Forensic Genealogy and Probabilistic Genotyping in the 1974 Ted Bundy Cold Case

Forensic Genealogy and Probabilistic Genotyping in the 1974 Ted Bundy Cold Case

Modern forensic genetics resolves half-century-old cold cases not through a single breakthrough, but by chaining three distinct analytical layers—whole-genome sequencing of degraded samples, probabilistic genotyping of complex mixtures, and Investigative Genetic Genealogy (IGG)—to bypass the structural limits of traditional short tandem repeat (STR) profiling. When investigators linked Ted Bundy to a 52-year-old homicide, the evidentiary chain relied on overcoming fifty years of hydrolytic and oxidative DNA damage that rendered legacy forensic tools useless. Understanding how this linkage occurred requires dissecting the biochemical decay of biological evidence, the mathematical models used to separate overlapping genetic profiles, and the statistical mechanics of distant kinship inference.

The Biochemical Bottleneck of Legacy Forensics

Standard forensic identification relies on Polymerase Chain Reaction (PCR) amplification of 20 core autosomal STR loci, the standard required for upload to the Combined DNA Index System (CODIS). STR markers consist of repeating sequences of two to six base pairs. Because these target regions span roughly 100 to 400 base pairs in length, the DNA template must remain intact across that entire span for PCR primers to bind and copy the sequence.

Biological evidence stored for five decades—particularly outside climate-controlled facilities—undergoes severe molecular degradation. Three primary mechanisms destroy the integrity of legacy samples:

  1. Hydrolytic Depurination: Water molecules cleave the glycosidic bonds between purine bases (adenine and guanine) and the sugar-phosphate backbone, creating abasic sites that destabilize the DNA strand and cause single-strand breaks.
  2. Oxidative Base Modification: Free radicals alter nucleotide structures, blocking DNA polymerase from reading the template during amplification.
  3. Environmental Cross-Linking: Exposure to UV light, humidity, and chemical contaminants creates covalent bonds between DNA strands and surrounding proteins, preventing strand separation during the denaturation phase of PCR.

When average fragment lengths drop below 100 base pairs, STR profiling fails entirely. In a 52-year-old cold case, extracting a clean, full-locus STR profile from trace biological material on clothing, ligatures, or fingernails is statistically improbable. Investigators face allelic dropout—where one or both alleles at a given locus fail to amplify—resulting in partial, inconclusive profiles that cannot meet the match thresholds required for database hits.

Single Nucleotide Polymorphisms and Whole-Genome Sequencing

To extract viable data from fragmented templates, forensic laboratories shift from length-based STR analysis to sequence-based Single Nucleotide Polymorphism (SNP) genotyping. Unlike STRs, which require long, unbroken strands of DNA, a SNP represents a variation at a single nucleotide position. Because the target region is a single base pair flanked by short primer-binding sites, SNP assays successfully interrogate DNA fragments as short as 30 to 50 base pairs.

In high-profile cold cases involving historical figures like Ted Bundy, laboratories employ Whole-Genome Sequencing (WGS) or targeted SNP microarray capture rather than capillary electrophoresis. The operational workflow follows a strict sequence to prevent contamination from modern exogenous DNA:

  • Ultra-Clean Extraction: Samples undergo enzymatic digestion and silica-based purification in positive-pressure cleanrooms equipped with UV sterilization to isolate endogenous human DNA from microbial contaminants.
  • Library Preparation for Damaged Templates: Technicians use single-stranded DNA library preparation techniques. Standard double-stranded methods lose fragmented molecules with single-strand nicks; single-stranded protocols capture both intact and nicked strands, doubling the yield from degraded samples.
  • High-Throughput Sequencing: Short-read sequencing platforms generate millions of overlapping reads across hundreds of thousands of SNP markers distributed throughout the 22 autosomal chromosomes and the X and Y chromosomes.

This shift changes the unit of forensic currency. Instead of comparing 20 highly polymorphic STR loci against a criminal database, analysts generate a dense digital map containing upwards of 500,000 to 1,000,000 SNPs. This density provides sufficient statistical power to establish identity or biological kinship even when 70% to 80% of the original genomic material has decayed.

Probabilistic Genotyping of Low-Template Mixtures

Cold case evidence rarely contains a single, uncontaminated biological source. Evidence collected in the 1970s—before modern chain-of-custody protocols mandated sterile gloves, masks, and individual paper packaging—frequently exhibits commingled DNA from the victim, the perpetrator, and decades of handling by detectives, lab technicians, and court clerks.

Traditional binary interpretation methods fail when analyzing low-template mixtures. Under binary rules, analysts apply fixed stochastic thresholds to peak heights on an electropherogram; any signal falling below the threshold is discarded to avoid mistaking instrument noise for true alleles. When dealing with degraded perpetrator DNA mixed with abundant victim DNA, the perpetrator's alleles consistently fall below this binary threshold.

Probabilistic genotyping software resolves this limitation by replacing binary cutoffs with continuous mathematical modeling. Using Markov Chain Monte Carlo (MCMC) simulations, the software evaluates millions of possible genotype combinations to calculate a Likelihood Ratio (LR). The LR compares the probability of observing the mixed DNA evidence under two competing hypotheses:

$$LR = \frac{P(E \mid H_p)}{P(E \mid H_d)}$$

Where $E$ represents the observed genetic evidence, $H_p$ is the prosecution hypothesis (e.g., the victim and a specific suspect contributed to the mixture), and $H_d$ is the defense hypothesis (e.g., the victim and an unknown, unrelated individual contributed to the mixture).

