gn-uploader - gn-uploader: Data uploader for GeneNetwork

Project Goals

The project seeks to handle the checking of data files for correct syntax and other errors before allowing the code to be uploaded.

The files are "tab-separated" values (TSV) files, and must conform to the following criteria:

Line-Level Checks

Must be tab-separated

Cell-Level Checks

No empty data cells
no data cells with spurious characters like `eeeee`, `5.555iloveguix`, etc.
decimal numbers must conform to the following criteria:
- when checking an average file decimal numbers must contain exactly three places to the right side of the dot.
- when checking a standard error file decimal numbers must contain six or greater places to the right side of the dot.
- there must be a number to the left side of the dot (e.g. 0.55555 is allowed but .55555 is not).
check line endings to make sure they are Unix and not DOS
check strain headers against a source of truth (see strains.csv)

Guix Channel

This repository can be used as a guix channel.

(channel
 (name 'gn-uploader)
 (url "https://git.genenetwork.org/gn-uploader")
 (branch "main")
 (introduction
  (make-channel-introduction
   "93c77842315d304abbfc27d78d98b7d42da32a61"
   (openpgp-fingerprint
    "F370 F409 854B 90E3 52F3  AB01 362B 0BB8 B81D 5A42"))))

Configuration

This system requires that some configurations be in place to operate correctly. The following sections detail the configuration variables.

System Configurations

Config: `LOG_LEVEL`

Default: WARNING

We use logging to track the operation of the system, noting major events, and conditions in the system as it runs. The LOG_LEVEL variable controls how much of that information we want to track, and the severity of said information.

Config: `SECRET_KEY`

Mandatory

This is a key used for signing cookies for more secure tamper-proof sessions. Without it, Flask will generate a random key for each running thread/process of the system, leading to some really difficult to debug consequences.

Avoid the aggravation! Just set a secret key and be done with it.

For some more information on Flask's SECRET_KEY have a look at the following resources:

Config: `UPLOAD_FOLDER`

Default: /tmp/qc_app_files

This is the directory where all uploaded files are stored.

Files are to be stored here temporary, while they are still being processed. Down the line, we will probably have a different location for files that have passed QC and whose data has been accepted into the system.

Development Note: Other files we store here that we will need to extract to a separate configuration are:

Files with outputs from background processes triggered by this system
list others here…

Config: `SQL_URI`

Mandatory

This defines the connection URI to the database.

This is a MANDATORY configuration variable: without it, the uploader will not work as expected, if at all.

Presently, we use the MariaDB database, and thus, the URI to the database will take the form

mysql://<username>:<password>@<hostname-or-ip>:<port>/<database-name>

The parts of the URI of note are:

<username>: This is the username to use to connect to the MariaDB server
<password>: A password to authenticate the user with the MariaDB server
<hostname-or-ip>: The domain name or IP address of the computer hosting the MariaDB server
<port>: The port to connect to for the MariaDB server, on the computer hosting MariaDB
<database-name>: The database to connect to

Redis Configurations

Config: `REDIS_URL`

Default: redis://

This is the connection URI to the Redis server. The default connection parameter above tries to connect to a redis server on the same host that the uploader is running on.

Note to Fred: We should probably rename this REDIS_URI for consistency.

The Redis connection URI has the general form:

redis://<username>:<password>@<hostname-or-ip>:<port>/<dbnum>

and the parts of the URI of note are:

<username>: This is the username to use to connect to the Redis server
<password>: A password to authenticate the user with the Redis server
<hostname-or-ip>: The domain name or IP address of the computer hosting the Redis server
<port>: The port to connect to for the Redis server, on the computer hosting Redis
<dbnum>: A number identifying the database to connect to

Here is [a resource](https://github.com/lettuce-io/lettuce-core/wiki/Redis-URI-and-connection-details) on other ways to construct Redis connection URIs.

