If I've learned one thing getting into the tedium of small segment matches, it's that AncestryDNA has some serious issues with laziness and defaulting to what they presumably view as the "safest" estimate. I noticed this as I was consistently running into my Virginia & Tennessee Native ancestors in segments Ancestry had designated as Native Mexican. Strangely, I never found this in the reverse and I also noted that segments erroneously labelled in my results as Mexican, when passed to my daughter in a slightly smaller segment was accurately labelled. I've noticed this is also the case when it comes to the English & NW European and Germanic designations.

This is when I started running several generations of our tests through the various admixture utilities on Gedmatch, hoping I could gain some understanding of the actual mixture contributions and variations I'd come to understand are simply painted over by whichever group was most dominantly present. I also hoped I could gain some insight into the relatively numerous unassigned segments in granddad's results.

I did a bit of preliminary research to determine which tools to focus on as there's dozens of combinations and most are oriented toward different populations and parts of the world. I settled on 4 I felt were pretty accurate when running kits I had a clearer picture of. One major consideration was finding a good group of utilities that could work with LG's diverse paternal ancestry.

The process, while tedious, was fairly straight-forward. I positioned his maternal and paternal chromosomes with Ancestry's ethnicity signifiers and positioned the same chromosome as analyzed by the admixture tool. I also marked off the unassigned areas for easier comparison.

While of course there can be little certainty when it comes to an ever-evolving science, I was able to recognize patterns in these unassigned segments:



The Unassigned Chromosome19 segment (9,538,807-20,633,104) shows a strong North African component across all calculators. This was most obvious in the Dodecad World9 results where it appears as a distinctive brown African segment. While not a part of our traditional understanding of our family background, it is well in alignment with our family history connecting to Granada and Albacete in Spain - regions with significant Moorish influence. Most of the hyper-mixed areas in gramps' genome appear characteristic of the complex history of interactions in Iberia for centuries.

The chromosome 3 segment (122,394,456-127,426,477) consistently shows Mediterranean ancestry with Middle Eastern elements. It displays a mix of markers which all overlap in the western Mediterranean, particularly Spain & North Africa in Dodecad K12b, while MDLP-22 shows it as early Atlantic Mediterranean with Near East components. Many of us show relatively significant ancestry from the Levant and Anatolia. Being aware that many Sephardi Jews emigrated to Mexico during the Inquisition, especially to remote regions like Sonora, I have suspected the Levantine contribution could come from conversos.

This may not be provable anytime soon however given the methods currently used in developing the reference populations used to estimate ethnic origins of genes/segments. Modern Spaniards must be thoroughly and substantially mixed today, it will not be until methods and techniques get us to the point where we can build our reference panels almost entirely on ancient genomes sequenced in large numbers from people who weren't moving around too much.

Chromosome 6 (24,537,710-43,773,255) isn't particularly interesting when contextualized alongside our other Mexican segments, showing what appears to be complex Native American ancestry with European admixture. It has that distinctive black spike in the MDLP-22 results and contains red Amerindian components in World9. This seems pretty typical and represented of the complex indigenous ancestry coming from colonial Mexico. I went ahead and assigned that one as Mexican without hesitation.

Chromosome 8 had the last mystery segment (1-10,546,816) which consistently shows a West Asian/Mediterranean mix with significant orange West-Asian markers in MDLP-22 and a mix of N. African and SW Asian elements in Dodecad K12b. Again, more Mediterranean/Middle Eastern admixture.

What I think I found most annoying was that in analyzing the different results over this process, I actually ended up with more segments I feel should probably be recategorized, where I had gone in hoping to reduce the ambiguity. I've started doing work with the kits of ancient people and it's genuinely one of the things that gives my life meaning - watching something that once seemed basically impossible: mapping the genome of an ancient person becoming such a common/established thing that I can compare my dna to hundreds of them - my biggest issue now is keeping up on my reference list.

As I was looking through old family photos and imagining how I could best improve and display them, I kept running up against a difficult issue - what do I do with this portrait? In normal circumstances, I would treat it the same as a photograph, ie: remove noise, sharpen, improve color balance. This portrait however, depicts my ggg-grandmother Miquella Sierra-Limon and it suffers from a practice in historical portraiture called whitewashing.

Miquella died in 1908 from eclampsia while giving birth to her 9th child, a son, who also died. It is probable that the portrait was commissioned posthumously, perhaps decades later as suggested by the outfit in which she's depicted - bowties on women became a fashion in the 1920s and '30s but were basically unheard of beyond the realm of menswear prior to that time. The artist may have based their rendering of her image on the features of her adult children who had lighter skin and less prominent Mestizo features.

