Archive | 2019

Shameful DNA reporting by PBS

These days too many media outlets seem to go for sensationalism rather than facts, but not usually PBS. I watch their news hour every night and expect it to have thoughtful, in depth reporting, unlike the sound bites from many other news shows.

Thursday night’s piece on the privacy issues from DNA testing was a travesty. This misleading headline on their web site about that segment is not what I have come to expect from PBS – “Genetic genealogy can help solve cold cases. It can also accuse the wrong person.”

No, genetic genealogy does not put the wrong people behind bars. Autosomal DNA is just a very accurate tip that points police to a person or family. In order to make an arrest they next collect the suspect’s DNA from discarded items to see if it is a match to the crime scene DNA. It is those results they take to court, not the genetic genealogy theory.

The wrong person accused scenario that they refer to happened several years ago from using Y DNA, not the full autosomal DNA currently used by genetic genealogists. PBS interviewed Michael Usry about his experience of being suspected of a horrific murder because his Y matched the crime scene and even he suggested that it was not such a bad thing to catch murderers and rapists by using the DNA of their cousins. Click here for an article from 2015 about the Michael Usry case that explains what happened back then.

Note that the Y is only one of 46 chromosomes (in 23 pairs but Y pairs with X) and it is the only one which changes extremely slowly. Therefore can reach back many hundreds of years. For example I have a cousin who has a perfect Y match with a 5th cousin where their common ancestor lived in the 1600s. So clearly, the Y is just not useful for law enforcement searches.

CeCe Moore explaining how autosomal DNA led to a killer – screenshot from PBS News Hour Wednesday Nov 6

Wednesday night’s episode (click here) did a wonderful job of explaining how autosomal DNA is used to solve cold cases. PBS interviewed both famous genetic genealogist Cece Moore and Curtis Rogers, the founder of GEDmatch. I even made my husband watch it so he could better understand what I do when I help people with their DNA results. He enjoyed it.

So I was expecting Thursday’s follow up episode on genetic privacy to explain to me why so many people are worried about this issue. Personally if any of my cousins are violent criminals I am happy to out them, but I get that people did not expect this use when they uploaded to genealogy sites, so their permission is needed.

Instead Thursday’s show was highly inaccurate. So I wrote this blog post… will I cancel my PBS membership? We will see.

 

UPDATE 8-Nov-2019: A number of readers have pointed out the glaring innacuracy of the November 7 episode claiming “that Brian Dripps had been convicted of killing Angie Dodge and is serving a life sentence” when in fact the case has not even gone to trial yet. Although he confessed, he is now pleading not guilty. A recent non-sensationalist summary of that case is here: https://www.ishinews.com/events/unraveling-the-twisted-case-of-angie-dodge/

UPDATE 9-Nov-2019: The PBS NewsHour has corrected the Brian Dripps statement in November 7 episode on their website and also retracted the incorrect information towards the end of their November 8 episode.

Automation to Find the Common Ancestors in the Trees of your DNA Matches

Recently I gave a presentation on many of the great new DNA tools that have come out this year. The talk focused on how both Ancestry and MyHeritage figure out the likely ways in which you are related to a DNA match from the other trees on their sites (click here for my slides). This left very little time to go into the details of my favorite third party tools that can do similar magic, so I promised the attendees a blog post…

The three tools I use the most for finding common ancestors are:

They all have their strengths but none are free. Think of the endless hours professional programmers have spent making these tools and be grateful.

Mainly I use them for unknown parentage cases where the tree is not yet known. However they are also useful for your genealogy. For example, Genetic Affairs will look at the trees on Family Tree DNA and show you any common ancestors it finds with the path to your matches:

My Dad shares DNA with many descendants of this couple as shown by Genetic Affairs from his FtDNA results

There are many other tools I could not do without, like the online relationship calculator at DNApainter (click here) but this article is about automation to find common ancestors. Read on for my summary of the strengths and weaknesses of each of the tools that do that.

Continue reading

Why don’t we share more DNA with a Sibling?

When I first started doing DNA testing and I saw that full siblings shared only about 2800 cM of their DNA, I was surprised. I expected more. Then I thought about it. We have 23 pairs of chromosomes but the testing process cannot separate the two parts of each pair so our matches are seen as if we had only 23 unpaired chromosomes.

