from michael in slack re: app #51
Description
Michael
10:43 AM
I am working through some website analysis use cases and it has become clear to me that, in addition to the currently “active” selected set of samples, users must be able to name the set of samples and save the list of samples associated with the for the duration of the session or download it (the list of samples, not the data) for future use, or upload a named list of samples that has been previously downloaded. The session may have any number of named sample sets. Operations like “intersect” will take two sample sets. The “active set” will be just like the other named sets except it will have a default name that doesn’t need to be explicitly assigned. We may want to provide named sets that we think people may be interested in such as “Harbison 2004 ChIP-ChiP” which would be a default name for all samples in that dataset. Users could apply subsetting operations to that set, like any other named set, for example on growth media.
Michael
10:46 AM
I realize this is a front-end functionality, but I think it’s one that’s worth keeping in mind as you work on the data store. I also wonder, @chase, if there might be some value in beginning to plan the front end now, since you have people to work on it. This is essentially a required functionality, not a specific page layout or sequence of pages. However, it does imply that there should be a way to view a list of named samples that exist in the current session and to perform arbitrary unions and intersections on them. This is probably it’s own page that could perhaps be built sooner than designing specific graphical outputs and so on, since it is basically a tabular output.
I realize this is a front-end functionality, but I think it’s one that’s worth keeping in mind as you work on the data store. I also wonder, @chase, if there might be some value in beginning to plan the front end now, since you have people to work on it. This is essentially a required functionality, not a specific page layout or sequence of pages. However, it does imply that there should be a way to view a list of named samples that exist in the current session and to perform arbitrary unions and intersections on them. This is probably it’s own page that could perhaps be built sooner than designing specific graphical outputs and so on, since it is basically a tabular output.
Michael
5 minutes ago
I like the idea of thinking of “dataset” as just a default, always-available name for a set of binding or PR samples. Users can assign names to arbitrary sets of binding or PR samples but these names are always available. I think we may want to have sets of binding samples and sets of PR samples be different “types” meaning they will be listed separately in the list of named samples and treated separately in the analysis code. So on this view “dataset” is a feature but it doesn’t have to be a privileged feature. It is privileged only in that it is a default sample set with a default name.
10:57
Also, users should be able to go to a data download page where all the named sample sets are displayed and they can choose one or more to download the actual data values from. Possibly one file for each named dataset that is selected for download.