Although no photographs of Miquella exist that I have yet found, we do have a photograph of her full sister Viviana. Taking in consideration what I know about Miquella's ethnic background through both traditional papertrail genealogy and genetic evidence from segments matched to her and her parents, I made the decision to alter the image in the portrait to better reflect her heritage. In this case, I feel there are less ethical issues around making such alterations since the portrait itself was likely not created in life but rather based on the appearance of living family members and their recollections of her appearance.

My goal was to maintain the original portrait's essence while making it more reflective of a person of Miquella's background in the time in which she lived.

I've been aware throughout this process of the heavy reliance on text and its potential to make the site less interesting for family to browse through so thus far I have largely avoided including photos of our distant ancestors.

I decided I wanted to see if 20+ years of dicking around on Photoshop might have given me enough skill to make some of these photos worth displaying. This is my first effort; a very old, grainy, and color-deteriorated photo of my great-great-grandfather Berchal Sr, presumably in his military uniform.

There are still some changes I could make but I'd like to work through a stack of photos to get a carousel up before I start overthinking and overworking individual photos - I do think it's leagues more clear and visible than it was before and that's all I was going for, especially considering how little experience I have doing digital restorations.

After several months of configuring and re-configuring the maps and tree structure, I finally have everything more-or-less the way I want it so that I can get back to work adding and transcribing sources and even expanding the tree! There is now a functioning big map showing every place where a previously recorded ancestor has been associated. It can be viewed fullscreen or on the page, and markers at each location contain a link routing to a page showing all people in my tree associated with that state (if in the US) or country.

This means that ancestors can be browsed and accessed via the traditional tree, via the map, and soon more will be linked to the DNA matches where they've been identified as the source of that (or those) segment(s)! This is a much slower process, requiring clustering and analysis of numerous single-segment DNA matches to confirm a shared MRCA (most recent common ancestor).

This has taken a long time to pull together, but it's finally at the point I was aiming for - where I can now just enter information into my database, upload documents, and everything will be automatically and dynamically inserted into the proper place within the site. It's very exciting that I can now get back to the actual work and stop worrying about the site display/accessibility!

I've largely finished getting the site working so it's pulling properly from the database, which allows me to focus more on my genome painting project. Working with my grandfather's GEDmatch kit, I've been assigning shared DNA segments to their originating ancestors. I've found the simplest method involves analyzing matches who share a single segment with him. These matches typically share about 15-30 cM and have a most recent common ancestor (MRCA) from the 18th or early 19th century.

This research led to a serendipitous discovery while sorting through DNA matches on Ancestry - I recognized someone I knew, my high school Algebra teacher. This was the first time I'd ever recognized a DNA match, and coincidentally, he was also the only teacher I regularly saw outside of school. My friends and I used to show up at his house on random nights when we were in his neighborhood and wake him up. He would often hang out with us while we chain-smoked cigarettes on his porch. I even adopted a kitten from him once, and by another stroke of luck, I had recently seen him on Facebook's "People You May Know."

My teacher's match data showed that he and my grandfather share a single segment on chromosome 9, where my grandfather had no other matches. I messaged him about this coincidence and asked if he had any ideas about our shared lineage. During our conversation about his family history, he mentioned that he never knew his father and only had his name and estimated birth year. I asked him to upload his test to GEDmatch for a more detailed segment comparison. After he shared his kit number, some genealogical sleuthing revealed our shared ancestor was only one generation removed from where my previous research had ended.

The segment traced back to Isaiah VanZandt, a member of the old New York Dutch family best known as progenitors of the wealthy, ranching VanZandts of Texas (after whom a Texas county is named). Notably, this is the same family that produced the legendary folk singer Townes VanZandt, who ranks second only to Bob Dylan among my favorite artists in the genre.

Using GEDmatch's segment search, I found other matches sharing portions of this segment with both of them, allowing me to construct a cluster on this otherwise sparse chromosome on his paternal side. The fact that my teacher is a half-cousin to my grandfather makes this match particularly valuable for genetic genealogy. Usually, at this genetic distance where you share a single segment, matches trace back to an ancestral couple, making it impossible to determine which individual contributed the segment. However, with half-cousins, you can trace back to a specific individual. In this case, my grandfather descends from one of Isaiah's wives, while my teacher descends from the man's other wife, Margaret Thompson.

We were both surprised to discover such a large segment of DNA remained intact, inherited from a man born in the colonial period - more than 40 years before the American Revolution. Neat.