.hpp
++++ b/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/LargeInt.hpp
+@@ -35,7 +35,7 @@
+ #include <stdint.h>
+ #include <algorithm>
+ #include <iostream>
+-#include <hdf5/hdf5.h>
++#include <hdf5.h>
+ 
+ #include <gatb/system/api/Exception.hpp>
+ #include <gatb/system/api/config.hpp>
+diff --git a/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt128.hpp b/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt128.hpp
+index 60be5d5..25ae75e 100644
+--- a/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt128.hpp
++++ b/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt128.hpp
+@@ -33,7 +33,7 @@
+ /********************************************************************************/
+ 
+ #include <iostream>
+-#include <hdf5/hdf5.h>
++#include <hdf5.h>
+ 
+ #include <gatb/system/api/types.hpp>
+ #include <gatb/tools/misc/api/Abundance.hpp>
+diff --git a/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt16.hpp b/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt16.hpp
+index 6a71bb0..b9205df 100644
+--- a/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt16.hpp
++++ b/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt16.hpp
+@@ -31,7 +31,7 @@
+ #include <iostream>
+ #include <gatb/system/api/types.hpp>
+ #include <gatb/tools/misc/api/Abundance.hpp>
+-#include <hdf5/hdf5.h>
++#include <hdf5.h>
+ 
+ /********************************************************************************/
+ namespace gatb  {
+diff --git a/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt32.hpp b/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt32.hpp
+index c22b892..62e6586 100644
+--- a/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt32.hpp
++++ b/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt32.hpp
+@@ -31,7 +31,7 @@
+ #include <iostream>
+ #include <gatb/system/api/types.hpp>
+ #include <gatb/tools/misc/api/Abundance.hpp>
+-#include <hdf5/hdf5.h>
++#include <hdf5.h>
+ 
+ /********************************************************************************/
+ namespace gatb  {
+diff --git a/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt64.hpp b/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt64.hpp
+index c06aaab..e0befba 100644
+--- a/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt64.hpp
++++ b/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt64.hpp
+@@ -31,7 +31,7 @@
+ #include <iostream>
+ #include <gatb/system/api/types.hpp>
+ #include <gatb/tools/misc/api/Abundance.hpp>
+-#include <hdf5/hdf5.h>
++#include <hdf5.h>
+ 
+ extern const unsigned char revcomp_4NT[];
+ extern const unsigned char comp_NT    [];
+diff --git a/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt8.hpp b/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt8.hpp
+index 9659874..0c79ff6 100644
+--- a/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt8.hpp
++++ b/thirdparty/gatb-core/gatb-core/src/gatb/tools/math/NativeInt8.hpp
+@@ -31,7 +31,7 @@
+ #include <iostream>
+ #include <gatb/system/api/types.hpp>
+ #include <gatb/tools/misc/api/Abundance.hpp>
+-#include <hdf5/hdf5.h>
++#include <hdf5.h>
+ 
+ /********************************************************************************/
+ namespace gatb  {
+diff --git a/thirdparty/gatb-core/gatb-core/src/gatb/tools/misc/api/Abundance.hpp b/thirdparty/gatb-core/gatb-core/src/gatb/tools/misc/api/Abundance.hpp
+index 3cb84f8..cd5d382 100644
+--- a/thirdparty/gatb-core/gatb-core/src/gatb/tools/misc/api/Abundance.hpp
++++ b/thirdparty/gatb-core/gatb-core/src/gatb/tools/misc/api/Abundance.hpp
+@@ -31,7 +31,7 @@
+ /********************************************************************************/
+ 
+ #include <sys/types.h>
+-#include <hdf5/hdf5.h>
++#include <hdf5.h>
+ 
+ /********************************************************************************/
+ namespace gatb      {
+diff --git a/thirdparty/gatb-core/gatb-core/src/gatb/tools/misc/api/IHistogram.hpp b/thirdparty/gatb-core/gatb-core/src/gatb/tools/misc/api/IHistogram.hpp
+index b8f6c79..a040832 100644
+--- a/thirdparty/gatb-core/gatb-core/src/gatb/tools/misc/api/IHistogram.hpp
++++ b/thirdparty/gatb-core/gatb-core/src/gatb/tools/misc/api/IHistogram.hpp
+@@ -28,7 +28,7 @@
+ 
+ #include <gatb/system/api/ISmartPointer.hpp>
+ #include <gatb/tools/storage/impl/Storage.hpp>
+-#include <hdf5/hdf5.h>
++#include <hdf5.h>
+ 
+ /********************************************************************************/
+ namespace gatb      {
