MRS is full text retrieval software optimized for large, flat file databanks containing biological and medical information. After indexing these databanks, they can be searched using a command line interface and when running the server, using a web browser or using web services (SOAP, REST).
Searching can be done using simple keyword searches, but also using more advanced boolean queries, searches limited to parts of a document, searches for 'strings' (sequences of words). Results are sorted by relevance, when applicable. MRS also has a protein BLAST search engine built in.
Previous versions used to be hard to set up and configure, version 6 tries to address this by combining all configuration information in one place and offering a graphical interface to manipulate this configuration. The update process is fully automated and easy to follow.
MRS runs on both Linux and Windows although a recent C++ compiler is required to build it. The Linux version comes with an init.d script. For windows, I still have to implement the code to run MRS as a service.
See the changelog file for an overview of what has changed since version 5.
As an example, we will index the source code for MRS. First go to the Admin pages and click the 'add' button in the Databanks tab page. Select the newly created databank (first of the list) and enter an ID (e.g. m6src) and check the Enabled checkbox.
Then you enter the path to the source files. If the path you enter is a relative path, MRS will prepend it with the 'raw files' directory specified in the Main tab. If the path is absolute however, it will take this path directly. So we enter for Source files:
/home/me/mrs-6.0.0/*.{cpp,h,inl,c}
Now if you tick the Recursive checkbox, MRS will include all files that have a name that ends with either .cpp, .h, .inl or .c found in any directory under /home/me/mrs-6.0.0/.
As parser use the generic parser. This parser simply indexes all text and assigns an incrementing number as ID to each document. It assumes each file is a document.
Now go the Main tab and click the Restart button (to save the configuration file) and in a terminal type
mrs build m6src
Restart the server again to load the newly created databank and you'll see the m6src is now searchable.
Building MRS
You need gcc 8.3.0 and boost >= 1.65 to build MRS. Intel compilers are supported and Visual Studio 2010 is OK too.
MRS comes with a Dockerfile for building mrs. It can also be used in production. To build the docker image, run the following in the root project folder:
docker build -t mrs .
To run the container in development mode, allowing you to change the source on the host (your machine) and have the changes picked up in the container, run:
docker run -v /home/jon/projects/mrs:/app -p 18090:18090 -it mrs
This will run interactively and the MRS web interface is accessible on the host at http://localhost:18090.
Your production environment might differ so take the following with a pinch of salt. Instead of mapping the source on the local machine, you probably want to map the data so it's not destroyed if the container gets deleted.
docker run -v /srv/mrs/data:/srv/mrs-data -p 18090:18090 mrs
This maps the data on the host at /srv/mrs/data to the mount point
/srv/mrs-data in the container, allowing mrs in the container to write data
to the host.
You need to have the following to build MRS from source code:
- gcc 4.6
- libboost >= 1.65
- libperl-dev Development files for Perl
Then run:
./configure
make
make install
You can set several options using the configure script, see ./configure --help to see which options are available.
After installing you have a single executable called 'mrs'. This program works with command as the first parameter followed by options depending on the command.
Command can be one of:
blast Do a blast search
build (Re-)build a databank
dump Dump index data
entry Retrieve and print an entry
fetch Fetch/mirror remote data for a databank
info Display information and statistics for a databank
query Perform a search in a databank
server Start or Stop a server session, or ask the status
vacuum Clean up a databank reclaiming unused disk space
validate Perform a set of validation tests
update Same as build, but does a fetch first
password Generate password for use in configuration file
The first thing you should do after installing MRS 6 is run the password command and add a new user/password for the admin account. The password is stored encrypted in the configuration file. You then start a server using the command:
mrs server start
If this was successful you can now access the MRS website at the address of the local machine and the port specified (default: 18090). Use your web browser to go to this address and click the Admin link on the right top of the page. Then follow the instructions there.
The default setup will automatically update the enabled databanks. You can update databanks manually as well of course using the 'mrs update db' command.
Building MRS on Windows is not trivial, don't try it unless you have experience building Windows software using Visual Studio.
First of all, you need to have MSVC set up correctly to build 64 bit executables. You then need to build the Boost libraries with static runtime libraries. The mrs project file assumes you've installed Boost in C:\Boost and you're using version 1.65. Also, make sure you build Boost with zlib and bz2 support. The way I did it is, download boost, extract it, run boostrap and then:
bjam link=static runtime-link=static threading=multi address-model=64
-sBZIP2_SOURCE="C:/Users/maarten/projects/bzip2-1.0.6"
-sZLIB_SOURCE="C:/Users/maarten/projects/zlib-1.2.5"
-sICU_PATH="C:/Users/maarten/projects/icu" stage install
Then you need to build a custom Perl interpreter. Download the source code and edit the file perl-5.14.1/win32/Makefile uncomment the line containing:
CCTYPE = MSVC100
and build and install perl using nmake. This should install a perl in C:\Perl and here you can also file the perl514.dll file which you need to place in the m6\msvc\x64{Debug,Release} folders.
Libzeep is contained in the source tar but the solution file assumes it is located in the same directory as the m6 directory. Either move it, or change the paths in the m6 project file.