Glycam网站的本地配置

环境准备

glycam源码编译需要c++17标准，服务器上现有gcc/g++版本为4.8.5，不支持c++17标准，安装更高版本的gcc/g++需要源码编译和更新的glibc库，很麻烦。因此使用conda环境，在conda环境中安装高本版的gcc/g+g++。glycam需要依赖swig，但服务上swig版本为2.0.10，glycam在该版本上有bug，因此也在conda环境中安装更新版本的swig。为了避免conda包冲突，在创建conda环境时就将所有需要安装的包一并安装。

(base) [chpeng@localhost ~]$ conda create -n glycam libboost gcc_linux-64 gxx_linux-64 python=3.8 swig=3.0.12

获取glycam源码

先克隆gems源码，由于github用不了，使用镜像网站，将github.com替换为github.com.cnpmjs.org

(glycam) [chpeng@localhost glycam]$ git clone https://github.com.cnpmjs.org/GLYCAM-Web/gems.git
正克隆到 'gems'...
remote: Enumerating objects: 6859, done.
remote: Counting objects: 100% (2042/2042), done.
remote: Compressing objects: 100% (929/929), done.
remote: Total 6859 (delta 1531), reused 1582 (delta 1100), pack-reused 4817
接收对象中: 100% (6859/6859), 19.82 MiB | 10.87 MiB/s, done.
处理 delta 中: 100% (4766/4766), done.

克隆完毕后，当前目录存在gems目录，进入gems目录克隆gmml源码

(glycam) [chpeng@localhost gems]$ git clone https://github.com.cnpmjs.org/GLYCAM-Web/gmml.git
正克隆到 'gmml'...
remote: Enumerating objects: 28527, done.
remote: Counting objects: 100% (4976/4976), done.
remote: Compressing objects: 100% (1863/1863), done.
remote: Total 28527 (delta 3752), reused 4046 (delta 3018), pack-reused 23551
接收对象中: 100% (28527/28527), 92.35 MiB | 12.19 MiB/s, done.
处理 delta 中: 100% (18018/18018), done.

编译

设置环境变量

(glycam) [chpeng@localhost gems]$ export PYTHON_HOME=/home/chpeng/anaconda3/envs/glycam/include/python3.8
(glycam) [chpeng@localhost gems]$ export GEMSHOME=/home/chpeng/glycam/gems
(glycam) [chpeng@localhost gems]$ export GEMSMAKEPROCS=20
(glycam) [chpeng@localhost gems]$ export CPLUS_INCLUDE_PATH=/home/chpeng/anaconda3/envs/glycam/include:$CPLUS_INCLUDE_PATH

PYTHON_HOME指定当前conda环境中python3.8的headers目录，GEMSHOME指定gems源码目录，GEMSMAKEPROCS指定编译时需要多少个核进行编译，CPLUS_INCLUDE_PATH将/home/chpeng/anaconda3/envs/glycam/include下的headers加入搜索路径，主要是为了使用boost库。

修改make.sh脚本

make.sh脚本中会克隆github上的源码，所有需要修改make.sh，将github.com替换为github.com.cnpmjs.org。

替换完后的部分代码:

    if [ ! -d "${theDir}/.git" ]; then
        echo ""
        echo "MD_Utils repo does not exist. Attempting to clone."
        git clone -b ${mdBranch} https://github.com.cnpmjs.org/GLYCAM-Web/MD_Utils.git ${theDir}
        if [ ! -d "${theDir}/.git" ]; then
                echo ""
                echo "Error:  Unable to clone MD_Utils.  Some functions will be unavailable."
                echo "You can try again on your own using the following command."
                echo "You will not need to remake GEMS or GMML after cloning."
                echo ""
                echo "git clone -b ${mdBranch} https://github.com.cnpmjs.org/GLYCAM-Web/MD_Utils.git ${theDir}"
                return 1
        fi
        echo "Cloning of MD_Utils was successfil"
        return 0
    fi

进行编译

conda的gcc_linux-64 gxx_linux-64安装后会自动将CC和CXX变量设置为新安装的gcc和g++，不会对gcc和g++进行设置。glycam编译时不使用CC和CXX变量。glycam需要编译gmml源码，gmml源码使用gmml/make.sh生成Makeflie，其中会调用qmake生成Makefile，我们可以使用qmake的环境变量QMAKE_CC和QMAKE_CXX指定gcc和g++的路径，此外，还需使用QMAKE_LINKER指定link程序为新安装的g++，打开gmml/make.sh文件，找到生成Makefile的那一行，大概在163行左右，添加QMAKE_CC，QMAKE_CXX和QMAKE_LINKER变量，修改后像下面这个样子

