News for hackers. Stuff that matters.
Re:Idaho murders: Suspect was identified through DNA using g... by javajosh on Monday January 02, 2023 @10:44PM (#34227344)
>What I was surprised to learn is that even fingerprints contain DNA, not just bodily fluids.

A single skin cell contains all your DNA. However it's not enough to sequence. Call me an old cynic, but this smells like parallel reconstruction to cover up the real story.

Re:Idaho murders: Suspect was identified through DNA using g... by cjensen on Monday January 02, 2023 @11:26PM (#34227654)
"However it's not enough to sequence"

I'm not sure what you mean by that. The nice thing about DNA is you can make more of it if you don't have enough. [1].

[1] https://en.wikipedia.org/wiki/Polymerase_chain_reaction

Re:Idaho murders: Suspect was identified through DNA using g... by javajosh on Tuesday January 03, 2023 @12:00AM (#34227877)
How do you ensure that you don't amplify the wrong sample? That would seem to be the crux of the matter, especially for a sample from a fingerprint. Seems like most of the time you'll end up sequencing non-human DNA, given the disgustingness of the biological environment.

EDIT: ...or another persons DNA. Note that in general you aren't sequencing against a desired match, but generating a target for some future sequence (obtained from a suspect) to match against.

Re:Idaho murders: Suspect was identified through DNA using g... by asdff on Tuesday January 03, 2023 @12:33AM (#34228083)
If you take a sample at a crime scene and its a match to your suspect, then there's no question what it is. If you sequence non human DNA you won't get a match. If you sequence the forensic lab technician's DNA by accident when taking the fingerprint sample you still won't get a match to the suspect.
Re:Idaho murders: Suspect was identified through DNA using g... by candiodari on Tuesday January 03, 2023 @07:46AM (#34230706)
YES there is.

PCR can make things up when it goes wrong, which can be as easy as getting the acidity wrong (acidity will cut DNA, which will then recombine into new sequences randomly, so will UV light. Both of these factors, you can bet your firstborn, were present in the sample). PCR (the kind used for sequencing, not the kind used for identification of viruses) does not fail when it is not given DNA or when it's given somehow bad DNA, as everyone in a microbiology lab has probably experienced, it just makes things up. If there is any DNA in the sample to trigger PCR to start, it will "finish". Just ... potentially with a lot of made up endings.

Given what exactly DNA is, and how this was collected, the odds of errors external to the testing procedure itself is pretty damn high (tiny sample, contaminated by loads of frankly incompetent people (police, whatever else you think of them, can seriously improve handling of samples. I'm not even arguing foul play here), contaminated/partially eaten by bacteria, irradiated by UV (which first mutates, then destroys DNA faster than you can say skin cancer), then this was inserted into a database with millions of samples.

Oh, and labs have a financial incentive to make this evidence stick.

Then the output: sequences are a LOT of high probability guesses (millions). So you will never have a 100% match between a sequence, as in a string, and a sequence you found. The output of gene sequencing is not the full genome, but a large amount of short sequences that don't quite fit, but almost, where an algorithm then extracts what might be called "globbing pattern" (if DNA is AGCCAGCAATA, the match might be AG?CA*AAGA (note the mistake at the end T->G). Yes, this is exaggerated, and yes, there's no '*', there's a special marker for "up to X letters missing here", where X can vary).

Important: these patterns contain mistakes. There are ZERO actual DNA sequences that match the output of sequencing. That's where probability comes in.

You do not get the actual sequence (although a "highest probability" match without ? and * can be produced, with the very important caveat that it's a guess with not all that high a probability. That's why they probably matched by matching the "globbing patterns" from samples in the database vs the evidence and then selecting the highest number of matches, rather than comparing an actual DNA "string").

One very relevant question to ask is "what was the probability of this match and what was the probability of the second best match?" (because it's going to be something like ~85% for best match and 75 or more for second best match, and there will be 100+ 20% matches)

And then there's the statistics problem. The police will focus on the odds that THIS match, BY ITSELF, happened. But this is not one match! This is the outcome of millions of matches! The vast majority were negative matches, and then a few hundred/thousand partial matches. That's how the misdirection works. The correct question to ask is "what are the odds, given a true negative rate, of a match if there shouldn't be a match?".

