Reproducible research is a fundamental responsibility of scientists, but the best practices for achieving it are not established in computational biology. The Synapse “Provenance” system is one of many solutions you can use to make your work reproducible by you and others.
Provenance is a concept describing the origin of something; in Synapse it is used to describe the connections between workflow steps that derive a particular file of results. Data analysis often involves multiple steps to go from a raw data file to a finished analysis. Synapse’s Provenance Tools allow users to keep track of each step involved in an analysis, and share those steps with other users.
The model Synapse uses for provenance is based on the W3C provenance spec where items are derived from an activity which has components that were used and components that were executed. Think of the used items as input files and executed items as software or code. Both used and executed can either be items in Synapse or URLs such as a link to a github commit or a link to specific version of a software tool.
The Synapse clients for command line, Python, and R support creating and editing of provenance relationships. The Web client allows editing of provenance once the file has been uploaded.
On the right is a Synapse visualization of provenance relationships that is demonstrated in the following section using our programmatic and web clients. In this example, we have two scripts, one that generates random numbers and another that takes a list of numbers and computes their squares. The project’s workflow looks like the image to the right.
Let’s begin with a script that generates a list of normally distributed random numbers and saves the output to a file. For example, you have an R script file called generate_random_data.R and you’ve saved the output to a data file called random_numbers.txt. We’ll begin by uploading the files to Synapse and then set their provenance.
For this example, we’ll use a
Project that already exists (Wondrous Research Example : syn1901847). The code file is already saved in Synapse with synId
syn7205215 so we’ll upload the data file to this
Project, or in Synapse terminology, the project will be the parent of the new entities.
As the random_numbers.txt file was generated from the above script, we are going to specify this using provenance.
There are a couple ways to set provenance information for a Synapse entity. The
executed arguments specify resources used and code executed in the process of creating the entity. Code can be stored in Synapse(as we did in the previous step) or, better yet, linked by URL to a source code versioning system like GitHub or SVN. As an example, we’ll specify 2 somewhat contrived sources of provenance:
Once the data file is uploaded, it will provide the synId assigned to it. In this case, the data file’s synId is
To continue our example above, we’ll now add some new results from our initial data file. We’re going to take the results in random_numbers.txt and square them. The script to square the numbers will be square.R and we’ll save the output to a data file, squares.txt. As with the previous example, the code file is already saved in Synapse, so we’ll upload the data file and set its provenance.
To view the provenance relationships you’ve created:
Activity is a Synapse object that helps keep track of what objects were ‘used’ in an analysis step … as well as what objects were generated. Thus, all relationships between Synapse objects and an
Activity are governed by dependencies. That is, an
Activity needs to know what it ‘used’ – and outputs need to know what
Activity they were ‘generatedBy’. A couple of points for clarity:
Activitycan ‘use’ many things (i.e. many inputs to an analysis)
If an activity isn’t assigned to an entity and then stored, a separate graph will be created for each file that the activity generated. The following example is used to assign the same activity to multiple files resulting in one provenance graph:
|command line docs|
Try posting a question to our Forum.
Let us know what was unclear or what has not been covered. Reader feedback is key to making the documentation better, so please let us know or open an issue in our Github repository (Sage-Bionetworks/synapseDocs).