Config: `JOBS_TTL_SECONDS`

Default: 1209600 seconds (14 days)

This is the "Time-To-Live" for the keys in redis. Most of the keys we store in Redis from this system do not need to be retained forever. The TTL determines how long (in seconds) to keep the keys around for.

Config: `GNQC_REDIS_PREFIX`

Default: GNQC

This system could find itself sharing a Redis server with other (related) applications such as [GN2](https://github.com/genenetwork/genenetwork2) and [GN3](https://github.com/genenetwork/genenetwork3).

The GNQC_REDIS_PREFIX configuration is prepended to the keys being saved by this system to redis. For example, if GNQC_REDIS_PREFIX=GNQC and you want to set up a hash with the key users, the effective key in redis will be GNQC:users.

Development

For reproducibility, this project is developed using guix.

To launch a guix shell for development, do

guix shell --container --network --share=/some/host/directory=/the/upload/directory --development --file=guix.scm

which environment that is isolated from the rest of your system.

We share a host directory with the container (that is writeable by the user that started the web application) to serve as the upload directory for the application.

Run the CLI version

Run the CLI version of the application, with

python3 -m scripts.qc --help

Run the web version

You need to have a running Redis instance, and configure the application to connect to it.

To run the web-version of the qc app in development mode, you need to set up a few environment variables

export FLASK_APP=wsgi.py
export FLASK_ENV=development
export QCAPP_INSTANCE_PATH=/path/to/directory/with/config.py

then you can run the application with

flask run

Checks

To run the linter over the code base, run:

pylint setup.py tests quality_control qc_app r_qtl scripts

To check for correct type usage in the application, run:

mypy --show-error-codes .

Run unit tests with:

$ export QCAPP_CONF=</path/to/configuration/file.py>
$ pytest -m unit_test

To run ALL tests (not just unit tests):

$ export QCAPP_CONF=</path/to/configuration/file.py>
$ pytest

Deploying/Installing QC

NOTE: All tests should pass before you deploy the uploader!

CLI: Docker

Generate the docker image file with

guix pack -f docker -S /bin=bin genenetwork-qc

That creates the image file with a path such as:

/gnu/store/ibz5qlwzv0lyply2by7422z0c6jfaa6s-genenetwork-qc-docker-pack.tar.gz

You can now load this file into docker withe

docker load < /gnu/store/ibz5qlwzv0lyply2by7422z0c6jfaa6s-genenetwork-qc-docker-pack.tar.gz

and from there, you can run the application as detailed in the Running QC: CLI-Version section below

CLI: Guix

The application can be installed using guix by pointing to the guix.scm file as follows:

guix package [-p /path/to/qc/profile] -f guix.scm

Web-Version

TODO Document deployment details for the web version of GeneNetwork QC better

Running QC

Command-Line Version

Install the application as shown in the Installing QC section above.

To run qc against a file, the syntax is:

qc [--strainsfile <strainsfile-path>] [--verbose] <filetype> <filepath>

where

<filetype> is one of "average" or "standard-error"
<filepath> is either an absolute path to the file, or a path relative to the current working directory
if the --strainsfile option is not provided, it will default to the one in the root directory of this repository
the --verbose option is a flag, defaulting to False that controls the display of optional progress messages

To view the usage information for the application, run

qc --help

Web Version

TODO Document usage of the web-UI version of the application

Docker

Download the docker image file from the releases page of the application and load it to docker with something like:

docker load < genenetwork-qc-0.0.1-1-oxu472i-docker.tar.gz

replacing genenetwork-qc-0.0.1-1-oxu472i.tar.gz with the actual name of the release you downloaded

Run the application with something like:

docker run -v /path/to/qnqc_py/tests/test_data:/data -ti \
    genenetwork-qc:latest /bin/qc average /data/average_error_at_end_200MB.tsv

replacing /path/to/qnqc_py/tests/test_data with the path to the folder where the file you want to check is in, and average_error_at_end_200MB.tsv with the name of the file you want to check for errors.