Runtime view

This section describes the concrete behaviour, interactions, and pathways that data take within Seedcase. “Runtime” in this case refers to how the software works “in action”.

Login and authentication

Almost all users will need to log into the Seedcase-managed Data Resources. The steps for logging in and having their permission levels checked follows the sequence described in the figure below.

Figure 1: Login and authentication sequence of a registered user.

Data input

The overall aim of this section is to describe the general path that data takes through a Seedcase Data Resource, from input into the final output. Specifically, these items are described as:

  • Input: Because we currently focus on health research, the type of input data and metadata will be what is typically generated from health studies.
  • Output: The final output object would be the input data stored together as a single database, or at least multiple databases and files explicitly linked in such a way that it conceptually represents a single database.
  • Path: The computational and programmatic steps the input data and metadata takes from being uploaded by a human (and potentially automatically by a program) into Seedcase, passing through quality control checks, adding to the changelog, and storage into the database.

Expected type of input data

Given the (current) focus on health data as well as our own experiences, we make some assumptions about the type of data that will be input into Seedcase. Health data tends to consist of specific types of data:

  • Clinical: This data is typically collected during patient visits to doctors. Depending on the country or adminitrative region, there will likely already be well-established data processing and storage pipelines in place.
  • Register: This type of data is highly dependent on the country or region. Generally, this data is collected for national or regional administrative purposes, such as, recording employment status, income, address, medication purchases, and diagnoses. Like the routine clinical data, the pipelines in place for processing and storage of this data are usually very extensive and established.
  • Biological sample analysis: This type of data is generated from biological samples, like blood, saliva, semen, hair, or urine. Data generated from sample analytic techniques often produce large volumes of data per person. Samples may be generated in larger established laboratories or in smaller research groups, depending on how what analytic technology is used and how new it is. The structure and format of the generated data also tends to be highly variable and depends heavily on the technology used, sometimes requiring specialized software to process and output.
  • Survey or questionnaire: This type of data is often done based on a given study’s aims and research questions. There are hundreds of different questionnaires that can have highly specific purposes and uses for their data. They are also highly variable in the volume of data collected based on the survey, and on the format of the data.

Expected flow of input data

The above described data tends to fit into, mostly, two categories for data input.

  • Routine or continuous collection, where ingested data into Seedcase would occur as soon as the data was collected from one “observational unit”1 or very shortly afterwards. Clinical data as well as survey or questionnaire data may likely fall under this category.
  • Batch collection, where ingested data occurs some time after the data was collected and from multiple observational units. Biological sample data would fall under this category, since laboratories usually run several samples at once and input data after internal quality control checks and machine-specific data processing. While register-based data does get collected continuously, direct access to it is only given on a batch basis, usually once every year. Survey data may also come in batches, depending on the questionnaire and software used for its collection.

For sources of data from routine collection with well-established data input processes, the data input pipeline would likely involve redirecting these data sources from their generation into Seedcase via a direct call to the API so the data continues on to the backend and eventual data storage.

Sources of data that don’t have well-established data input processes, such as from hospitals or medical laboratories, would need to use the Seedcase data batch-input Web Portal. This Portal would only accept data that is in a pre-defined format (as determined and created by the data owner) and would include documentation, and potentially automation scripts, on how to pre-process the data prior to uploading it.

These uploaded files might be a variety of file types, like .csv, .xls, or .txt). Only users with the correct permission levels are allowed to upload data. In either case it will be the administrator who will be doing the initial upload, as that will entail setting up tables and allocating space in the raw data file storage. The second way of getting data into the Data Resource is by manually enter it by an authorized user.

Once the data is submitted through the Portal, it would get sent in an encrypted, legally-compliant format to the server and stored in the way defined by the API and common data model.

Manual data entry: Done in one session

The approved user will open the login screen in the Web Portal. They will enter their credentials which will be transmitted to the API layer. The API Security layer will check with the list of users and permissions in the database and confirm that the specific user has permission to enter data into a specific table (or set of tables) in the database.

