Merge pull request #203 from voutcn/refactor

v1.2.1

.travis.yml

@@ -6,12 +6,19 @@ language: python
 python:
   - "2.7"
   - "3.4"
-script: git submodule update --init && mkdir build && cd build && cmake .. -DCMAKE_BUILD_TYPE=Debug -DCOVERAGE=ON && make -j2 simple_test
+script:
+  - mkdir build
+  - cd build && cmake .. -DCMAKE_BUILD_TYPE=Debug -DCOVERAGE=ON
+  - make -j2 simple_test
+  - sudo make install
+  - megahit --test
+  - megahit --test --kmin-1pass
+  - megahit --test --no-hw-accel
 after_success:
   # Create lcov report
   - lcov --capture --directory . --output-file coverage.info
   - lcov --remove coverage.info '/usr/*' --output-file coverage.info # filter system-files
-  - lcov --remove coverage.info '*xxhash/*' --output-file coverage.info # filter xxhash-files
+  - lcov --remove coverage.info '*xxHash/*' --output-file coverage.info # filter xxhash-files
   - lcov --remove coverage.info '*sparsepp/*' --output-file coverage.info # filter sparsepp-files
   - lcov --list coverage.info # debug info
   # Uploading report to CodeCov

CHANGELOG.md

@@ -1,3 +1,8 @@
+### 1.2.1-beta / 2019-03-30 PST
+-   Added `--no-hw-accel` option for users whose CPUs do not support BMI2/POPCNT
+-   Added `--test` option for testing
+-   Compilable with CMake 2.8 and g++4.8
+
 ### 1.2.0-beta / 2019-03-24 PST
 
 Heavily refactored the whole project:

CMakeLists.txt

@@ -1,11 +1,15 @@
 cmake_minimum_required(VERSION 2.8)
 project(megahit)
 set(CMAKE_VERBOSE_MAKEFILE ON)
-set(CMAKE_CXX_STANDARD 11)
 
-option(USE_POPCNT "Use popcnt hardware instructions" ON)
-option(USE_BMI2 "Use bmi2 hardware instructions" ON)
+if (CMAKE_VERSION VERSION_LESS "3.1")
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
+else ()
+    set(CMAKE_CXX_STANDARD 11)
+endif ()
+
 option(COVERAGE "Generate coverage report" OFF)
+option(STATIC_BUILD "Build static executation" OFF)
 
 include_directories(src)
 include_directories(src/sparsepp)
@@ -27,38 +31,47 @@ LIST(APPEND OTHER_SOURCE
         src/sequence_manager.cpp
         )
 
+if (STATIC_BUILD)
+    set(CMAKE_FIND_LIBRARY_SUFFIXES ".a")
+endif (STATIC_BUILD)
+
 find_package(ZLIB REQUIRED)
 find_package(OpenMP REQUIRED)
 
-#set(STATIC_LIBS -lz -ldl -lpthread -lgomp -static-libstdc++ -static-libgcc -static)
-
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${OpenMP_CXX_FLAGS}")
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -DXXH_INLINE_ALL -D__STDC_FORMAT_MACROS")
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wall -Wno-unused-function")
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fprefetch-loop-arrays -funroll-loops")
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -D__XFILE__='\"$(subst ${CMAKE_SOURCE_DIR}/,,$(abspath $<))\"'")
 
-
-if (USE_POPCNT)
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mpopcnt")
-endif (USE_POPCNT)
-
-if (USE_BMI2)
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mbmi2 -DUSE_BMI2")
-endif (USE_BMI2)
+set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} ${ZLIB_EXE_LINKER_FLAGS} ${OpenMP_EXE_LINKER_FLAGS}")
 
 if (COVERAGE)
     set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} --coverage")
-endif(COVERAGE)
+    set(COV_PY "coverage run")
+endif (COVERAGE)
 
+set(CMAKE_CXX_FLAGS_RELEASE "-O2 -DNDEBUG")
 set(CMAKE_CXX_FLAGS_DEBUG "-g -ggdb -O1")
 
 
 message(STATUS "Build type: ${CMAKE_BUILD_TYPE}: ${CMAKE_CXX_FLAGS}")
 
 add_executable(megahit_core ${OTHER_SOURCE} ${ASMBL_SOURCE} ${IDBA_SOURCE} ${SDBG_SOURCE} ${LCASM_SOURCE}
         ${CX1_SOURCE} ${TOOLKIT_SOURCE})
-target_link_libraries(megahit_core ${ZLIB_LIBRARIES} ${OpenMP_CXX_LIBRARIES})
+add_executable(megahit_core_no_hw_accel ${OTHER_SOURCE} ${ASMBL_SOURCE} ${IDBA_SOURCE} ${SDBG_SOURCE} ${LCASM_SOURCE}
+        ${CX1_SOURCE} ${TOOLKIT_SOURCE})
+set_target_properties(megahit_core PROPERTIES COMPILE_FLAGS "-mbmi2 -DUSE_BMI2 -mpopcnt")
+
+if (STATIC_BUILD)
+    # TODO too dirty
+    set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -Wl,--whole-archive -lpthread -Wl,--no-whole-archive -static")
+    set_target_properties(megahit_core megahit_core_no_hw_accel PROPERTIES LINK_SEARCH_START_STATIC ON)
+    set_target_properties(megahit_core megahit_core_no_hw_accel PROPERTIES LINK_SEARCH_END_STATIC ON)
+endif (STATIC_BUILD)
+
+target_link_libraries(megahit_core ${ZLIB_LIBRARIES})
+target_link_libraries(megahit_core_no_hw_accel ${ZLIB_LIBRARIES})
 