 qmake -project -t lib -o gmml.pro "QMAKE_CXXFLAGS += -Wall -W -std=c++17 ${DEBUGOPTIONS} ${OPTIMIZE}" "QMAKE_CFLAGS += -Wall -W ${DEBUGOPTIONS}" "${NO_OPTIMIZE}" "DEFINES += _REENTRANT" "CONFIG = no_lflag_merge" "unix:LIBS = -L/usr/lib/x86_64-linux-gnu -lpthread" "OBJECTS_DIR = build" "DESTDIR = lib" "QMAKE_CC = /home/chpeng/anaconda3/envs/glycam/bin/x86_64-conda-linux-gnu-cc" "QMAKE_CXX = /home/chpeng/anaconda3/envs/glycam/bin/x86_64-conda-linux-gnu-c++" "QMAKE_LINK = /home/chpeng/anaconda3/envs/glycam/bin/x86_64-conda-linux-gnu-c++" -r src/ includes/ -nopwd

修改完毕后并保存退出，在gems目录下运行make.sh

(glycam) [chpeng@localhost gems]$ pwd
/home/chpeng/glycam/gems
(glycam) [chpeng@localhost gems]$ ./make.sh

编译到一半时会出现下面的错误

/home/chpeng/anaconda3/envs/glycam/bin/../lib/gcc/x86_64-conda-linux-gnu/11.2.0/../../../../x86_64-conda-linux-gnu/bin/ld: build/main.o: in function `main':
/home/chpeng/glycam/gems/gmml/internalPrograms/GlycoproteinBuilder/main.cpp:6: multiple definition of `main'; build/main.o:/home/chpeng/glycam/gems/gmml/internalPrograms/GlycoproteinBuilder/main.cpp:6: first defined here
collect2: error: ld returned 1 exit status
make: *** [lib/libgmml.so.1.0.0] 错误 1

这是因为gmml/Makefile中OBJECTS变量含有两个build/main.o造成的，修改gmml/Makefile，删去一个build/main.o只保留一个，然后保存。在gmml目录下运行make -j 10 -f Makefile继续编译（由于gmml/Makefile是make.sh自动产生的，因此我们只能直接修改Makefile，让gmml编译完后，在运行make.sh程序，接着gmml编译后的步骤运行，因为每次运行make.sh，gmml/Makefile会被覆盖成新的文件）。

(glycam) [chpeng@localhost gmml]$ pwd
/home/chpeng/glycam/gems/gmml
(glycam) [chpeng@localhost gmml]$ make -j 10 -f Makefile

gmml编译完成后，回到gems目录，执行make.sh继续编译

(glycam) [chpeng@localhost gmml]$ cd ..
(glycam) [chpeng@localhost gems]$ pwd
/home/chpeng/glycam/gems
(glycam) [chpeng@localhost gems]$ ./make.sh

然后会出现下面的错误

Compiling wrapped gmml library in python ...
g++: 错误：unrecognized command line option ‘-std=c++17’


Warning:  gmml python interface has not been compiled correctly

这是因为make.sh在编译python模块时，直接使用g++，但是g++指向的是系统上旧的g++，我们将make.sh里的g++改为新安装的g++，修改后类似下面这样：

    if [ -f $PYTHON_FILE ]; then
        echo "Using $PYTHON_FILE header file."
        if [ -f "gmml_wrap.cxx" ]; then
            echo "Compiling wrapped gmml library in python ..."
            /home/chpeng/anaconda3/envs/glycam/bin/x86_64-conda-linux-gnu-c++ -std=c++17 -O3 -fPIC -c gmml_wrap.cxx -I"$PYTHON_HEADER_HOME"
        else
            echo "Warning:  gmml_wrap.cxx does not exist."
        fi
    else
        echo "Warning:  $PYTHON_FILE not found !"
    fi


    echo ""
    if [[ -f "gmml_wrap.o" ]]; then
        echo "Building python interface ..."
        /home/chpeng/anaconda3/envs/glycam/bin/x86_64-conda-linux-gnu-c++ -std=c++17 -shared gmml/build/*.o gmml_wrap.o -o _gmml.so
    else
        echo "Warning:  gmml python interface has not been compiled correctly."
    fi

然后使用make.sh继续编译

(glycam) [chpeng@localhost gems]$ alias g++="/home/chpeng/anaconda3/envs/glycam/bin/x86_64-conda-linux-gnu-c++"
(glycam) [chpeng@localhost gems]$ ./make.sh

gems/gmml/internalPrograms/GlycoproteinBuilder/bin/gpBuilder和/home/chpeng/glycam/gems/gmml/internalPrograms/CarbohydrateBuilder/bin/carbBuilder为gmmly源代码里提供的，是gmml作者自己编译的，不能直接拿来用（会报动态链接库和glibc库的错误），我们需要删除他们再重新编译一个。