Why? Let's calculate: true negative rate 99.999%, a million people in the database. Odds of a false positive? 1-(99.999% ^ 1000000).

The odds of them fingering a random suspect when there is no positive match is thus: 99.99546023401903%.

As you can see, the more people in the DNA database, the less you can trust the results. Of course, you cannot, honestly, sidestep this problem by checking 10 databases.

This DNA match can be a starting point for an investigation. It is not valid evidence, despite the reputation DNA has. But you can bet anything the prosecutor will attempt to use DNA evidence as the word of God in a trial.

And where there is no question about at all, is that the police used people's DNA (not the murderer's) for something those people have never given permission for.

Re:Idaho murders: Suspect was identified through DNA using g... by dekhn on Thursday January 05, 2023 @10:28AM (#34260996)
Acidity does not cause DNA to recombine into new sequences spontaneously.

Also your details about error rates in sequencing aren't correct either (you often get completely accurate reads of reasonable lengths).

Re:Idaho murders: Suspect was identified through DNA using g... by dylan604 on Tuesday January 03, 2023 @01:07AM (#34228307)
how can there be a wrong sample? if the sample leads to someone that they can prosecute, it can't be wrong. only guilty leave DNA samples at crime scenes. this is known.
Re:Idaho murders: Suspect was identified through DNA using g... by kenmacd on Tuesday January 03, 2023 @09:52AM (#34232061)
Pretty sure this is sarcasm, but for general interest I'll mention the Phantom of Heilbronn (https://en.wikipedia.org/wiki/Phantom_of_Heilbronn), "responsible" for multiple murders, but really just an employee at the company that made the cotton swabs that got used for DNA collection.
Re:Idaho murders: Suspect was identified through DNA using g... by dylan604 on Tuesday January 03, 2023 @05:37PM (#34239147)
Yes on the sarcasm, but even better on you for posting an interesting link for a reply.
Re:Idaho murders: Suspect was identified through DNA using g... by mtqwerty on Tuesday January 03, 2023 @12:01AM (#34227884)
PCR will only amplify DNA that exists in a sample. If you get a partial sample, running it through PCR will just give you more of what you already have. Helpful if you need a specific concentration of DNA to run an assay but not helpful otherwise.

It is possible to link a person to a partial DNA sample with something like patterns in SNPs but PCR won't help you create a full genome from a partial one.

Not sure if OP was misspeaking when they said "not enough to sequence" but they may have meant, not enough DNA to get a definitive match with the database samples.

Re:Idaho murders: Suspect was identified through DNA using g... by smitty1e on Monday January 02, 2023 @11:02PM (#34227500)
These days, skepticism is always in order.
Re:Idaho murders: Suspect was identified through DNA using g... by ROTMetro on Monday January 02, 2023 @11:44PM (#34227771)
Ironic that the original purpose of fingerprint matching itself is subjective pseudo science, just like bite mark comparisons, ballistic comparisons, and shoe and tire prints. But that doesn't stop them from being used.
Re:Idaho murders: Suspect was identified through DNA using g... by sn_master on Monday January 02, 2023 @11:48PM (#34227798)
Bite mark was proven false, but what's suspicious about matching fingerprints?
Re:Idaho murders: Suspect was identified through DNA using g... by giantg2 on Monday January 02, 2023 @11:51PM (#34227813)
I don't know how true it is but I've heard that they haven't done any studies to find out how likely it is that two people have the same prints. I have also heard that there is no objective criteria for things like number of matching points to determine a "match". But maybe the original commenter has additonal info, or other people have info to refute it.
Re:Idaho murders: Suspect was identified through DNA using g... by tedunangst on Tuesday January 03, 2023 @12:14AM (#34227972)
Re:Idaho murders: Suspect was identified through DNA using g... by mc32 on Monday January 02, 2023 @11:56PM (#34227851)
The only things I can think of is that partials can match other fingers’s fingerprints and the temporal aspect —when was the print left, in the act of committing the crime or before or after?

Still it’s not pseudoscience like lie detectors, bitemark and bloodsplatter analyses.