+diff --git a/thirdparty/gatb-core/gatb-core/src/gatb/tools/storage/impl/CollectionHDF5.hpp b/thirdparty/gatb-core/gatb-core/src/gatb/tools/storage/impl/CollectionHDF5.hpp
+index 2645abd..fad48c0 100644
+--- a/thirdparty/gatb-core/gatb-core/src/gatb/tools/storage/impl/CollectionHDF5.hpp
++++ b/thirdparty/gatb-core/gatb-core/src/gatb/tools/storage/impl/CollectionHDF5.hpp
+@@ -40,7 +40,7 @@
+ #include <string>
+ #include <vector>
+ #include <stdarg.h>
+-#include <hdf5/hdf5.h>
++#include <hdf5.h>
+ 
+ /********************************************************************************/
+ namespace gatb      {
+diff --git a/thirdparty/gatb-core/gatb-core/src/gatb/tools/storage/impl/CollectionHDF5Patch.hpp b/thirdparty/gatb-core/gatb-core/src/gatb/tools/storage/impl/CollectionHDF5Patch.hpp
+index a92b729..66d552f 100644
+--- a/thirdparty/gatb-core/gatb-core/src/gatb/tools/storage/impl/CollectionHDF5Patch.hpp
++++ b/thirdparty/gatb-core/gatb-core/src/gatb/tools/storage/impl/CollectionHDF5Patch.hpp
+@@ -40,7 +40,7 @@
+ #include <string>
+ #include <vector>
+ #include <stdarg.h>
+-#include <hdf5/hdf5.h>
++#include <hdf5.h>
+ 
+ /********************************************************************************/
+ namespace gatb      {
+diff --git a/thirdparty/gatb-core/gatb-core/src/gatb/tools/storage/impl/StorageHDF5.hpp b/thirdparty/gatb-core/gatb-core/src/gatb/tools/storage/impl/StorageHDF5.hpp
+index 29e0949..0565cc4 100644
+--- a/thirdparty/gatb-core/gatb-core/src/gatb/tools/storage/impl/StorageHDF5.hpp
++++ b/thirdparty/gatb-core/gatb-core/src/gatb/tools/storage/impl/StorageHDF5.hpp
+@@ -33,7 +33,7 @@
+ #include <gatb/tools/storage/impl/CollectionHDF5.hpp>
+ #include <gatb/tools/storage/impl/CollectionHDF5Patch.hpp>
+ #include <gatb/system/impl/System.hpp>
+-#include <hdf5/hdf5.h>
++#include <hdf5.h>
+ #include <sstream>
+ 
+ /********************************************************************************/
+diff --git a/thirdparty/gatb-core/gatb-core/thirdparty/CMakeLists.txt b/thirdparty/gatb-core/gatb-core/thirdparty/CMakeLists.txt
+index 6e0b5c4..34aef28 100644
+--- a/thirdparty/gatb-core/gatb-core/thirdparty/CMakeLists.txt
++++ b/thirdparty/gatb-core/gatb-core/thirdparty/CMakeLists.txt
+@@ -1,54 +1,3 @@
+-################################################################################
+-#  HDF5 GENERATION 
+-################################################################################
+-
+-#SET (HDF5_ENABLE_THREADSAFE       ON)
+-#SET (H5_HAVE_THREADSAFE 1)
+-
+-##########  MOMENTARY DEACTIVATED => CRASH ON MACOS TO BE INVESTIGATED  ##########
+-SET (HDF5_BUILD_TOOLS             ON CACHE BOOL "Build HDF5 Tools")
+-#SET (CMAKE_EXE_LINKER_FLAGS       "-lpthread -lz")
+-
+-SET (HDF5_EXTERNALLY_CONFIGURED     ON)
+-
+-#SET (HDF5_INSTALL_BIN_DIR           ${PROJECT_BINARY_DIR}/bin/${CMAKE_BUILD_TYPE})
+-#SET (HDF5_INSTALL_LIB_DIR           ${PROJECT_BINARY_DIR}/lib/${CMAKE_BUILD_TYPE})
+-SET (HDF5_INSTALL_BIN_DIR           bin)
+-SET (HDF5_INSTALL_LIB_DIR           lib)
+-
+-SET (HDF5_INSTALL_INCLUDE_DIR       ${PROJECT_BINARY_DIR}/include/${CMAKE_BUILD_TYPE}/hdf5)
+-SET (HDF5_INSTALL_DATA_DIR          ${PROJECT_BINARY_DIR}/share/${CMAKE_BUILD_TYPE})
+-SET (HDF5_INSTALL_CMAKE_DIR         ${PROJECT_BINARY_DIR}/share/${CMAKE_BUILD_TYPE})
+-
+-IF (NOT DEFINED GATB_CORE_INSTALL_EXCLUDE)
+-    SET (HDF5_EXPORTED_TARGETS          "gatb-hdf5")
+-ENDIF()
+-
+-IF (NOT DEFINED GATB_CORE_EXCLUDE_HDF5_ZLIB)
+-	OPTION (HDF5_ENABLE_Z_LIB_SUPPORT "Enable Zlib Filters" ON)
+-ENDIF()
+-
+-# We don't want warnings from HDF5 compilation
+-set (COMPILE_DEFINITIONS  "${COMPILE_DEFINITIONS}  -w")
+-add_definitions (${COMPILE_DEFINITIONS})
+-
+-# add HDF5 generation
+-ADD_SUBDIRECTORY (hdf5)
+-
+-# We add a custom target for copying header files.
+-add_custom_target (hdf5_postbuild ALL)
+-
+-# We build the output directory
+-add_custom_command (TARGET hdf5_postbuild POST_BUILD  COMMAND ${CMAKE_COMMAND} -E make_directory ${HDF5_INSTALL_INCLUDE_DIR})
+-
+-# We define all the header files to be copied
+-file (GLOB headerfiles ${PROJECT_SOURCE_DIR}/thirdparty/hdf5/src/*.h ${PROJECT_BINARY_DIR}/thirdparty/hdf5/H5pubconf.h)
+-
+-# We copy each header file
+-foreach (header ${headerfiles})
+-  add_custom_command (TARGET hdf5_postbuild POST_BUILD  COMMAND ${CMAKE_COMMAND} -E copy_if_different ${header} ${HDF5_INSTALL_INCLUDE_DIR}  )
+-endforeach()
+-
+ # include other smaller libraries (json, Boophf)
+ 
+ add_custom_target (thirdparty_copy ALL)