 add_custom_target(
         megahit
@@ -69,15 +82,14 @@ set(TEST_DATA ${CMAKE_SOURCE_DIR}/test_data)
 
 add_custom_target(
         simple_test
-        COMMAND rm -rf megahit_out*
-        COMMAND ./megahit --12 ${TEST_DATA}/r1.il.fa.gz,${TEST_DATA}/r2.il.fa.bz2 -1 ${TEST_DATA}/r3_1.fa -2 ${TEST_DATA}/r3_2.fa -r ${TEST_DATA}/r4.fa -o megahit_out1 -t 2 --keep-tmp-files
-        COMMAND rm -rf megahit_out*
-        COMMAND ./megahit --12 ${TEST_DATA}/r1.il.fa.gz,${TEST_DATA}/r2.il.fa.bz2 -1 ${TEST_DATA}/r3_1.fa -2 ${TEST_DATA}/r3_2.fa -r ${TEST_DATA}/r4.fa -o megahit_out2 -t 2 --keep-tmp-files --kmin-1pass
+        COMMAND ./megahit --test -t 2 --keep-tmp-files
+        COMMAND ./megahit --test -t 2 --keep-tmp-files --no-hw-accel
+        COMMAND ./megahit --test -t 2 --keep-tmp-files --kmin-1pass
 )
 
-add_dependencies(megahit megahit_core)
+add_dependencies(megahit megahit_core megahit_core_no_hw_accel)
 add_dependencies(simple_test megahit)
 
-install(TARGETS megahit_core DESTINATION bin)
+install(TARGETS megahit_core megahit_core_no_hw_accel DESTINATION bin)
 install(PROGRAMS src/megahit DESTINATION bin)
 install(DIRECTORY test_data DESTINATION share/${PROJECT_NAME})

Dockerfile

@@ -9,3 +9,5 @@ WORKDIR build
 RUN cmake -DCMAKE_BUILD_TYPE=Release ..
 RUN make -j4
 RUN make install
+RUN megahit --test --no-hw-accel
+RUN megahit --test --no-hw-accel --kmin-1pass

README.md

@@ -1,97 +1,76 @@
 MEGAHIT
 =======
 
-[![BioConda Install](https://img.shields.io/conda/dn/bioconda/megahit.svg?style=flag&label=BioConda%20install)](https://anaconda.org/bioconda/megahit)
-[![Build Status](https://travis-ci.org/voutcn/megahit.svg?branch=master)](https://travis-ci.org/voutcn/megahit)
-[![codecov](https://codecov.io/gh/voutcn/megahit/branch/master/graph/badge.svg)](https://codecov.io/gh/voutcn/megahit)
+[![BioConda Install](https://img.shields.io/conda/dn/bioconda/megahit.svg?style=flag&label=BioConda%20install)](https://anaconda.org/bioconda/megahit) [![Build Status](https://travis-ci.org/voutcn/megahit.svg?branch=master)](https://travis-ci.org/voutcn/megahit) [![codecov](https://codecov.io/gh/voutcn/megahit/branch/master/graph/badge.svg)](https://codecov.io/gh/voutcn/megahit)
 
 MEGAHIT is an ultra-fast and memory-efficient NGS assembler. It is optimized for metagenomes, but also works well on generic single genome assembly (small or mammalian size) and single-cell assembly.
 
-*News*
+*News: try v1.2.x!*
 ------
 
-MEGAHIT v1.2.0-beta is released. Main changes include
+MEGAHIT v1.2.x (beta) is released. Compared to v1.1.x, its changes include
 
 -   faster and more memory-efficient than before, by using [BMI2 instructions](https://en.wikipedia.org/wiki/Bit_Manipulation_Instruction_Sets), [sparsepp](https://github.com/greg7mdp/sparsepp) and [xxhash](https://github.com/Cyan4973/xxHash)
 -   refactored with C++11 features
 -   use CMake to build the project
 -   removal of GPU support
 
-It is highly recommended to use v1.2.0-beta. Past versions can be found at the [release](https://github.com/voutcn/megahit/releases) page.
+Please follow the instructions in [Getting Started](#gst) to try this new version.
+Past versions can be found at the [release](https://github.com/voutcn/megahit/releases) page.
 
