Biocodicology - A Deeper Dive

[Bernd]

Excellent, the NIST paper is exactly what I was looking for.
The field of forensics is a powerful ally because it is well staffed and funded for obvious reasons.

We clearly see the limitations and pitfalls of analyzing complex DNA mixtures here and considering the DNA on the VM is additionally highly degraded, from an unknown number of unknown people without reference sequences and contaminated by microorganisms, deconvoluting this haystack is not going to be an easy task, even with advanced models like probabilistic genotyping. Yet I'm more optimistic now than before, with enough sampling material (and funding) it should very much be possible to get useful data from both humans and especially the animals used for vellum.

It would be interesting to conduct a test experiment tailored to our needs - eg create a complex mixture of DNA from 10 known people in concentrations expected to find on a manuscript, artificially degrade it to the expected level and then see if these 10 profiles can be resolved by NextGen sequencing and if we get accurate, less or even more reads by artifacts. As the NIST paper states most mixture interpretation was only done with a large amount of fresh DNA from 2 to max 5 persons and does not necessarily reflect the conditions of (semi)ancient DNA, thus greatly reducing the realiability of these forensic models.