pkgs/applications/science/biology/tebreak/default.nix

@@ -0,0 +1,42 @@
+{ stdenv, fetchFromGitHub, last, exonerate, minia, python3Packages, bwa
+, samtools, findutils }:
+
+python3Packages.buildPythonApplication rec {
+  pname = "tebreak";
+  version = "1.0";
+
+  src = fetchFromGitHub {
+    owner = "adamewing";
+    repo = "tebreak";
+    rev = version;
+    sha256 = "194av17wz66n4zxyi56mbkik31j2wmkly5i9qmxgaxymhavzi3kq";
+  };
+
+  nativeBuildInputs = [ findutils python3Packages.cython ];
+  propagatedBuildInputs = with python3Packages; [
+    pysam
+    scipy
+    bx-python
+    scikit-bio
+  ];
+
+  preConfigure = ''
+    # patch the paths to all required software
+    for f in $(find . -type f) ; do
+      sed -i "s|'bwa'|'${bwa}/bin/bwa'|" $f
+      sed -i "s|'minia'|'${minia}/bin/minia'|" $f
+      sed -i "s|'exonerate'|'${exonerate}/bin/exonerate'|" $f
+      sed -i "s|'samtools'|'${samtools}/bin/samtools'|" $f
+      sed -i "s|'lastal'|'${last}/bin/lastal'|" $f
+      sed -i "s|'lastdb'|'${last}/bin/lastdb'|" $f
+    done
+  '';
+
+  meta = with stdenv.lib; {
+    description = "Find and characterise transposable element insertions";
+    homepage = "https://github.com/adamewing/tebreak";
+    license = licenses.mit;
+    maintainers = with maintainers; [ jbedo ];
+    platforms = platforms.x86_64;
+  };
+}

