News for hackers. Stuff that matters.
Re:Seattle police use family DNA, latest science to identify... by droithomme on Saturday May 18, 2019 @10:19PM (#19951007)
I agree with what you say but more interesting to me is that the way DNA is used in forensics is pseudoscientific nonsense, along with bite and footprint matching.

As a separate issue, this guy is dead and has no ability to defend himself in court such as by providing an alibi. In the Golden State Killer case, the first "match" they found in their cold hit/blind match search (using an extremely small number of 9 to 13 SNPs) found a guy that had a solid alibi. They shrugged, kept looking, and found some other guy who didn't have an alibi. Which proves nothing given the state of cold hit against massive databank methodology combined with a bare minimum of competence at genetics and statistics. Such is the unprofessional state of so-called DNA forensic "science".

Frank Wypych, as an undisputable legal fact, is innocent until proven guilty. Since there's no been trial or jury verdict LEGALLY HE IS INNOCENT. That is the legal fact. Anyone claiming otherwise, such as the experts in this article, is lying and should forever more not be trusted to speak their own weight much less decide who is guilty of a capital crime.

Re:Seattle police use family DNA, latest science to identify... by dahinds on Sunday May 19, 2019 @12:36AM (#19951461)
You seem to think that these cases are being solved using old school CODIS database matching but that's just false. This case and the Golden State Killer case were solved using modern distant relative matching methods that are widely used in the genetic genealogy community and are far more powerful than traditional forensic DNA matching and typically use 500,000 or more markers.
[dead]
Re:Seattle police use family DNA, latest science to identify... by dang on Sunday May 19, 2019 @09:23PM (#19956822)
It is seriously not ok to be this aggressive or post personal attacks to HN. What incentive does anyone have to participate here if they're going to be harangued like this? Most people have genuine knowledge about the field in which they work, so harassing them about their employer is particularly counterproductive. If we allow commenters to disincentivize that, this place gets less interesting.

Please review the guidelines and don't ever do this again, regardless of how strongly you feel about some issue.

https://news.ycombinator.com/newsguidelines.html

Edit: it looks like we've had to ask you a great many times before to stop breaking the site guidelines in your posts here. I assume you know that we ban accounts that won't stop doing that. Would you mind taking the spirit of this site to heart instead? We're looking for curious, thoughtful conversation.

Re:Seattle police use family DNA, latest science to identify... by droithomme on Sunday May 19, 2019 @11:22PM (#19957303)
Hey Dan. Do you agree that Anne Wojcicki is a very close personal friend of Sam Altman?

What issues I have been discussing with my long time personal friend David here are inappropriate?

Is ycombinator just a sham site to push BS and ban legitimate questions and calling out of scams, all in the service of market manipulation and PR?

Please ban me Dan. It is long overdue and completely in line with Sam's agenda.

Re:Seattle police use family DNA, latest science to identify... by dang on Monday May 20, 2019 @01:25AM (#19957750)
I don't know who Sam's friends are or how that could possibly be relevant here.

If dahinds is your friend and I misread friendly banter as aggression, that's...far from what I was expecting to hear. But in that case your comments should include enough information to make it clear that they aren't the egregious violations of the site guidelines that they sounded like.

I don't want to ban you; I'd much rather persuade you to use HN in the spirit that it's intended. You've also posted some great comments, so that shouldn't be hard.

https://news.ycombinator.com/newsguidelines.html

Re:Seattle police use family DNA, latest science to identify... by dahinds on Monday May 20, 2019 @03:30PM (#19964107)
As far as I know it wasn't friendly banter
Re:Seattle police use family DNA, latest science to identify... by dahinds on Sunday May 19, 2019 @02:52AM (#19951849)
Most of this new class of DNA searches, including the one in the article, are being carried out by Parabon NanoLabs:

http://docs.parabon.com/pub/Parabon_Snapshot_Scientific_Post...

https://www.genomeweb.com/applied-markets/parabon-nanolabs-g... [sorry, paywalled]:

"Parabon uses Illumina’s SNP microarray platform. It doesn't disclose which particular array product it uses but said that the array contains more than 98 percent of the SNPs that are on the chip used by Ancestry.com." Ancestry.com uses a ~700K SNP array. I never claimed full sequences were used.

I do work at 23andMe, I'm not aware of criminal activity there, overt or otherwise. I think we try to be responsible and transparent about how we use customer data. We are not a HIPAA covered entity, but we are GDPR compliant.

Re:Seattle police use family DNA, latest science to identify... by droithomme on Sunday May 19, 2019 @11:56AM (#19953912)
DNA is not medical data? It's obvious that it is and disingenuous to pretend it is not in order to avoid compliance with the law.

Galvin's sample, along with other cold cases, was pulled out of evidence storage and analyzed in 2002. Your assertion is that that analysis did not in fact use the legal tools at the time for this, but instead used the microarray Paradon currently uses with the same ~700k SNP points, and that this microarray was somehow commercially available and being used by law enforcement to analyze cold cases in 2002? That is a remarkable claim.