-Getting Started
+<a name="gst"></a>Getting Started
 ---------------
 
-### Run with docker (recommended)
+### Running with Linux binaries or docker images (recommended)
 
-``` bash
-# in the directory with your input reads
-docker run -v $(pwd):/workspace -w /workspace --user $(id -u):$(id -g) vout/megahit \
-  megahit -1 YOUR_PE_READ_1.gz -2 YOUR_PE_READ_2.fq.gz -o YOUR_OUTPUT_DIR
+``` sh
+https://github.com/voutcn/megahit/releases/download/v1.2.1-beta/MEGAHIT-1.2.1-beta-Linux-static.tar.gz
+tar zvxf MEGAHIT-1.2.1-beta-Linux-static
+cd MEGAHIT-1.2.1-beta-Linux-static/bin/
+./megahit --test  # run on a toy dataset
+./megahit -1 YOUR_PE_READ_1.gz -2 YOUR_PE_READ_2.fq.gz -o YOUR_OUTPUT_DIR
 ```
 
-If your CPU does not support BMI2 or POPCNT instructions (typically you will see exit code -4), please use the docker images from `vout/megahit-no-popcnt-bmi2`. i.e.
+If your CPU does not support BMI2 and/or POPCNT, you may see "exit code -4". In this case, run MEGAHIT with `--no-hw-accel` option.
 
-``` bash
+You can also run MEGAHIT with its docker images.
+
+``` sh
 # in the directory with your input reads
-docker run -v $(pwd):/workspace -w /workspace --user $(id -u):$(id -g) vout/megahit-no-popcnt-bmi2 \
+docker run -v $(pwd):/workspace -w /workspace --user $(id -u):$(id -g) vout/megahit \
   megahit -1 YOUR_PE_READ_1.gz -2 YOUR_PE_READ_2.fq.gz -o YOUR_OUTPUT_DIR
 ```
 
-### Build from source
+### Building from source
 
 #### Prerequisites
 
--   For building: zlib, cmake, gcc/g++ &gt;= 5
+-   For building: zlib, cmake &gt;= 2.8, g++ &gt;= 4.8.4
 -   For running: gzip and bzip2
 
-#### Build and test
-
-1.  Obtain the source code
-
-    ``` bash
-    git clone https://github.com/voutcn/megahit.git
-    cd megahit
-    git submodule update --init
-    ```
-
-2.  Create the build directory
-
-    ``` bash
-    mkdir build && cd build
-    ```
-
-3.  Run cmake
-
-    ``` bash
-    cmake -DCMAKE_BUILD_TYPE=release ..
-    ```
-
-    If your CPU does not support BMI2 instructions (uncommon), run the following command instead
-
-        cmake -DUSE_BMI2=OFF -DCMAKE_BUILD_TYPE=release ..
-
-4.  Compile & test
-
-    ``` bash
-    make -j4
-    make simple_test  # will test MEGAHIT with a toy dataset
-    ```
-
-    If you need to install Megahit to your PATH, run `make install` in the build directory.
+``` sh
+git clone https://github.com/voutcn/megahit.git
+cd megahit
+git submodule update --init
+mkdir build && cd build
+cmake .. -DCMAKE_BUILD_TYPE=Release  # add -DCMAKE_INSTALL_PREFIX=YOUR_PREFIX if needed
+make -j4
+make simple_test  # will test MEGAHIT with a toy dataset
+# make install if needed
+```
 
 Usage
 -----
 
 To run MEGAHIT with default parameters:
 
-``` bash
+``` sh
 megahit -1 YOUR_PE_READ_1.fq.gz -2 YOUR_PE_READ_2.fq.gz -r YOUR_SE_READ.fq.gz -o YOUR_OUTPUT_DIR
 ```
 
 If you did not install Megahit to your PATH, just run Megahit with full-path, e.g.
 
-``` bash
+``` sh
 /PATH/TO/MEGAHIT/build/megahit
 ```
 

src/assembly/low_depth_remover.cpp

@@ -40,7 +40,7 @@ bool RemoveLocalLowDepth(UnitigGraph &graph, double min_depth, uint32_t max_len,
                          bool permanent_rm, uint32_t *num_removed) {
   bool is_changed = false;
   bool need_refresh = false;
-  auto &removed = *num_removed;
+  uint32_t removed = 0;
 
 #pragma omp parallel for reduction(+: removed)
   for (UnitigGraph::size_type i = 0; i < graph.size(); ++i) {
@@ -79,6 +79,7 @@ bool RemoveLocalLowDepth(UnitigGraph &graph, double min_depth, uint32_t max_len,
     bool set_changed = !permanent_rm;
     graph.Refresh(set_changed);
   }
+  *num_removed = removed;
   return is_changed;
 }
 

src/definitions.h

@@ -24,7 +24,7 @@
 #include <stdint.h>
 
 #ifndef PACKAGE_VERSION
-#define PACKAGE_VERSION "v1.2.0-beta"
+#define PACKAGE_VERSION "v1.2.1-beta"
 #endif
 
 #include "sdbg/sdbg_def.h"