nda-tools

To submit data to the National Institute of Mental Health Data Archives (NDA), users must validate their data to ensure it complies with the required format. This is done by using the NDA Validation and Upload tool. Additionally, users can package and download data from NDA as well. If the associated data is downloaded from S3, temporary federated AWS tokens are required. A Python package and command line clients have been developed to allow users to programmatically validate, package, submit, and/or download data. Validation, Submission Package, and Data Submission web services.

Getting Started

Installing Python

The user will need a Python distribution to use the client. Run the following from a terminal/command prompt to determine if Python is already installed:

python3 --version

Notes:

• If Python has already been installed, users should see version information. If not, you will need to download and install it from Python.org.
• The user may need administrative rights, root, or sudo privileges to install a Python distribution.
• Python may be installed but not available on the system path. Please consult Python installation and usage documentation: Python3

Installing pip

Since Python 3.4, pip is included by default with the Python binary. You can check the version with:

pip3 --version

If pip is installed, then you should see version information. If not, you should install pip. First, download it from https://bootstrap.pypa.io/get-pip.py, then run the following to install for your user.

python3 get-pip.py --user

Notes:

• Pip may be installed but not available on the system path. Please consult Python installation and usage documentation.

Installing the client

These instructions will help you get setup to run the client.

Simply enter the following command into your terminal or command prompt to install nda-tools:

pip install nda-tools

This will automatically install the nda-tools package, including the command line scripts and required packages.

Notes:

• If the nda-tools needs special permission try:
  • pip install nda-tools --user
• If multiple versions of python or pip exists on the operation machine, the command prompt will not recognize the nda-tools script. Try the following command instead:
  • python -m NDATools.clientscripts.[NDAtoolcommand]
• If a deprecated version of the tool is already installed, it'll prompt the user to upgrade. To update, follow the prompt command.

Credentials

While not needed solely for validation, if you want to create a package and submit your data to the NDA, you must have an active account with us. This can be requested from the NDA website. You can read more about what is needed for contributing data into the NDA here.

Keyring

Keyring is a Python package that leverages the operating system's credential manager to securely store and retrieve user credentials. To improve security on nda-tools, password storage in the settings.cfg and the password flag have been replaced with keyring. Furthermore, the keyring implementation supports long-running workflows.

Updating Stored Passwords with keyring

All Operating Systems

For users of any operating system, the password can be updated with:

keyring.set_password('nda-tools', USERNAME, NEW_PASSWORD)

Mac / Windows

Mac and Windows users may use Keychain and Credentials Manager, respectively, to update their passwords.

To update your password with keyring, run:

• keyring.set_password('nda-tools', 'YOUR_USERNAME', 'NEW_PASSWORD'),

replacing YOUR_USERNAME and NEW_PASSWORD with your NDA username and new password. You can read more from the Keyring Documentation.

If you do not have any entries stored via keyring, you will be prompted to enter the password. If authentication is successful, nda-tools will store your password via keyring. Subsequent usage of nda-tools will retrieve the password automatically and securely from keyring.

Linux

Linux users may need to install a backend implementation of keyring since they may not have a native credentials manager such as those included with the Mac and Windows operating systems. If the keyring backend is missing, nda-tools will print the following message:

If there is no backend set up for keyring, you may try pip install secretstorage --upgrade keyrings.alt

For Ubuntu users,

apt-get install -y gnome-keyring

You are Now Ready to Run the Client.

Please note that if you encounter SSL errors when running the client, you may need to re-run pip installation of requests, with pip install pip install requests[secure] which will install some additional packages with more support for SSL connections.

Using the Client

To view options available for the Validation Tool Python client, enter the following command:

vtcmd -h

or to view options available for the Download Python client, enter:

downloadcmd -h

Configuring the Client

• If your command-line inputs have special characters (i.e., passwords) or spaces (i.e., in directory/filenames), you may need to enclose them in quotations.
  • If you are using windows, use double-quotes: " "
  • If you are using Mac OSX or Linux, use single-quotes: ' '
• Upon your first run, the client will prompt you to enter your username and password, which it will store in your operating system's credential manager. You may go back and edit your credentials at any time.

The ~.NDATools\settings.cfg file provided with the client contains configurable options for Endpoints, Files, and User information.

