curatr

This respository contains the open-source spectral library curation tool: Curatr. It accompanies the paper

Andrew Palmer, Prasad Phapale, Dominik Fay, Theodore Alexandrov, Curatr: a web application for creating, curating, and sharing a mass spectral library, 2017 biorxiv

It is written in python using the django web framework and is being developed by the Alexandrov Team at EMBL Heidelberg.

Requirements

• Ubuntu 14.04 or 16.04
• git 2.6
• python 3.6

Installation

We recommend installing curatr and its python dependencies inside a conda virtual environment as follows:

Install miniconda

```
cd $HOME/Downloads
wget https://repo.continuum.io/miniconda/Miniconda2-latest-Linux-x86_64.sh
bash Miniconda2-latest-Linux-x86_64.sh
source .bashrc
conda update conda
```

Download curatr

Create a convenient directory, for example curatr and clone the repository there:

mkdir curatr
cd curatr
git clone https://github.com/alexandrovteam/curatr

Initialize and activate a 'curatr' environment with all python dependencies

```
conda create -y -n venv python=3.6
source activate venv
pip install --upgrade pip
pip install pip-tools
conda install cython
conda install -y -c openbabel openbabel
pip-sync
```

Run Server in debug mode

Django Settings

Make and edit a local copy of the django settings file

cd mcf_standard_browser/mcf_standard_browser/mcf_standard_browser/
cp settings_template.py settings.py

Open settings.py and edit the following fields

• SECRET_KEY: replace the secret key with a securely generated key e.g. copy-and-paste from a python terminal

  python
  

  from the python terminal:

  import random
  print(''.join([random.SystemRandom().choice('abcdefghijklmnopqrstuvwxyz0123456789!@#$%^&*(-_=+)') for i in range(50)]))

• CELERYD_CONCURRENCY: set this number to the desired amount of workers (degree of parallelism) or remove the line to use all CPUs

Initialize the DB

python manage.py migrate

Create an admin user

python manage.py createsuperuser

Set up a message broker (example: RabbitMQ)

1. [Install RabbitMQ] (https://www.rabbitmq.com/download.html)

   echo "deb http://www.rabbitmq.com/debian/ testing main" >> /etc/apt/sources.list
   curl http://www.rabbitmq.com/rabbitmq-signing-key-public.asc | sudo apt-key add -
   sudo apt-get update
   sudo apt-get install rabbitmq-server
   

2. (Optional) Configure a user

   # RabbitMQ installs with a default username / password of guest / guest
   # you can change that by creating a new user
   rabbitmqctl add_user myuser mypassword
   rabbitmqctl add_vhost myvhost
   rabbitmqctl set_permissions -p myvhost myuser ".*" ".*" ".*"
   

   And adjust the URL in settings.py: BROKER_URL = "amqp://myuser:mypassword@localhost:5672/myvhost". Find more details [here] (http://www.rabbitmq.com/man/rabbitmqctl.1.man.html)

3. Sync your DB:

   python manage.py migrate djcelery
   

Run Server

1. Start RabbitMQ: sudo service rabbitmq-server start (on Ubuntu)
2. Start Celery: python manage.py celeryd. Useful parameters:
   • --verbosity=2
   • --loglevel=DEBUG
3. Start the server: python manage.py runserver

Curatr should now be visible on localhost:8000

Getting Going

We have provided an example molecular library and data in the media directory

1. Add standards to library
   • Single
     1. Add molecule: See detailed screenshots in the manuscript supporting information
     2. Add standard: See detailed screenshots in the manuscript supporting information
   • Batch
     1. From the "Curate" menu select "Add Standard"
     2. Click 'Add Batch'
     3. Select a properly formatted .tsv file containing the standards to add this method will silently overwrite existing files allowing batch updating
     4. The task runs asynchronously so after a few seconds hit refresh
2. Add adducts. Curate -> Add Adduct [nM+X]^(y)
   • NM = n (number of metabolite molecules)
   • Delta formula = X (e.g. "-H2O+H")
   • Charge = y
3. Add data: See detailed screenshots in the manuscript supporting information
4. Curate data: See detailed screenshots in the manuscript supporting information

Advanced

• Further users can be configured from http://localhost:8000/admin

Deployment

It is a bad idea to run the django development server in a production environment. It's probably OK to run behind a corporate/university firewall but should not be exposed to the internet. There are excellent detail guides available for various deployment platforms e.g. from mozilla on deploying to Heroku

Licence

The source code in this repository is distributed under the Apache 2.0 licence.