Welcome @alexander.indra
Yes, the datasets are abstracted with persistent unique identifiers. This is a bit different from how custom file naming works on the command line. And while you can custom name files in Galaxy (in your workflow), this isn’t really specific enough. Instead, we have a few other methods you can try.
The most direct is to group input data into a collection folder, and add in a name tag to the datasets. You can run all samples together through all steps. Some protocols can make use of group tags for different conditions too.
This video shows a simple example where the # hashtag is added directly.
Adding tags one by one is tedious, so consider doing this in a batch.
Tag elements from a tabular file of tags (link at EU)
Finally, there are methods to tag and track data within workflows. As a example, the original sample #sample propagating tag could be retained and a processing step tool non-propagating tag could be added, so when a dataset has both, the contents of the data are very clear. And the API supports using Galaxy with file tracking at a detailed level that doesn’t involve the web application.
With that context, would this work for your process? It sounds like you are running an assembly job on some samples, then downloading the history into an archive, and wanting to know which files belong to which sample and which step.
The json file in the RO-Crate history archive can be parsed to map from tag(s) to the dataset name.
In the next release, we’ll be adding in a pre-parsed mapping file that will do some of this for you. I put some examples about how to parse the json in here when creating the feature request, maybe helps? Please ask of you have questions.
I’m glad you ask and follow up questions are welcome. Tracking samples is a really important function and we definitely have methods to do this. Which method to use depends on the overall experimental design. If you want to share more about your process I can try to help more. 