The matching segments are listed and totaled as if there was just one side of each pair even though it is using the data from both sides for the comparison; that is why they are called half identical regions or HIRs. Click here for my blog post with a deeper discussion of why doing comparisons that way can create false matching segments.

My brother and I compared at 23andme. The darker segments are the FIRs. Click it for the full image.

Siblings are usually listed as sharing about 50% of their DNA but that is only the half identical part that most companies measure. How many places did they get the same DNA from both parents on both pairs of a chromosome? In other words how many fully identical regions (FIRs) do siblings share? In my experience, they share about 800 cM or so that are fully identical. Thus the total of the FIRs and HIRs is about 3500-3700 cM; the same amount expected of a parent match. That intuitively makes much more sense to me.

So what would be the logical basis for this? Each parent gives each child half of their DNA but not all of it will be the same half. Roughly half of what each parent gives me will be the same as what my brother got so about 2 times 1800 for 3600; yet most full siblings are shown as sharing only about 2600-2800 cM of half identical segments. The rest of that 3600 is found in the fully identical segments.

A recent query I got was “I only share 2553 cMs with my sibling are we full or half? By the online calculator at DNApainter they are full siblings sharing a low amount, but another possibility is the father of one sibling was the brother of the other. That is called ¾ sibings. To tell the difference total up the FIRs and the HIRs. If they total less than 3000 rather than the 3600 for full siblings then they are most likely three quarters siblings or some similar relationship.

23andme totals both the FIRs and the HIRs unlike any other testing company, as shown above, however those totals include the X which throws it off a bit for my calculations.

Wherever you tested, you can get the FIRs and the HIRs by comparing the two kits at GEDmatch.com.

Here’s how. Presuming you have both uploaded to GEDmatch (and hopefully both opted in to helping law enforcement identify violent criminals and victims – click here for my post on that), start with a One-to-One comparison of the two kits.

Continue reading

Talking About Many New DNA Tools

Tuesday I will be presenting the latest version of my talk on solving unknown parentage cases in a virtual conference hosted by the Utah genealogical society (click here for more information). In the past, I relied heavily on the tools at DNAgedcom, but now there are several new tools that are even more exciting.

The basic methodology for unknown parentage searches is to DNA test everywhere. Then look through the trees of your matches to see what ancestors are in common. Build trees down from those common ancestors looking for where the different families meet in a marriage. Then find a child of that marriage who was in the right place at the right time to be the missing parent or grandparent or …

A major difficulty is that many people test DNA without providing a tree. Usually you have to try to build trees for them. Another problem is that building trees down from those common ancestors is incredibly labor intensive when the families are large and the matches are distant. My latest strategy for difficult cases is to recruit several search angels to build the different trees.

There are now tools that automate building trees!
Continue reading

Why you might want to test for DNA health results at MyHeritage

The live streaming from the MyHeritage conference in Amsterdam this past weekend was both informative and enjoyable. I was even able to watch without getting up in the middle of the night as they stayed available afterwards. Many of the talks included information that was not new to me, although I was pleased to finally be able to watch Leah Larkin’s WATO presentation.

Yaniv Erlich

The talk that impressed me the most was the one by Yaniv Erlich in the middle of day 2 about the DNA health results that MyHeritage is now providing (on a reduced price sale just through Tuesday, Sept 10). With the acquisition of Promethease and SNPedia they have greatly expanded their access to medical genetic knowledge. SNPedia is always my go to resource for looking up specific genes, for example click here to see what it says about the BRCA genes. While Promethease is where you can upload your raw DNA from wherever you tested to get health results (sadly written in formal medical talk) based on current but not always solid research (see comments below) and is linked to SNPedia.

One of the important take aways for me was that if MyHeritage finds a bad mutation in your test, they will use a different method, Sanger sequencing, to confirm the result! This almost completely eliminates false positives. Still if you are diagnosed with a really bad variation, my opinion is that you should double check even further by getting a doctor ordered test like the one from Color Genomics.

Yaniv also emphasized that they have worked hard to present the results in an understandable format. There are some examples of this on their blog (click here).

Continue reading