pkgs/development/python-modules/bx-python/default.nix

@@ -0,0 +1,39 @@
+{ lib, fetchFromGitHub, buildPythonPackage, isPy27, numpy, cython, zlib, six
+, python-lzo, nose }:
+
+buildPythonPackage rec {
+  pname = "bx-python";
+  version = "0.8.6";
+  disabled = isPy27;
+
+  src = fetchFromGitHub {
+    owner = "bxlab";
+    repo = "bx-python";
+    rev = "v${version}";
+    sha256 = "1i40vmn8n83vqcpqj843riv9vp16s753jc4wc90p0cmrnhmzcv13";
+  };
+
+  nativeBuildInputs = [ cython ];
+  buildInputs = [ zlib ];
+  propagatedBuildInputs = [ numpy six python-lzo ];
+  checkInputs = [ nose ];
+
+  postInstall = ''
+    cp -r scripts/* $out/bin
+
+    # This is a small hack; the test suit uses the scripts which need to
+    # be patched. Linking the patched scripts in $out back to the
+    # working directory allows the tests to run
+    rm -rf scripts
+    ln -s $out/bin scripts
+  '';
+
+  meta = with lib; {
+    homepage = "https://github.com/bxlab/bx-python";
+    description =
+      "Tools for manipulating biological data, particularly multiple sequence alignments";
+    license = licenses.mit;
+    maintainers = [ maintainers.jbedo ];
+    platforms = [ "x86_64-linux" ];
+  };
+}

pkgs/development/python-modules/python-lzo/default.nix

@@ -0,0 +1,22 @@
+{ lib, fetchPypi, buildPythonPackage, lzo, nose }:
+
+buildPythonPackage rec {
+  pname = "python-lzo";
+  version = "1.12";
+
+  src = fetchPypi {
+    inherit pname version;
+    sha256 = "0iakqgd51n1cd7r3lpdylm2rgbmd16y74cra9kcapwg84mlf9a4p";
+  };
+
+  buildInputs = [ lzo ];
+  propagatedBuildInputs = [ ];
+  checkInputs = [ nose ];
+
+  meta = with lib; {
+    homepage = "https://github.com/jd-boyd/python-lzo";
+    description = "Python bindings for the LZO data compression library";
+    license = licenses.gpl2;
+    maintainers = [ maintainers.jbedo ];
+  };
+}

pkgs/top-level/all-packages.nix

@@ -23672,10 +23672,16 @@ in
 
   kssd = callPackage ../applications/science/biology/kssd { };
 
+  last = callPackage ../applications/science/biology/last { };
+
   macse = callPackage ../applications/science/biology/macse { };
 
   migrate = callPackage ../applications/science/biology/migrate { };
 
+  minia = callPackage ../applications/science/biology/minia {
+    boost = boost159;
+  };
+
   mirtk = callPackage ../development/libraries/science/biology/mirtk { };
 
   muscle = callPackage ../applications/science/biology/muscle { };
@@ -23762,6 +23768,8 @@ in
 
   svaba = callPackage ../applications/science/biology/svaba { };
 
+  tebreak = callPackage ../applications/science/biology/tebreak { };
+
   trimal = callPackage ../applications/science/biology/trimal { };
 
   truvari = callPackage ../applications/science/biology/truvari { };

pkgs/top-level/python-packages.nix

@@ -1702,6 +1702,10 @@ in {
 
   bumps = callPackage ../development/python-modules/bumps {};
 
+  bx-python = callPackage ../development/python-modules/bx-python {
+    inherit (pkgs) zlib;
+  };
+
   cached-property = callPackage ../development/python-modules/cached-property { };
 
   caffe = toPythonModule (pkgs.caffe.override {
@@ -2940,6 +2944,10 @@ in {
 
   python-ctags3 = callPackage ../development/python-modules/python-ctags3 { };
 
+  python-lzo = callPackage ../development/python-modules/python-lzo {
+    inherit (pkgs) lzo;
+  };
+
   junos-eznc = callPackage ../development/python-modules/junos-eznc {};
 
   raven = callPackage ../development/python-modules/raven { };