Once this check is complete the frontend will receive permission from the API Security layer to display the data entry form. The user completes all fields in the form and clicks “Save and Submit”. This sends the data to the API layer where it is confirmed as valid, parcelled up and submitted to the database. The database will then write the data into a new record in the table (or tables). Once done the database will confirm successful entry of data to the API which will in turn send the confirmation back to the user via the frontend.

Figure 2: Logged in user who manually writes a new row to the Data Resource.

Manual data entry: Done in multiple sessions

There may be situations where an approved user will be prevented from completing the data entry form in one session. In that case it would be beneficial if there is an option of saving the data as it is, and be able to return to the data entry at a later time. Much of the initial workflow is the same as above, until the user is interrupted and selects “Save” instead of “Save and Submit”. This will send the data to the API with a flag showing that fields may be incomplete, thus preventing the API from rejecting the data due to NULL values. The API will submit the data to the database along the incomplete flag.

When the user at a later time goes back to the data entry they will be presented with the option of completing any incomplete records as well as entering new data. If they click on “Complete Records” they are shown the records that they have started but not submitted. Once they select a partially completed record the frontend will request the currently completed items from the database via the API layer before displaying the entry form with the completed fields.

Once the user has completed more data they can either click on “Save” or “Save and Submit”. The first option will put them back to the top of this workflow, the second will send the data back to the API layer for validation. Once the data is validated it will be submitted to the database. The database will then write the data into a new record in the table (or tables) and update the flag to show the record is complete. Once done the database will confirm successful entry of data to the API which will in turn send the confirmation back to the user via the front end.

Figure 3: Logged in user enters data manually in more than one session

Requesting data

There are two main ways to request for data from a Data Resource managed by Seedcase:

  1. Submitting an application to request data, which will create a new data project.
  2. Re-using parts of an existing data project.
Figure 4: Logged in user who is requesting data from a Data Resource.
The “alt” box shows the alternative workflows depending on whether the user requests a new data projects or re-uses an existing data project for the request.

Applying for a new data project

Users who want to request data can browse the metadata, which includes a list of variables and datasets available as well as a data dictionary and changelog, in the Seedcase-managed Data Resource. Similar to an online shopping cart, users who want specific variables in the Data Resource will be able to select them from the list of variables. These selected variables will go into the request application as a new data project, which would be used to automate the data extraction process. In order to add the variables to a new data project application, the user would have to be registered and logged in.

Aside from the selected list of requested variables, the user would need to write some information about the rationale for and intended use of the data as well as other project-specific metadata.

Admin users would approve any new data project applications. When an application is approved, it would trigger automated processes to run the data extraction and subsetting process. It would also bundle it and prepare it for transfer to the applicant user’s secure server or other location where the data will be stored. Admin users can also modify the application form to comply with local regulations and legal requirements, such as removing or adding variables (apart from the variable list), that may be required.

The user can update the data project at any point in the data project’s life-cycle. This includes requesting more variables or modifying the data project description and aim. The admin users would then need to approve these updates.

Re-using an existing data project

Because data projects are stored (and optionally displayed on the Web Portal) as JSON or other machine-readable format, users who want to copy and modify an existing data project can extract and use the details from that project as part of a new data project. The data projects listing on the Web Portal will include a button to either download the JSON metadata or to copy the project into a new data project application.

Browsing projects and results

There are at least three routes, we anticipate the external user might take when browsing the Data Resource:

  1. Reading about the Data Resource, its history, organizational structure and ownership, and any other details in a typical “About” section.
  2. Browsing the completed and ongoing projects that are using or have used the Data Resource, including viewing some of the basic results of the completed projects.
  3. Viewing some basic aggregate statistics of some key variables in the Data Resource. The aggregate statistics would be generated programmatically.

Footnotes

  1. Observational unit is the “entity” that the data was collected from at a given point in time, such as a human participant in a cohort study or a rat in an animal study at a specific time point.↩︎