Typically, you won't need to change entries in the 'Endpoints' section; however, you might wish to modify the 'Files' and 'User' sections with preferred locations for validation results, user login, and AWS credentials information.

• While arguments are not positional, the first argument should be the list of files to validate.

  • The list of files has no command-line switch so it can get interpreted as part of a preceding argument.
  • For example, there is no way to differentiate whether the csv file is part of the -l argument or a second argument:

   vtcmd -l "Users/[youruser]/Documents/MultipleDataTypes" \
   "Users/[youruser]/Documents/MultipleDataTypes/Stage_Testing_BigFiles_genomics_sample03.csv"
  

Files for Validation

It is required that you know the full path to the csv files that are going to be validated. Furthermore, if your data includes manifests and/or associated files (i.e., genomics files, imaging files, etc.), you must also know the full path to these files, which should be entered as an optional command-line argument. Otherwise, the client will prompt you to enter a list of directories where any additional files are stored. You can also list a bucket, optional prefix, and your AWS credentials if the associated files are in AWS.

Please Note: When listing the directory for associated files, include the folder up to but not including the file name listed in the csv file.

Example:

If the associated file name is in Users/[youruser]/Documents/MultipleDataTypes/data/1G_file.fastq and is listed in your csv file as:

data/1G_file.fastq

then the directory you will enter is:

Users/[youruser]/Documents/MultipleDataTypes

You should not include the 'data/' folder as part of the directory name.

• Check all files properties and make sure the user has all Permissions allowed.

To start validation, you must enter a list of files (or a file path if not in the current directory), separated by a space:

vtcmd MultipleDataTypes/genomics_sample03.csv testdata/with_associated_files/genomics_sample03.csv 

If your data includes manifest files, you must enter the directories where the manifest files are located, separated by a space:

vtcmd submission_data/sample_imagingcollection01.csv  -m submission_data/Manifests

If there are associated files, enter the directories where they are found, separated by a space:

vtcmd MultipleDataTypes/genomics_sample03.csv testdata/with_associated_files/genomics_sample03.csv -l MultipleDataTypes testdata/with_associated_files 

If the files are located somewhere other than the current working directory, then you must enter the full path to the files:

vtcmd MultipleDataTypes/genomics_sample03.csv testdata/with_associated_files/genomics_sample03.csv -l Users/[youruser]/Downloads/SubmissionData testdata/with_associated_files

If your associated files are in S3, then you must include the bucket name, access key, and secret key.

• The access and secret key can be stored in the settings.cfg file as well.

vtcmd MultipleDataTypes/genomics_sample03.csv testdata/with_associated_files/genomics_sample03.csv -s3 my_bucket -ak XXXXXXXXXXXXXX -sk XXXXXXXXXXXXXX

Note: You can also upload associated files saved locally and in s3. Just make sure to include the directory where the local files are saved (-l path/to/local/associated/files)

To create a package, enter "-b" at the end of your command line argument. You can also enter your username, AWS credentials, Collection ID or alternate endpoint title, and the title and description of your submission, or you can enter this information later when prompted by the client. The client will not begin building the submission package until:

• All your files are validated
• All associated files have been located on your local drive or in S3

Once package submission and upload are complete, you will receive an email in your inbox from NDA confirming your submission was successful. A local version of the package will be saved automatically to ~\nda-tools\vtcmd\submission_package\ folder and can be found on the collection submission tab on the NDA site.

Fixing QA Errors

A QA check is performed on all data after it has been submitted to NDA for inconsistencies in data-points including sex, subjeckey, interview age and interview date. If any problems are found with the data, an email will be sent to the users who created the submission along with a UUID called a QA Token which can be used to fix the errors in the submission.

To fix the data in NDA for your submission, you need to replace all of the csv files which contained errors in your original submission. To do this you must:

1. Retrieve the csv files with that were used to create the original submission and which contain data that needs to be corrected. This includes all csv files where data needs to be added, removed or updated.
2. Correct the files by adding, removing or updating information as needed.
3. Run the vtcmd with the -qa command line argument. Specify the value of the QA token which you should have received via email with the -qa argument. Then list all of the csv files that you made corrections to. If there was a csv file from the original submission that did not contain any changes, it is not necessary to supply the file as an argument at this time.
   

