/GNPS_LCMSDashboard

Primary LanguagePythonOtherNOASSERTION

GNPS LCMS Visualization Dashboard

Example Sources of Data

  1. GNPS Analysis Tasks - mzspec:GNPS:TASK-d93bdbb5cdda40e48975e6e18a45c3ce-f.mwang87/data/Yao_Streptomyces/roseosporus/0518_s_BuOH.mzXML:scan:171
  2. GNPS/MassIVE public datasets - mzspec:MSV000084951:AH22
  3. MassiVE Proteomics datasets - mzspec:MSV000079514:Adult_CD4Tcells_bRP_Elite_28_f01
  4. MassiVE Proteomics dataset large - mzspec:MSV000083508:ccms_peak_centroided/pituitary_hypophysis/Trypsin_HCD_QExactiveplus/01697_A01_P018020_S00_N01_R2.mzML:scan:62886
  5. Metabolights public datasets - mzspec:MTBLS1124:QC07.mzML

Instrument Vendors for Metabolomics

  1. Thermo - mzspec:MSV000084951:AH22
  2. Thermo MS3 - mzspec:MSV000084765:Leptocheline_MS3_DDA_IT_5
  3. Sciex - mzspec:MSV000085042:QC1_pos-QC1
  4. Bruker - mzspec:MSV000086015:StdMix_02__GA2_01_55623
  5. Waters - mzspec:MSV000084977:OEPKS7_B_1_neg
  6. Agilent - mzspec:MSV000084060:KM0001

Data Types for Proteomics

  1. Sciex - SWATH - mzspec:MSV000085570:170425_01_Edith_120417_CCF_01

Example Use Cases

Quick analysis of QC data

Here is the USI for a QC run

mzspec:MSV000085852:QC_0

What we can easily do is paste in the QC molecules and pull them out in one fell swoop:

271.0315;278.1902;279.0909;285.0205;311.0805;314.1381

You can try it out at this URL

Quickly Compare Multiple files

mzspec:MSV000085852:QC_0 mzspec:MSV000085852:QC_1 mzspec:MSV000085852:QC_2

271.0315;278.1902;279.0909;285.0205;311.0805;314.1381

Test Link

API

We aim to provide several APIs to programmatically get data.

Image Preview of MS run

/mspreview?usi=<usi>

Development

There are several ways to get GNPS Dashboard working locally, our preferred and recommended way is using docker/docker-compose as it provides a more consistent experience.

The initial steps are identical:

  1. Fork the GNPS Dashboard repository
  2. Clone down to your system
  3. Download Feature Finding tools (Dinosaur and MZmine 2) with get scripts

Docker

To get everything up and running, we've created a make target for you to get docker up and running:

make server-compose-interactive

The requirements on your local system are:

  1. Docker
  2. Docker Compose

This will bring the server up on http://localhost:6548.

Conda

  1. Install Python3 within conda
  2. Install all packages from the requirements.txt
  3. Install packages via conda
  4. Start the dashboard locally (defaults to http://localhost:5000)

Example shell

# make sure to have Python3 installed via conda (preferably 3.8)
conda install -c conda-forge datashader
conda install -c conda-forge openjdk

# install requirements
pip install -r requirements.txt

# run or debug the GNPS Dashboard with Python 3 on http://localhost:5000
python ./app.py

# on problem, maybe install the following (tested on Windows 10 with WSL2 Ubuntu) 
sudo apt-get install libffi-dev

Supporting new data sources

Since we utilize a USI to find datasets, there are a limited number of locations we can grab data from. If you want to provide a new data source, you'll have to implement the following

  1. USI Specification that denotes what the resource is and how to get data
  2. Update the code in download.py, specifically in _resolve_usi_remotelink to implement how to get the remote URL for your new USI.

Testing

To run our unit tests,

cd test
pip install pytest
pip install pytest-xdist
pip install pytest-profiling
make all

Useful links for developers

Dash and plotly documentations

Production Deployment

One major thing about production deployemnts is the DNS routing. You want to do the following steps to have everything route properly:

  1. Create a DNS entry in your DNS server for the domain you want to use (e.g. dashboard.gnps2.org) and point it to the server you're running this on
  2. Copy .env_template to .env and update the domain name to the one you want to use
  3. Run a reverse proxy (https://github.com/mwang87/GNPS_ExtensionsReverseProxy)
  4. Run in production mode make server-compose-production