Open e-t-k opened 8 years ago
We could have a "latest" flag and only show sample groups that are the latest version. While viewing the latest version of a sample group the precious versions could be visible.
On Wed, Sep 7, 2016, 7:53 PM e-t-k notifications@github.com wrote:
Linh comments that it is difficult to keep track of which sample group to use when they are identical except for the version number. eg for most of them, use v2; except for one of them, use v3 (ie, always use the latest version -- but it's frustrating to scan through & figure out which one that is.)
fix? create 'archived' sample groups that are kept for provenance purposes but removed from list of potential groups to run OA with ? I don't know ...
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/UCSC-MedBook/patient-care/issues/94, or mute the thread https://github.com/notifications/unsubscribe-auth/ABRIv1jNY7Yf1I9rGCgiQsflt5MyfKeBks5qnvn8gaJpZM4J3L_X .
Here is the use case, cohort one has 80 samples and a bunch of analysis.
Another 20 samples come in a few months later so a new version of the samples groups are created with new analysis.
People always want to go back and look at old analysis and see why the results have changed. Not much need to run on the older versions. We just need a clear indication on reports of which version was used. 99% of the time you want to run with the latest version. Sometimes you may want to look at the members of the old version but that is about it.
Linh comments that it is difficult to keep track of which sample group to use when they are identical except for the version number. eg for most of them, use v2; except for one of them, use v3 (ie, always use the latest version -- but it's frustrating to scan through & figure out which one that is.)
fix? create 'archived' sample groups that are kept for provenance purposes but removed from list of potential groups to run OA with ? I don't know ...