1. doxeylab
  2. metannotate



Table of Contents

Support and Requirements
Installing Base/Command Line Version Concurrency
Built-in Webserver
A Note on Cache
Known Issues
Configure With Apache and Celeryd

Installation Instructions

Packaged Installation (Simpler)

This is a simpler to install with Docker packaging (similar to running in virtual box). However, you won't be able to (or it is inconvenient to) use the command line version of MetAnntoate.

It works on a Mac and on Linux, however, be sure that your laptop meets required RAM and space requirement.

Follow instructions here https://bitbucket.org/doxeylab/metannotateinstaller

The rest of the README is for Ubuntu 14.04+ users with fair technical knowledge. It is required if you want to use the command line version of MetAnnotate.

Support and Requirements

Debian/Ubuntu, with at least 6GB of space + the space required to store the Refseq database.

The following packages are required. Note that in One Command Install, the dependencies are _automatically_installed

  • python-dev (version 2.7)
  • build-essential
  • default-jre
  • git
  • wget
  • python-mysqldb
  • rabbitmq-server
  • libssl-dev
  • libffi-dev
  • sqlite3

One Command Install (install all dependencies, command-line and webserver)

For Ubuntu:14.04:

NOTE machine root password is required and will be asked for

Step 1:

#cd to home directory
sudo apt-get update
if [ ! `which git` ]; then
  sudo apt-get install -y git
git clone --depth=1 https://bitbucket.org/doxeylab/metannotate.git
cd metannotate
bash one_command_install.sh
# enter password as required

Sample run:

python run_metannotate.py --orf_files=data/MetagenomeTest.fa --hmm_files=data/hmms/RPOB.HMM --reference_database=data/ReferenceTest.fa --output_dir=test_output --tmp_dir=test_tmp --run_mode=both

Note that in the example above, a [tiny] test reference database was specified to make process faster. If not specified, the default data/Refseq.fa database is used. You should now see run outputs in test_output directory.

For more options:

python run_metannotate.py --help


You can speed up metannotate by specifying a greater concurrency in metannotate/concurrency.txt. This will have the effect of increasing concurrency for HMMER, FastTree, and pplacer commands.

Built-in Webserver

The web server is installed as part of One Command Install

There are two files that provide additional configuration:

  • metagenome_directories_root.txt
  • metagenome_directories.txt

These files are already placed in the main metannotate directory.

metagenome_directories_root.txt contains the root path for all metagenome directories that will be read by the program. metagenome_directories.txt lists all the directories in that root directory that should be read as metagenome directories (in the case that you have other directories in the root directory that shouldn't be interpreted as metagenome directories).

To start server:

bash start-server.sh 
# enter password as prompted

To stop server:

bash stop.sh
# enter password as prompted

Output of server is at out.txt for debugging purposes.

Occasionally (when no job is running), you can run rm tmp/* to free up disk space.

A Note on Cache

MetAnnotate has a cache layer that is cleaned every week. You can modify your Cronjob by typing crontab -e to edit the clean up interval.

Repeating the same HMM files is much faster with cache (95% speed up).

Known Issues

If an ORF file has no HMM hit, the script will terminate and job will error out.

Configure with Apache and Celeryd (More technical)

Configuring the app to run with an Apache server and an existing celery daemon process is safer and ensures that the tool runs on startup.

Create a Celery Daemon Process

# Names of nodes to start
#   most will only start one node:


# Absolute or relative path to the 'celery' command:

# App instance to use

# Where to chdir at start.
CELERYD_CHDIR=""  # Change as needed to directory containing tasks.py.

# Extra command-line arguments to the worker

# %N will be replaced with the first part of the nodename.

# Workers should run as an unprivileged user.
#   You need to create this user manually (or you can choose
#   a user/group combination that already exists, e.g. nobody).

# If enabled pid and log directories will be created if missing,
# and owned by the userid/group configured.
  • sudo update-rc.d celeryd defaults
  • sudo /etc/init.d/celeryd restart

Integrating with Apache.

  • sudo apt-get install libapache2-mod-wsgi
  • sudo vim /etc/apache2/sites-available/000-default.conf
<VirtualHost *:80>
  #ServerName www.example.com

  ServerAdmin webmaster@localhost
  DocumentRoot /var/www/html

  WSGIDaemonProcess metannotate
  WSGIProcessGroup metannotate
  WSGIScriptAlias / /path/to/app.wsgi # Change as needed.

  <Directory /path/to/dir/containing/app.wsgi> # change as needed.
    Options All
    AllowOverride All
    Require all granted

  ErrorLog ${APACHE_LOG_DIR}/error.log
  CustomLog ${APACHE_LOG_DIR}/access.log combined

Change Permissions of IO Directories.

  • sudo chgrp www-data realinput/
  • sudo chown www-data realinput/
  • sudo chmod g+w realinput/
  • sudo chgrp www-data input/
  • sudo chown www-data input/
  • sudo chmod g+w input/
  • sudo chgrp www-data output/
  • sudo chown www-data output/
  • sudo chmod g+w output/
  • sudo chgrp www-data databases/
  • sudo chown www-data databases/
  • sudo chmod g+w databases/
  • sudo chgrp www-data tmp/
  • sudo chown www-data tmp/
  • sudo chmod g+w tmp/
  • sudo chgrp www-data session/
  • sudo chown www-data session/
  • sudo chmod g+w session/


Upon a restart of the system, all components will be restarted as well. If necessary, the components can be restarted manually using the following commands:

  • sudo /etc/init.d/rabbitmq-server restart
  • sudo /etc/init.d/celeryd restart
  • sudo /etc/init.d/apache2 reload