I think I should create some random new groups as this doesn't really count as "fascinated". I just finished two and half month run to impute (estimate/predict) cataloged variations in OpenSNP individuals (although this was on top of half a year calculation to prepare for stratifying them by ethnicity).
Now, of course, the problem is I really CBA to do anything about visualizing the data - a fact I could blame on either being INTP personality or Aries sun-sign, or just the old cop-out of being lazy and bored. One justifiable reservation is "we" (meaning humans, mainly) don't yet seem to have a good grasp on visualizing genetic data.
So I'm kinda wondering on the off chance someone has ideas or favorites for the visualization. For sure, most people don't actually need or benefit from the classic location-based "vector" representation, also bearing in mind the HUGE expanse of the genome.
I'm trying to think of some more haplotype based approach, with advanced information it could track the inheritance of traits and genes (along with genealogy) up and down along the tree, but it's still a real challenge to make it informative; and perhaps there are even more intuitive ways to represent genetic information that would be obvious even without technical knowledge.
I should put some illustrative images here, but did I mention I'm lazy, and that might poison the intuitive ideas ^^
Faust76 Faust76
70+, M
2 Responses Aug 16, 2014

actually, what i have in mind is quite abit more complex, but is based off of this geometric equation.

Of course it depends a lot on what one is hoping to visualize, the data I'm considering now is relatively simple, linear. Circos based visualizations for bioinformatics have been popular, but then they're mostly useful for non-linear information, like phyogenetics or epigenetics.
With Google et. all working on their full genome sequences they're faced with similar visualization problems, as well as several solutions of course, but it's still supposed to be fun to think about :) When you deal with genome data, especially imputed, much of it will be very identical of course, but at the same time you're most interested in the small differences and discontinuities. But LD plots are even more awful.

what i had in mind was a combination of linear and circos based structuring, wich should provide you with an easy to understand 3D graph that could be used to represent both relatively simple data through more complex data.

*simple and complex data...lol not through complex data(typo)

are you familiar with the vortex mathematics symbol? i think you could probably use the same structuring to create a more dynamic vissual representation.

No, sorry, not familiar with that but I guess now I'll have to check it out :) I was contemplating posting the "How many ancestors do you have" video http://youtu.be/tgXu19LNYEk that's been making rounds here, but the author/channel has pretty interesting video about 3D time volumes. Of course none of that gives a direct recipe, but food for thought on packing enormous vectors into comprehensible spaces.