Skip to content

HTS Bioinf - ELLA production

Scope

ELLA: software for interpretation of genetic variants.

This procedure describes tasks related to managing ELLA in production:

  • Handling import and deletion of analyses
  • Managing gene panels
  • Managing users

Responsibility

Responsible person: A bioinformatician on production duty. ELLA core developers can also assist with the tasks described here, to offload work.


ELLA setup

Location of the application

The source code and other files related to ELLA are located at /ess/p22/data/durable/production/ella. The directories and scripts mentioned later in this document are relative to this unless given with absolute path.

Instances

ELLA runs on the TSD VM p22-ella-01. There are four instances of ELLA:

Instance URL Supervisor URL Description
prod http://p22-ella-01:5000 http://p22-ella-01:9007 Production. Should be up at all times.
staging http://p22-ella-01:7000 http://p22-ella-01:9008 Used to test new features and releases.
test http://p22-ella-01:7001 http://p22-ella-01:9009 Used for various (user) testing purposes.*
validation http://p22-ella-01:7002 http://p22-ella-01:9010 Used to validate new analysis methods.
* User testing purposes could be training of new users, testing new filters, new features etc.

Supervisors

Each instance of ELLA is controlled by a Supervisor process accessible at http://p22-ella-01:9000 (superdupervisor) running locally on the VM. Each instance corresponds to a Singularity container that can be started/stopped from the superdupervisor. Inside each Singularity container instance is another Supervisor process controlling the programs ELLA is running. These programs are:

  • api: The ELLA application
  • polling: Program for communicating with anno/anno-targets
  • watcher: Program for watching ella-<instance>/data/analyses/incoming for new analyses, and importing them.

ELLA's command line interface (CLI)

ELLA's CLI (ella-cli) can be accessed through ./ops/<instance>-cli.sh. Run ella-cli --help to see available options. See also official documentation.

Production routines

Importing analyses to ELLA

New analyses are automatically imported into ELLA when delivered from:

  1. The bioinformatic pipeline.
  2. Reanalysis from anno/anno-targets (ordered through ELLA import).

New analyses are delivered to ella-<instance>/data/analyses/incoming. The import is done by the ELLA program analysis-watcher.

White-/blacklists

Each instance has two associated files that determine which analyses should be imported:

  • ops/<instance>-watcher-whitelist.txt
  • ops/<instance>-watcher-blacklist.txt

To allow an analysis / project to be imported, add a regex to the whitelist file. To block an analysis / project from being imported, add a regex to the blacklist file.

Troubleshooting missing analyses

The watcher checks for new analyses (and updates to the whitelist / blacklist files) every thirty seconds. Check the watcher logs from the Supervisor UI to see if an import succeeds, or if not, why it does not succeed.

Common causes of analyses not being imported include:

  • Not matching whitelist or matching blacklist
  • Gene panel not in database
  • Analysis already in database
  • Analysis already in ella-<instance>/data/analyses/imported

The presence of an analysis in imported but not in the database suggests either that the analysis has been deleted from the database or that the database has not been updated. The latter should never be the case for the prod-instance.

The presence of an analysis in imported but not in the UI suggests that the corresponding gene panel is not tied to the user group under which the user is currently signed in. See below for possible solutions to this.

Deleting analyses

Warning

Deleting an analysis will remove all work performed on that analysis. This action cannot be undone. Should the need arise to access it, the application logs contain information on the work performed on all analyses, including deleted ones.

To delete an imported analysis:

  1. Search for and open the analysis in ELLA. Note down the analysis ID from the URL (ending with /workflows/analyses/{analysis ID}/). Double check that the ID is correct.

  2. Open ELLA's CLI and run:

    ella-cli delete analysis {analysis ID}
    

    Enter the reason after the prompt, e.g. "Requested by {username} on 190425 by email".

  3. When you are asked for confirmation, double check that the analysis name is correct. Type "y" and Enter to confirm.

Adding new gene panels

To add new gene panels, they must be first imported into the database and then added to the relevant user groups.

I. Import gene panels into the database

  1. Copy the relevant files into ella-<instance>/data/fixtures/genepanels

  2. Add them to the local Git repository (there is one in the fixtures directory).

  3. Open ELLA's CLI and run for each gene panel:

    ella-cli deposit genepanel --folder ../data/fixtures/genepanels/{gene panel folder name}
    

    Warning

    Make sure there is no / at the end of the gene panel folder name.

II. Add new gene panels to user groups

After you have added all the gene panels you want, you need to update the relevant user groups:

  1. Enter the directory ella-<instance>/data/fixtures and edit the file create_usergroups.py. Add the gene panels to the relevant groups. Please add them in alphabetical order.

  2. Run:

    source ../../../sw/sourceme.
    python create_usergroups.py > usergroups-new.json
    
  3. Perform a diff on the old one like this diff usergroups-new.json usergroups.json and make sure everything looks fine.

  4. Run:

    mv usergroups-new.json usergroups.json
    
  5. Commit the files to Git (provide reason in commit message):

    git add create_usergroups.py usergroups.json
    git commit -m "your-message"
    
  6. Go back to ella-cli in \<instance> and run

    ella-cli users add_groups ../data/fixtures/usergroups.json
    
  7. Confirm in ELLA that the gene panels are available for the corresponding user groups.

Changes to users

Reset password

If a user forgets his/her password:

  1. Reset the user's password using ELLA's CLI:

    ella-cli users reset_password <username>
    
  2. Send the new password in a personal email, including a note that the user needs to change the password before logging in again ("Change password" on ELLA login page).

Add or edit user

To add a new or edit an existing user (including changing the associated user group):

  1. Open ella-<instance>/data/fixtures/users.json and add or change the user with the corresponding user group. For OUS users, the OUS user name (same as in OUS emails) should be used.

  2. Commit the changes to Git (provide reason in commit message):

    git add users.json
    git commit -m "your-message"
    
  3. Open ELLA's CLI and run:

    ella-cli users add_many ../data/fixtures/users.json`
    

    This should print out the changes, including newly created user(s) and their temporary passwords.

  4. Provide the user name and password to each new user in a personal email, along with instructyions on how to proceed. The e-mail should contain something along the lines of:

    Har lagt deg til som bruker i ELLA, brukergruppe [user group]. Velkommen!
    
    Merk at innlogging i TSD via VMware (med brukernavn «p22-…» og passord fra TSD) må være i orden før du kan åpne ELLA. Se prosedyre «Bruk av TSD og ELLA» i eHåndbok (http://ehandbok.ous-hf.no/document/133297) for detaljer, vedlegget «TSD_brukerguide_...» har alt du trenger for å komme i gang. 
    
    Brukernavn i ELLA: 
    [user name]
    
    Midlertidig passord: 
    [temp password]
    
    NB: Passordet må endres før første innlogging, klikk «Change password» på innlogginssiden i ELLA.
    

Change your own user group (devs only)

Sometimes you may need to change your own user group to be able to test that the configuration is correct or investigate missing samples, find analysis IDs for deletion etc.

To change user group, you can use the following script:

./ops/<instance>-change-usergroup.sh {username} {usergroup}

Note

Changes using this method are not persisted when updating regular users and should only be used for developers. To change the user group for a regular user, see above.

Tip

The ELL user group has access to all analyses, and will work for most scenarios.

Deactivate user

Removing existing users is not possible, due to historical tracking of changes in the system. A user can, however, be deactivated in ELLA's CLI by running:

ella-cli users deactivate <username>

Other relevant documents