Software in the Crime Lab: What a Florida Defendant Should Ask For

On June 9, 2024, a federal judge in San Diego excluded a DNA result reporting that the profile from a gun swab was 542 million times more likely if the defendant contributed to the mixture than if the defendant didn’t. The number came from STRmix, a program that crime labs use to sort out DNA mixtures, and the judge found the government hadn’t shown the program was used on the kind of sample it had been tested to handle.

Software like that now sits between the evidence and the jury in DNA cases, and it has sat inside breath and blood alcohol testing for much longer. A lab report can give the software’s answer without the settings, the assumptions or the version that produced it, and those are the things I ask for first.

Follow this site in Google:

The Software Behind a DNA Number

An object touched by several people can yield a mixture of their DNA, with the profiles layered on top of each other. Probabilistic genotyping programs such as STRmix and TrueAllele take that mixture and produce a likelihood ratio: how much more likely the data is if a particular person contributed than if they didn’t. The output is a single number, and it’s the number the jury hears.

In the San Diego case, the court explained that STRmix relies on the analyst to tell it how many people contributed to the mixture, and the analyst decides that by reading peaks on a graph called an electropherogram. Every calculation after that rests on the analyst’s count. This older statistical software is a different thing from the newer machine learning tools that get most of the attention, and it’s already deciding cases.

The San Diego Ruling

In United States v. Ortiz, No. 21-CR-2503-GPC (S.D. Cal. June 9, 2024), the San Diego Police Department crime lab had validated STRmix for mixtures of up to five people. Its rule was that a six-person sample is unsuitable for comparison. The analyst counted five contributors and ran the program. The defense brought in a DNA scientist who testified that at several locations on the graph there were twelve peaks, which points to at least six people. The analyst read two of those peaks as artifacts of the testing process.

The court went with the defense on the facts. The DNA sample was small, and it was injected three times, each injection using up part of it. The analyst also hadn’t been told that the gun was found in a car with a mother and her daughter, and relatives share DNA markers, which can hide a contributor. The court cited a 2018 study of 31 laboratories in which analysts got the contributor count wrong on 64% of five-person samples and on every six-person sample, always guessing too low. It held that the government failed to show STRmix had done an analysis it was validated to perform, and it excluded the evidence under both Rule 702 (expert reliability) and Rule 403 (unfair prejudice).

The Fight Over Source Code

A second question is whether the defense can look inside the program at all. In State v. Pickett, No. A-4207-19T4, 466 N.J. Super. 270 (App. Div. 2021), the State’s witness was the man who designed TrueAllele, and he testified that it has about 170,000 lines of code and that reading it at ten lines an hour would take about eight and a half years. New Jersey’s appellate court held that when the State relies on a witness using novel probabilistic genotyping software, the defendant is entitled to the source code and related development records, under a protective order, to challenge the software’s reliability. The defendant first has to show a particularized need. On the trade secret argument, the court wrote: “Hiding the source code is not the answer.”

Ortiz and Pickett come from California and New Jersey. Neither one binds a Florida court, and Pickett was decided under the older Frye test (general acceptance in the field), which Florida no longer uses.

Where Florida Stands

Florida follows Daubert. Section 90.702, Florida Statutes, lets a qualified witness give an opinion only if it rests on sufficient facts or data, comes from reliable principles and methods, and applies those methods reliably to the facts of the case. The Florida Supreme Court adopted that standard as a rule of evidence in 2019. Ortiz turned on the third prong, reliable application to this sample, rather than on whether STRmix works in general.

Florida’s own appellate case on this software went the other way. In Daniels v. State, No. 4D19-822 (Fla. 4th DCA Feb. 24, 2021), the Fourth District called the issue one of first impression and affirmed the admission of TrueAllele results in a murder case. The defense argued that the program hadn’t been validated on data from the local sheriff’s lab. The court noted that the defense’s witness wasn’t familiar enough with TrueAllele to explain how that gap compromised the result, and that no appellate decision supported the argument. Daniels rejected an internal-validation challenge on its record. Ortiz shows how evidence that a particular sample fell outside what the lab had validated can lead to exclusion.

Beyond DNA: Breath and Blood

DUI lawyers were fighting over software long before DNA mixtures were. A breath machine is a small computer, and its code decides when a sample counts as valid, how it checks for interfering chemicals, and how readings are averaged. Florida’s leading appellate case on breath test code went against the defense. In State v. Bastos, 985 So. 2d 37 (Fla. 3d DCA 2008), the court held the source code for the Intoxilyzer 5000 wasn’t shown to be material and said there would need to be “a particularized showing demonstrating that observed discrepancies in the operation of the machine necessitate access to the source code.” It’s the same idea Pickett used: show the specific problem first. My breathalyzer source code page covers that fight in Florida.

Blood alcohol is run on a gas chromatograph, and the instrument’s software draws the start and end of each peak and measures its area. That area becomes the reported number. An analyst can accept the software’s integration or redraw it by hand, and the instrument’s audit trail records which happened. The printout shows the result, while the electronic raw data and the audit trail show how the lab got there. My gas chromatography page walks through where that process breaks.

The Records That Show What the Software Did

When DNA, a breath test or a blood result carries a case, I start by finding out what software touched the evidence. For DNA, I ask for the program’s name and exact version, the vendor’s developmental validation, this lab’s internal validation and the limits it set, the analyst’s contributor count with the bench notes behind it, and the raw electronic data so an independent DNA analyst can rerun it. I also ask whether any algorithm or facial recognition tool was used at any stage of the investigation, because a lead can come from software that never shows up in the lab file. For breath, I ask for the instrument’s software version, its inspection and maintenance records, and any error or exception codes. For blood, I ask for the electronic chromatographic data, the audit trail showing whether anyone reintegrated a peak, and the lab’s integration procedure.

For lab evidence more broadly, my forensic science defense page explains how I approach it, the DNA and serology defense page covers mixtures and touch DNA, the Daubert page covers the hearing where these challenges are made, and my DUI blood test defense page covers blood draws in DUI cases.

This post is general information about the law, not legal advice, and it does not create an attorney-client relationship. Past results do not guarantee a similar outcome. The law discussed is section 90.702, Florida Statutes; In re Amendments to the Florida Evidence Code, 278 So. 3d 551 (Fla. 2019); Daniels v. State, No. 4D19-822 (Fla. 4th DCA Feb. 24, 2021); and State v. Bastos, 985 So. 2d 37 (Fla. 3d DCA 2008). The out-of-state decisions discussed, which do not bind Florida courts, are United States v. Ortiz, No. 21-CR-2503-GPC, ECF No. 171 (S.D. Cal. June 9, 2024), and State v. Pickett, 466 N.J. Super. 270 (App. Div. 2021).

Available 24/7 for Immediate Defense

Your first strategy session is free. One call can start protecting your future today.
Get in Touch

You’re better Safir than sorry!

Arrested for DUI? Time matters. Complete the form to schedule a free strategy session with attorney Rory Safir. Your information is confidential, and we will follow up promptly.

200+
Client Testimonials
1 of 6
Forensic Lawyer-Scientists in Florida
4.9★
Google Rating
24/7
Availability

Let’s Go Over Your Case


Email Newsletter