Re:Seattle police use family DNA, latest science to identify... by dahinds on Sunday May 19, 2019 @02:23PM (#19954640)
You keep doubling down on this?

The article says that DNA from the clothing sample was sent to Parabon for analysis in July of 2016, and was used in a GEDmatch search at that time. There's nothing remarkable about my claim, it is right there in the article. Also it simply isn't possible to submit a CODIS profile to GEDmatch, they only accept SNP array data, the matching algorithms only work for dense SNP array data, and CODIS markers are not even SNPs, they are STRs.

I did not claim genetic information isn't "medical data". I said that 23andMe is not a "covered entity" under HIPAA. HIPAA only applies to "covered entities". Maybe you should say specifically what it is you think 23andMe is doing that is improper?

Re:Seattle police use family DNA, latest science to identify... by droithomme on Sunday May 19, 2019 @04:01PM (#19955170)
> specifically what it is you think 23andMe is doing that is improper

23andMe acquires, analyzes, stores, and trades in private medical data and does not comply with HIPAA in any way when doing so.

We both know that the DNA data you have is personal medical data. You have a PhD and are an executive at one of the largest companies in the world that is trading in DNA data so we both know you are very aware this is true.

> DNA from the clothing sample was sent to Parabon for analysis in July of 2016

We both agree the Galvin DNA sample was acquired and analyzed in 2002 as CODIS data and yielded no hits.

The article says "In July 2018, the suspect DNA from Galvin’s clothing was submitted to Reston" and then goes on to say "The suspect’s profile was reformatted and entered into GEDmatch". Parabon was not sent the actual original DNA from the underwear. Parabon was sent the DNA profile that was created in 2002. It was then reformatted by their data scientists so that they could do a cold hit search on GEDmatch. It yielded possible distant match on two individuals. They found a common ancestor, and created a descent tree of thousands of people, looking for people with a connection to Seattle. They exhumed one and did a modern microarray on that person, and compared that to the extremely sparse data from the 2002 test. They did not do a modern ~700k size microarray on the Galvin data.

You have professional colleagues at Parabon. Ask them. They will confirm what I have said.

Also as a professional biological statistician, you more than almost anyone know that cold hitting a sparse sample against a massive database, finding a possible remote connection, and then expanding the search from a 1800s possible ancestor to include all descendants who live in a large metropolitan area near the crime in a city many people migrated to over the decades is quite likely to find many matches and this sort of match is far from conclusive at all. In fact, you are even qualified to be an expert witness in such a case and testify regarding the actual odds.

We also both know that there are serious problems with testing amplified DNA samples on clothing that has been stored for 35 years and possibly handled by dozens of people.

We also know that even finding high quality single person DNA samples on clothing of a murder victim which exactly matches a suspect (which we did not do at all in this case here) does not prove that the person it matches was the killer and can not reasonably be deemed so without a trial in which a competent defense is presented and considered.

Re:Seattle police use family DNA, latest science to identify... by dahinds on Monday May 20, 2019 @04:03PM (#19964531)
I don't really think you're trying to engage in a good faith discussion, but in case you are:

Lots of companies that handle medical information are exempt from HIPAA. "Many organizations that use, collect, access, and disclose individually identifiable health information will not be covered entities, and thus, will not have to comply with the Privacy Rule."[1] Generally many of our practices meet or exceed HIPAA requirements, either because we're complying with some other rule (like GDPR, or Common Rule for protection of subjects in research), or because we think we should.

You also still haven't given a specific example of something you think we're doing that you don't like. For example acquiring, analyzing and storing medical data is sort of required in order to provide our service to our customers so unless you think DTC genomics is fundamentally without merit this doesn't seem objectionable. If your concern is the "trades in" I think you may not understand exactly how we operate. Our privacy statement says we won't share even de-identified individual-level data without obtaining consent to do so. We rarely do that, for nearly all of our work, we only share statistical results from pooling information from thousands of individuals, where there is no way to extract data about any individual that contributed to the analysis.

As for the original article, I'll stake my professional experience that it's simply not technically possible to do a GEDmatch lookup using "reformatted" CODIS data from 2002. The article says that DNA was submitted to Parabon, not a CODIS profile, Parabon uses SNP arrays, GEDmatch uses SNP arrays, and if you think somehow they didn't in this instance you're frankly delusional.

There certainly are issues with testing old DNA samples and I'm not arguing about any of the difficulties with establishing chain of custody for old evidence, the complexity of statistics for CODIS matching, asserting guilt without a trial, etc. I haven't made any statements about any of that.

[1] https://privacyruleandresearch.nih.gov/pr_06.asp

Re:Seattle police use family DNA, latest science to identify... by Dylan16807 on Sunday May 19, 2019 @05:26AM (#19952369)
It's not lying to call someone guilty based on sufficiently high non-court standards. Sometimes things skip courts, sometimes the prosecutor doesn't do their job, sometimes courts or juries get it wrong.

> They shrugged, kept looking, and found some other guy who didn't have an alibi.

And based on that, they got a search warrant.

Don't imply that their claim is based on the preliminary test.