Group genes together in an LSDB within LOVD3

[ifokkema]

It has been requested several times to be able to group gene databases together under a group or LSDB, within one LOVD3 instance. This is particularly useful for LOVD2 databases that used to have their own branding and disease-specific setup, and are now merged into the shared LOVD3 instance. Such a group should allow for, at least, a group-specific gene switcher. Other requirements are under debate.

• [ ] Call this groups or LSDBs?
• [ ] Groups should be configurable by managers and up.
• [ ] Groups should have some settings or meta data, but it hasn't been determined yet what;
  • [ ] Abbreviation.
  • [ ] Title.
  • [ ] Logo?
  • [ ] Header text or such?
• [ ] Genes part of a group should have their gene homepage updated; the group name and meta data should show up, and the gene switcher (or an additional one?) should be limited to the other genes in the same group.
  • [ ] Should this change be only when a group has been selected, or always? Genes exist that are related to >1 disease and may not always need/want branding (e.g. BRCA1 and MBTPS2).
• [ ] Should genes be able to join multiple groups? (see above) If so, how would we know which group's settings to apply to the gene homepage? Would groups need to be selected first? Perhaps only show branding when selected, or make it a gene setting?
• [ ] URL for showing list of groups: /groups.
• [ ] Groups should have their own URLs, so they are linkable. E.g. /groups/ABBREVIATION.
  • [ ] Perhaps even with an LOVD.nl/shared based link.

[leicray]

Call this groups or LSDBs?

In think that "groups" is better because it is a more neutral, generic, term.

Groups should be configurable by managers and up.

It seems reasonable that configuration privilege should be limited. However, the LOVD3 manual does not list the tasks that a manager may undertake. A simple table (somewhere near the start of the manual) that summarises each of the user types, and their associated privileges, would be a useful addition to the manual.

Genes exist that are related to >1 disease and may not always need/want branding (e.g. BRCA1 and MBTPS2).

I fully understand the issue here and it's tricky. Let's say that gene XYZ123 is shown to harbour a very few sequence variants that cause osteogenesis imperfecta, but pathogenic variants in the gene most commonly cause, say, excessively hairy ears. On the other hand, it is well established that COL1A1 and COL1A2 harbour variants that cause both osteogenesis imperfecta and Ehlers Danlos syndrome and joint branding would be appropriate. The key question is whether there is a threshold for determining when particular branding might be applied to a gene. That sounds like a philosophical issue and it's probably above my pay grade.

Should genes be able to join multiple groups? (see above) If so, how would we know which group's settings to apply to the gene homepage? Would groups need to be selected first? Perhaps only show branding when selected, or make it a gene setting?

The simple answer is yes (see above). The other issues are probably best resolved by open discussion with group "owners".