I apologize for jumping the gun on moving the DNA Gramplet documentation I did not realize the documentation was referring to an unreleased addon not available to anyone but yourself, which explains my confusion when I was attempting to update based on the existing version and seeing different results than your screenshots show :( Will you be sharing it for testing myself & possibly User:Matt are interested? --Gioto (talk) 22:21, 20 September 2020 (UTC)
- Added access to test version at DNA. I will be pushing this once another PR has been accepted (Addon:SyncAssociationis loosely connected to the DNA gramplet) GaryGriffin (talk) 17:25, 20 Sep 2020 (PST)
dna feedback from matt
GaryGriffin: Thank you for sharing. Testing seems to work fine. Some notes and ideas:
- The loading speed might become a problem. For 8 associations it already takes ~3 seconds to load.
- Overlapping segments are hidden below others and currently unaccessible. A way to draw overlapping segments in seperate rows might be an interesting feature in future.
- Matt: Hmmm, I have 18 Associations for one person and it is less than a second to draw for me. I am on a 2013 Mac, so not the fastest system. This may be more of a function of PyCairo implementation on the platform. Can you provide more info on platform and range of overlaps.
- Are most of your segments tied to a maternal/paternal side or unknown (which can definitely get busy)? Almost all of mine are M/P known. The drawing format is the standard one (for each segment an M and a P section). I fear I would add more confusion if I diverted from that. See DNApainter as an example.
- I draw in order of Association. I cannot think of how I would draw in an order that guaranteed the largest was always behind all smaller overlaps. I have only 1 totally obscured overlap, so hadn't thought about this problem. I only realized I had this when I made the segments more transparent (see next item).
- Maybe if all were less opaque, it would be an easier view? Change line 330 of dna.py to the following. This will change the opacity to 0.5 for M/P and 0.33 for U. The one I posted was 1 for M/P and 0.5 for U. Would appreciate your feedback on this change.
cr.set_source_rgba(rgb_color, rgb_color, rgb_color, 1/(chr_mult+1))
- talk) 17:40, 25 Sep 2020 (PST)
- GaryGriffin: I'm on Windows 10 using GrampsAIO64-5.1.3-1 and Pycairo 1.16.1. All segments are tied to a maternal/paternal side since I only added the known ones. I do have some segemnts overlapping since many of them are from the same common ancestor couple. Yes drawing one maternal and paternal row for each chromosome is standard, but the DNA Painter also has an option to expand and collapse these rows, so you can see all overlapping segments. (see: https://dnapainter.com/example, click on the chromosome number or the plus/minus icon)
- Matt (talk) 08:09, 26 September 2020 (UTC)
- Matt: So you don't see value in changing the transparency, right? Then I will leave as is and make a note in the docs. I was not aware of the expansion in DNApainter. Thanks for pointing that out. Let me ponder implementing that. I am concerned that it will become very busy. Especially if I do all open or all small (using an overall PLUS rather than a per chromosome PLUS). What are your thoughts on individual plus vs just an overall plus?
- talk) 17:11, 29 Sep 2020 (PST)
- GaryGriffin: No, I don't think that transparency is a solution. Unfortunately I don't know what would be the best way to implement it. I'm afraid you can only try it that way and then try other solutions if it isn't working well. Maybe it would be more efficent to try to create it as a view or report (which calulates it once for two selected persons) instead of a gramplet (which calculates it every time you change a person or switch to another active person)?
- Matt (talk) 14:47, 2 October 2020 (UTC)
- Matt: If you wanted just the view of one other person, you can run SyncAssociation to copy the Note from the first person to the associated person (if it is not already there) and then switch to the associated person and run the DNA gramplet. That would (probably) only have one pair of people in the view. I assumed that the value of this gramplet was when there were many associated people to look at the big picture. I thought that there would be only a small number (often just one) of people with very long Associated lists. Looking at the reciprocal view would be a way to reveal overlaps. But I didnt think that would actually be interesting normally.
- talk) 09:42, 04 Oct 2020 (PST)