With Y-DNA, the last I knew is FTDNA stated they were seeing a SNP mutation every 82 to 98 years. STRs (like y111) mutate much faster than SNPs.
Hiya, C.R. Those statements are correct, but they're in mixed context.
When references are made to a SNP mutation rate of ~83 years, the consideration is for any point mutation, any individual allele change across all of the ~23 million (out of 57 million) base pairs tested by the Big Y-700. And I believe the calculation includes all variants, whether a SNP has been named at a given locus or not: with so few nucleotides in the Y allocated to protein coding genes, and the avoidance of crossing over, just about any variant--at least within that 23.6 Mbp (megabase pair, or one million base pairs) target area--that hasn't yet met the basic criterion to be given a SNP name has an excellent possibility of being so labeled in the future. Since Big Y testing began, we've gone from 16,361 yDNA haplotree cataloged branches in September 2018 to 45,366 today. The dbSNP database currently has 2,548,155 named SNPs on file for the Y Chromosome.
The Iain McDonald paper I referenced uptopic does a pretty good analysis to arrive at 33 years as a blanket generalization for the generational interval along patrilineal lines. So if we assume 83 years is going to be roughly where the estimate settles for Big Y data, then we'd be talking about 2.56 generations per any single mutation among any of the ~23 million nucleotides. SNPs will mutate much slower than STRs individually, but if you're looking at an aggregated ~23 million independent events, the odds go way up that one will happen in any N period of time, or rolls of the dice.
The 1000 Genomes Project included 702 Y-STRs in their data, not far off what FTDNA now targets with the Big Y-700, but there are a total of around 4,500 known STRs on the Y...many in the 33.6 Mbp the Big Y doesn't test and that are just chock-full of tons of repetitions, or that are part of the two pseudoautosomal regions.
If we could look at data to give us an idea of how often--bad analogy again, but hey--rolling the dice 4,500 times would turn up a single STR count change, then we might have closer to an apples-to-apples comparison to an 83 years value across 23 million SNPs.
We're unlikely to ever get a decent value from even that exercise, though, because STRs can be flat-out quirky. In part that's because dad's germline DNA changes as he ages: unlike mom's ova, which are all generated via meiosis while she's still in the womb, dad's gametes are an on-demand operation, and as he ages his DNA goes through a process called deamination, a result of methylation, with the result that a child born when dad is 20 is likely to have genetic differences--especially in the Y since it doesn't recombine and so much of it doesn't have to "repair" itself at meiosis--compared to a child born when he's 40 or 50. That deamination process escalates as we age.
Biology rabbit hole. Sorry. Bottom line, though, is that STRs see over time, not uncommonly, back mutations, parallel mutations, and multi-step mutations...the latter where more than one repeat of the STR's sequence of alleles is added or subtracted in a single generation.
Three other fun STR facts. In addition to the father's age throwing us a curve-ball in accurately estimating a mutation rate, mutation rates of STRs seem to strongly correlate to the length of their repeat sequence (e.g., an STR that has only two or three repeated alleles will tend to mutate more slowly than one with, say, five or six alleles. Also, the number of repeats also seems to make a difference: STRs with a greater number of repeats will tend to mutate more quickly than STRs with a small number of repeats (e.g., two of the STRs with rapid mutation rates are DYS710 with up to 36 repeats, and DYS714 with up to 25 repeats).
Last up on the STR Trivia Tour, that latter effect--more repeats, more tendency to mutate--finally substantiated what a lot of us volunteer FTDNA project admins have been seeing for almost 20 years: the same STR seems to mutate at different rates in different haplogroups. Because different haplogroups can have signature repeat values for certain STRs, we now know that those different repeat numbers can lead to an experienced mutation rate difference among haplogroups of the same STR of up to around 20% (Claerhout, et al. "Determining Y-STR Mutation Rates in Deep-routing Genealogies: Identification of Haplogroup Differences." Forensic Science International, Genetics (May 2018) 34:1–10).
So with STRs, the difference in mutation rates between any two of them can be factors of magnitude. Most of my FTDNA projects are R-M269-centric, and to try to help project members make some sense of individual STRs (beyond FTDNA's TiP report, which isn't a terrible place to start, but is about as accurate as pinpointing a street address in San Diego by saying "Southern California") I maintain a spreadsheet that derives individual STR mutation rates from several sources (McDonald; Heinila; Burgarella; Willems; Ravid-Amir; and the NIST's STRBase: https://strbase.nist.gov). Number one in speed may be open to some debate because it's probably the most variable one tested, and that's the palindromic CDY with what I show as a rate of 0.03531 per generation, or a 3.5% chance of mutation per generation. Rounding out the top three are DYS710 (0.018279) and DYS712 (0.016378). At the tortoise end of the speed curve are DYS632 (a glacial 0.00007), DYS436 (0.000204), and DYS426 (0.000216).
Net message, I suppose, is that there's really no good way to lump all the known STRs together in order to arrive at an aggregate mutation rate estimation the same way we can using tens of millions of single nucleotide polymorphisms which, individually, are far more stable and mutate far more slowly. And generational estimates for genealogy using only Y-STRs can look far different once Y Chromosome full-sequence data can be included in the comparisons.
