FBI researcher says protein sequencing could help compare hairs when DNA fails

Dr. Joseph Donfack, a research biologist with the FBIs Counterterrorism and Forensic Science Research Unit, told an audience that protein sequencing of hair shafts shows promise as a forensic comparator when nuclear and mitochondrial DNA tests are inconclusive. "The answer to this question is probably yes," he said, describing trials that used liquid chromatographytandem mass spectrometry (LCMS/MS) to detect genetically varying peptides in hair.

Donfack opened with a formal disclaimer that his remarks were his personal opinions and not those of the Department of Justice or the U.S. Government. He framed the work as addressing a common forensic problem: a hair found at a scene that yields no usable DNA. Current practice often begins with microscopy and proceeds to DNA methods; when DNA fails, analysts have limited alternatives, he said.

The researcher described two linked studies. The first, a sensitivity study, used hair from a single 24-year-old female donor cut into five lengths from about 2.0 centimeters down to 0.1 centimeters. Donfack reported that peptide and protein counts fell as sample length decreased, but that hair-shaft keratins remained relatively stable while many non-keratin proteins dropped out. He said it was possible to obtain a protein profile from hair shafts 2 centimeters or shorter and that keratin proteins showed high sequence coverage and resistance to length-related loss.

For method details, Donfack explained the workflow: extract protein from the hair shaft, digest with trypsin to produce peptides, separate the peptides by nano-flow liquid chromatography and identify them with tandem mass spectrometry. He said many mass-spectral search engines cannot detect peptides that vary from reference protein sequences, so his team built a custom protein database containing targeted proteins and inferred sequence variants.

Citing prior work by Glendon Parker and colleagues at Lawrence Livermore National Laboratory, Donfack noted that Parkers group identified 77 genetically varying peptides (GVPs) from hairs of 66 subjects and about 600 SNPs confirmed by Sanger sequencing, but required about 10 milligrams of hair (described in the presentation as roughly 200 centimeters of shaft) to obtain that breadth of signal. That material requirement motivated the FBI team's sensitivity work.

The second study Donfack described was a hair-comparison experiment using three female donors (two of European descent and one of Asian descent), with buccal swabs collected for each donor to confirm SAPs by Sanger DNA sequencing. He presented genotype-like peptide profiles (SAPs) for the three subjects and said profiles appeared distinct, but he cautioned that no random-match probabilities have yet been calculated for these data.

A central analytical challenge Donfack emphasized is linkage among SAPs: "some of these SAPs originate from the same protein" or chromosome and therefore are not independent, he said, meaning that the product-rule approach used with STRs to calculate random-match probability will not work here. The team is developing statistical approaches, including consideration of microhaplotypes, to account for such linkage.

Looking ahead, Donfack said the group plans to enrich non-keratin signals (keratins currently dominate the profiles), expand the database to include many more SAPs so profiles become more discriminatory, and perform reproducibility studies to assess within-source variability (for example, hair collected from different scalp sites). "I wasn't a believer in the beginning, but after doing this work, seriously, I was amazed how sensitive LCMS is," he said, noting other experiments that recovered protein from samples as small as 300 micrometers.

The presentation concluded with acknowledgements of visiting collaborators, including Tracy Carlson, and an invitation for questions. In response to an audience question, Donfack said the team would begin with incremental reproducibility studies to determine the technology's practical limits before expanding scope.

The research is ongoing: Donfack reiterated technical promise but also identified clear next steps in database expansion, enrichment of non-keratin signals and development of statistical methods necessary before protein-based hair comparison could be used as a routine forensic tool.