The algorithm explicitly models biological artifacts that ruin manual interpretation:

💡 You might also like: The Cold Truth Hanging from the Window
  • Stutter Peaks: Slippage of DNA polymerase during amplification creates minor peaks one repeat unit shorter than the true allele. Probabilistic models calculate expected stutter ratios based on sequence motif and length, distinguishing stutter from a minor contributor's true allele.
  • Allelic Dropout Probabilities: Rather than ignoring missing alleles, the software assigns a mathematical probability to dropout events based on the estimated template mass and degradation index of the sample.
  • Degradation Curves: The model applies an exponential decay curve across molecular weight, accounting for the expected drop in signal intensity as fragment length increases.

By quantifying uncertainty rather than discarding suboptimal data, probabilistic genotyping isolates the minor contributor's genetic signature from a degraded mixture, generating a mathematically defensible profile suitable for downstream comparison.

Investigative Genetic Genealogy and Identity-by-Descent

If a direct reference sample from a deceased suspect like Ted Bundy is unavailable—or if the generated SNP profile cannot be uploaded directly to CODIS, which exclusively indexes STR markers—investigators deploy Investigative Genetic Genealogy (IGG).

IGG relies on Identity-by-Descent (IBD) segment analysis. Unlike STR matching, which looks for exact allelic matches at specific criminal loci, IBD measures continuous blocks of shared SNPs inherited without recombination from a common ancestor. The length of these shared segments, measured in centiMorgans (cM), directly correlates with generational distance:

  • Parent/Child: ~3,400–3,700 cM
  • First Cousins: ~550–1,200 cM
  • Third Cousins: ~30–150 cM
  • Fifth Cousins: ~0–30 cM

The operational execution of IGG follows a rigorous four-stage funnel:

  1. Database Ingestion: The crime scene SNP profile is formatted into a compatible digital file and uploaded to opt-in public genetic genealogy databases (such as GEDmatch or FamilyTreeDNA). Commercial direct-to-consumer databases like AncestryDNA and 23andMe prohibit law enforcement access via terms of service and technical firewalls.
  2. Triangulation and Clustering: Genealogists identify matches sharing between 40 cM and 200 cM with the unknown profile. By examining shared matches among these distant relatives, analysts group them into distinct genetic clusters corresponding to specific ancestral couples (Most Recent Common Ancestors, or MRCAs).
  3. Pedigree Reconstruction: Using public vital records, census data, obituaries, and historical archives, genealogists build family trees forward in time from the identified MRCAs, tracing the descendants down to the generation of the perpetrator.
  4. Target Isolation and Direct Confirmation: Intersection points between maternal and paternal lineage clusters narrow the suspect pool to a specific family branch. Investigators eliminate candidates based on age, geography, sex, and known timelines. Once a prime candidate (such as Bundy) is identified, the investigative lead is confirmed via direct comparison against a known reference sample—either from archived post-mortem biological specimens, pathology slides, or direct kinship testing of known biological relatives.

Operational Constraints and Methodological Failure Modes

Treating SNP sequencing and IGG as infallible solutions introduces severe operational and legal vulnerabilities. Practitioners managing cold case units must account for four primary failure modes:

  • Endogenous DNA Exhaustion: Extraction for whole-genome sequencing is destructive. If a 50-year-old swab contains fewer than 50 picograms of human DNA, a failed extraction permanently destroys the sole remaining physical evidence. Labs must quantify human-specific DNA via real-time quantitative PCR (qPCR) before committing samples to library preparation.
  • Pedigree Collapse and Endogamy: In populations with high rates of endogamy or historical geographic isolation, individuals inherit identical DNA segments through multiple ancestral pathways. This inflates shared centiMorgan totals, causing distant fifth cousins to appear genetically identical to second cousins and skewing generational estimates.
  • Misattributed Parentage Events (NPEs): Public genealogical trees assume biological parentage matches legal documentation. Unrecorded adoptions, infidelity, and sperm donation introduce silent breaks in documentary pedigrees. A single NPE between the MRCA and the suspect invalidates the documentary branch, requiring genealogists to pivot to collateral lines.
  • Imputation Errors in Low-Coverage Genomes: When sequencing coverage drops below 1x (meaning each base pair is read less than once on average), bioinformatics pipelines use statistical imputation—referencing population haplotype panels like the 1000 Genomes Project—to infer missing genotypes. Over-reliance on imputation in low-coverage samples introduces false-positive SNP calls, generating spurious IBD matches that waste hundreds of investigative hours.

The Standard Operating Blueprint for Legacy Cold Cases

Law enforcement agencies and forensic directors auditing pre-1990 homicide archives must abandon ad-hoc testing in favor of a strict triage protocol. Submitting degraded evidence directly to standard STR pipelines wastes limited biological material and delays resolution.

  1. Conduct Non-Destructive Triage: Prioritize items shielded from environmental hydrolysis—sealed envelopes, inside linings of garments, ligature knots, and fingernail clippings—over exposed surfaces.
  2. Enforce qPCR Quantification Gates: Establish strict template thresholds. Samples yielding degraded DNA with a high degradation index must bypass standard STR amplification entirely and route directly to single-stranded library preparation for SNP capture or WGS.
  3. Deploy Continuous Probabilistic Software: Mandate the use of validated probabilistic genotyping platforms for all mixed profiles, prohibiting manual binary threshold truncation on low-template extracts.
  4. Execute Parallel Reference Banking: Simultaneously locate and verify post-mortem biological artifacts (such as paraffin-embedded tissue blocks or archived blood cards) for known serial offenders active within the relevant geographic and temporal windows, enabling immediate 1-to-1 STR or SNP verification without awaiting multi-month genealogical tree reconstruction.
CW

Charles Williams

Charles Williams approaches each story with intellectual curiosity and a commitment to fairness, earning the trust of readers and sources alike.