(lywu) [chpeng@localhost CarbohydrateBuilder]$ pwd
/home/chpeng/glycam/gems/gmml/internalPrograms/CarbohydrateBuilder
(lywu) [chpeng@localhost CarbohydrateBuilder]$ export GEMSHOME=/home/chpeng/glycam/gems
(lywu) [chpeng@localhost CarbohydrateBuilder]$ g++ -std=c++17 -I $GEMSHOME/gmml -L$GEMSHOME/gmml/bin/ -Wl,-rpath,$GEMSHOME/gmml/lib/ main.cpp -lgmml -o bin/carbBuilder 


(lywu) [chpeng@localhost GlycoproteinBuilder]$ pwd
/home/chpeng/glycam/gems/gmml/internalPrograms/GlycoproteinBuilder
(lywu) [chpeng@localhost GlycoproteinBuilder]$ g++ -std=c++17 -I $GEMSHOME/gmml -L$GEMSHOME/gmml/bin/ -Wl,-rpath,$GEMSHOME/gmml/lib/ main.cpp -lgmml -o bin/gpBuilder

测试

在gems目录下，使用test_installation.bash进行测试

(glycam) [chpeng@localhost gems]$ pwd
/home/chpeng/glycam/gems
(glycam) [chpeng@localhost gems]$ ./test_installation.bash




This test should take less than 10 seconds to run on most modern computers.


This test will compare these files:


        updated_pdb.txt      --   this is the file the test should generate
        test_pdb.txt.save    --   this is the file to which it should be identical




Beginning test.






Checking for diffeences between test output and the standard output.






The test passed.

使用

CarbohydrateBuilder example

(lywu) [chpeng@localhost CarbohydrateBuilder]$ pwd
/home/chpeng/glycam/gems/gmml/internalPrograms/CarbohydrateBuilder
(lywu) [chpeng@localhost CarbohydrateBuilder]$ mkdir outputs
(lywu) [chpeng@localhost CarbohydrateBuilder]$ ./bin/carbBuilder exampleLibrary.txt _ outputs/ ../../dat/prep/GLYCAM_06j-1.prep

GlycoproteinBuilder example

(lywu) [chpeng@localhost GlycoproteinBuilder]$ pwd
/home/chpeng/glycam/gems/gmml/internalPrograms/GlycoproteinBuilder
(lywu) [chpeng@localhost GlycoproteinBuilder]$ ./bin/gpBuilder input.txt tests/tough/

gpBuilder的两个参数input.txt和tests/tough/用于拼接成最终的input文件，即拼接为tests/tough/input.txt，预测第一个参数指定输入文件名，第二个参数指定输入文件所在的目录，由于使用的是字符串拼接，因此tests/tough/最后面的'/'是必须的，否则程序会报错。

Docker image

FROM ubuntu:20.04


LABEL maintainer "fanxp <897488736@qq.com>"


# set env
ENV TZ=Asia/Shanghai
ENV LD_LIBRARY_PATH=/opt/gems/gmml/lib:$LD_LIBRARY_PATH
ENV PATH=/opt/gems/gmml/internalPrograms/GlycoproteinBuilder/bin:$PATH


# Set the timezone
RUN apt-get update && apt-get install -y --no-install-recommends --no-install-suggests tzdata && \
    ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone && \
    dpkg-reconfigure --frontend noninteractive tzdata && \
    rm -rf /var/lib/apt/lists/*


# add qt4 repository and install dependences
RUN apt-get update && \
    apt-get install -y --no-install-recommends --no-install-suggests software-properties-common && \
    add-apt-repository -y ppa:rock-core/qt4 && \
    apt-get install -y --no-install-recommends --no-install-suggests openssl git python3 \
            python3-dev swig qt4-qmake qt4-dev-tools qt4-default build-essential \
            ca-certificates vim libboost-dev && \
    rm -rf /var/lib/apt/lists/*


# compile glycam
RUN cd /opt && git clone https://github.com.cnpmjs.org/GLYCAM-Web/gems.git && \
    cd gems && git clone https://github.com.cnpmjs.org/GLYCAM-Web/gmml.git && \
    export PYTHON_HOME=/usr/include/python3.8 && \
    export GEMSHOME=/opt/gems && \
    export GEMSMAKEPROCS=20 && \
    sed -i 's/github\.com/github\.com\.cnpmjs\.org/g' make.sh && \
    ./make.sh

这个Docker image里没有对gems/gmml/internalPrograms/GlycoproteinBuilder/bin/gpBuilder和/home/chpeng/glycam/gems/gmml/internalPrograms/CarbohydrateBuilder/bin/carbBuilder进行重新编译。