For example, if the original submission consisted of file1.csv, file2.csv and file3.csv, and corrections needed to be made to file1.csv and file2.csv, the command to fix qa errors will look like:
vtcmd -b -qa f0d8ff08-cc38-4cb3-b6a4-39aff6f07f0e corrected-file1.csv corrected-file2.csv

Notice that file3.csv is excluded from the command because no changes needed to be made to that particular file.

Please note this command should be run once for a submission and should include all of the files that contain corrections to data. i.e do not run the vtcmd once for corrected-file1.csv and another time for corrected-file2.csv. If you accidentally omit files containing necessary changes when running the command, please contact the HelpDesk at [email protected].

Also note that the csv files should contain all of the data that was submitted originally. i.e. if a csv originally had 800 rows and only 3 rows needed to be changed, all 800 rows should be present in the csv when running the vtcmd, not just the 3 rows that contain changes. Any data that is left out of the csv will be reflected in data-expected numbers for the collection.

The script will not upload any associated files that were uploaded during the original submission. It will only be necessary to upload associated files if they appear in corrected csv files but not in any of the csv files from the original submission. This saves time during genomic and imaging submissions where associated files can take days to upload.

Downloading Data

To download data, you should use the downloadcmd command. This provides several options to download your NDA packaged data or a subset of the data. All files are downloaded automatically to the ~\nda-tools\downloadcmd\packages\ folder, but you can change this by indicating a new directory in the command line to save files. Please note: the maximum transfer limit of data is 20TB per month.

• Users can contact the NDA Help Desk at [email protected] and ask for their download threshold to [temporarily] be extended.

All Package Data

All packaged data can be downloaded by passing the package ID:

downloadcmd -dp <packageID>

Note: it will NOT download associated files unless you created your NDA package with associated files. Steps to download associated files are below.

Downloading .txt Files

The downloadcmd command has two options for downloading data inside .txt files. If you downloaded your NDA package, you will find meta-data .txt files, many of which represent data measures. Genomics, imaging, and other associated data will be listed in these .txt files as s3 links. If you would like to download all the s3 links in your .txt file, you can indicate so by passing the -ds flag.

downloadcmd -dp <packageID> -ds path/to/data/structure/file/image03.txt

The downloadcmd command has two options for downloading data inside .txt files. If you downloaded your NDA package, you will find meta-data .txt files, many of which represent data measures. Genomics, imaging, and other associated data will be listed in these .txt files as s3 links. If you would like to download all the s3 links in your .txt file, you can do so by passing the -ds flag.

downloadcmd -dp <packageID> -t path/to/all/s3/txt/file/alls3.txt

Downloading Files to S3 Bucket

The downloadcmd command can download your NDA package into your S3 bucket directly.

downloadcmd -dp <packageID> -s3 <s3 bucket>

This is the preferred way to download data from NDA for two reasons:

1. Downloading to another S3 bucket is considerably faster because the data doesn't leave AWS.

2. It allows us to download an unlimited amount of data from NDA to your bucket directly.

For S3-to-S3 copy operations to be successful, the S3 bucket supplied as the program argument must be configured to allow PUT object operations for arn:aws:sts::618523879050:federated-user/<username>, where <username> is your NDA username.

For non-public buckets, this will require an update to the bucket policy. The following statement should be added to allow the necessary permissions after replacing <your-s3-bucket> with the bucket name:

{
    "Sid": "AllowNDAUpload",
    "Effect": "Allow",
    "Principal": {
        "AWS": "arn:aws:iam::618523879050:federated-user/<username>"        
    },
    "Action": "s3:PutObject*",
    "Resource": "arn:aws:s3:::<your-s3-bucket>/*"
}

You may need to email your company/institution IT department to have this added for you.

Note: If your S3 bucket is encrypted with a customer-managed KMS key, then you will also need to update the policy of the key that is used to encrypt the bucket.

The following statement should be added to your key's policy:

{
    "Sid": "EnableUseForFederatedNDA",
    "Effect": "Allow",
    "Principal": {
        "AWS":  "arn:aws:iam::618523879050:user/DownloadManager"
    },
    "Action": ["kms:GenerateDataKey","kms:Decrypt"],
    "Resource": "*"
}

Further Assistance

If you have any problems with this Validation Tool Python client or would like to provide feedback/comments, please email us at [email protected] .

nda-tools