Re:Idaho murders: Suspect was identified through DNA using g... by thrwawy74 on Monday January 02, 2023 @11:45PM (#34227774)
Speaking as a non-expert, it sounds too good to be true with the regular reports of false-positives and false-negatives from DNA testing. I'm with this person: They did something illegal to save lives.
Re:Idaho murders: Suspect was identified through DNA using g... by vba616 on Monday January 02, 2023 @11:57PM (#34227859)
It seems reasonable that DNA evidence should not be used to convict someone alone, but if it's used as a starting point to find other evidence, well and good.

Couldn't that other evidence be perfectly legal but not publicized yet?

Re:Idaho murders: Suspect was identified through DNA using g... by asdff on Tuesday January 03, 2023 @12:44AM (#34228159)
These reports are not regular to be honest...

"Think of it this way: There are many thousands of paintings with blue backgrounds, but fewer with blue backgrounds and yellow flowers, and fewer still with blue backgrounds, yellow flowers, and a mounted knight in the foreground. When a forensic analyst compares alleles at 13 locations—the standard for most labs—the odds of two unrelated people matching at all of them are less than one in 1 billion." 1

In the past, there were issues with not getting these 13 sites sufficiently covered and losing statistical power. Today, to get around issues of certain markers dropping out for different reasons and to continue to ensure high statistical power, the FBI uses 20 markers now and samples must have a sufficient number of markers covered to be considered. 2

1. https://www.theatlantic.com/magazine/archive/2016/06/a-reaso...

2. https://www.nist.gov/news-events/news/2016/12/nist-research-...

Re:Idaho murders: Suspect was identified through DNA using g... by tinus_hn on Tuesday January 03, 2023 @11:42AM (#34233828)
> the odds of two unrelated people matching at all of them are less than one in 1 billion.

Sounds like on a world with 7 billion people for each crime there are on average 7 people that can be convicted using the incontrovertible evidence delivered by the DNA lab.

Re:Idaho murders: Suspect was identified through DNA using g... by asdff on Tuesday January 03, 2023 @12:03PM (#34234179)
Do all 7 people have a motive and no alibi? Probably not.
Re:Idaho murders: Suspect was identified through DNA using g... by tinus_hn on Tuesday January 03, 2023 @01:02PM (#34235148)
Who knows, chances are most of them are related.
Re:Idaho murders: Suspect was identified through DNA using g... by asdff on Tuesday January 03, 2023 @02:14PM (#34236472)
I think you know what a rare lightning strike of an even this would be and how without DNA evidence, conclusions would be even more dubious.
Re:Idaho murders: Suspect was identified through DNA using g... by tinus_hn on Tuesday January 03, 2023 @04:39PM (#34238465)
Sorry, if the chances are one in a billion, and there’s 7 billion to choose from it’s actually not rare at all. That’s why you should be very careful to either use this kind of evidence either to find a suspect or as proof, and not both. But I’m sure we all know there’s plenty of ‘tough on crime’ judges that wouldn’t care about convicting someone based on this kind of seemingly rare occurrence.
Re:Idaho murders: Suspect was identified through DNA using g... by asdff on Wednesday January 04, 2023 @02:29PM (#34250204)
Every piece of evidence presented in the court room has some probability of implicating someone else. That's why multiple pieces are generally presented.
Re:Idaho murders: Suspect was identified through DNA using g... by daemonk on Monday January 02, 2023 @11:46PM (#34227783)
It's difficult, but not impossible. There are plenty of whole genome amplification methods out there for single cell DNA-seq.
Re:Idaho murders: Suspect was identified through DNA using g... by asdff on Tuesday January 03, 2023 @12:36AM (#34228103)
You won't be collecting a single skin cell though, you will be collecting millions in that speck and there's plenty of redundancy there. Plus you don't need to cover the entire genome, just some variant sites that can give you enough statistical power to be confident in the identity of the sample.
Re:Idaho murders: Suspect was identified through DNA using g... by whycome on Tuesday January 03, 2023 @01:48PM (#34235996)
Was the article headline updated to include the “Police say”? Because it’s good to see that police statements aren’t just accepted as fact and also need to be verified