MitoFREQ: Mitochondrial Frequency Analysis

• MitoFREQ is a multifaceted concept that spans frequency‐based mitochondrial analyses including forensic estimation, sequence composition, dynamic oxidative responses, and spectral or relaxometric imaging.
• The forensic MitoFREQ method estimates mitogenome frequencies by combining top-level haplogroup data with the rarity of single nucleotide variants from large databases such as HelixMTdb and gnomAD.
• Additional applications of MitoFREQ involve triplet-frequency phylogenomics, mROS flash dynamics, 2PPT metabolic fingerprinting, and NV-center relaxometry to monitor mitochondrial state.

Searching arXiv for “MitoFREQ” and directly related papers to ground the article in the cited literature. MitoFREQ denotes a set of frequency-based approaches for characterizing mitochondrial information, but the term is not used uniformly across the literature. In its formal and explicit sense, MitoFREQ is a forensic method for estimating mitogenome population frequencies from top-level haplogroups and single nucleotide variants (Andersen et al., 15 Jan 2026). In adjacent work, the same label is used only conceptually or retrospectively for mitochondrial representations based on triplet-frequency composition, temporally patterned oxidative responses, spectrally resolved metabolic signatures, or frequency-sensitive relaxometry (Sadovsky, 2014, Wang et al., 2015, Hai et al., 15 Apr 2026, Su et al., 10 Dec 2025). This suggests that MitoFREQ is best understood as an umbrella notion for mitochondrial readouts in which frequency, spectral structure, or conditioned variant frequencies serve as the principal analytic substrate.

1. Terminological scope

Only one paper in the supplied literature formally introduces MitoFREQ as the name of a method: “MitoFREQ: A Novel Approach for Mitogenome Frequency Estimation from Top-level Haplogroups and Single Nucleotide Variants” (Andersen et al., 15 Jan 2026). The remaining papers are related by conceptual alignment rather than formal nomenclature. One treats the triplet-frequency composition of mitochondrial genomes as the relevant object of analysis and describes the $W_3$ dictionary as, in effect, a mitochondrial MitoFREQ pattern (Sadovsky, 2014). Another is presented as providing a mechanistic foundation for a frequency-based mitochondrial response regime marked by repeated mROS flashes and membrane-potential oscillations (Wang et al., 2015). A 2026 microscopy study is described as MitoFREQ-like because it extracts a spectrally discriminative mitochondrial metabolic fingerprint from intrinsic chromophores (Hai et al., 15 Apr 2026). A relaxometry study is similarly framed as aligned with a mitochondria-centered frequency-resolved sensing concept, because the NV center probes magnetic-noise fluctuations near $2.87~\mathrm{GHz}$ in the mitochondrial compartment (Su et al., 10 Dec 2025).

Usage	Core object	Representative paper
Formal forensic MitoFREQ	TLHG frequency combined with rarest SNV frequency	(Andersen et al., 15 Jan 2026)
Sequence-composition MitoFREQ	$W_3$ triplet-frequency dictionary in 63-dimensional space	(Sadovsky, 2014)
Dynamic response MitoFREQ	repeated mROS flashes and MMP oscillations during reversible fragmentation	(Wang et al., 2015)
Spectral/metabolic MitoFREQ	2PPT signatures of NADH and FAD in mitochondria	(Hai et al., 15 Apr 2026)
Relaxometric MitoFREQ	mitochondrial $T_1$ response to local paramagnetic noise	(Su et al., 10 Dec 2025)

A common misconception is to treat these usages as a single standardized framework. The supplied literature does not support that interpretation. A more accurate reading is that MitoFREQ names one specific forensic estimator and also serves, in broader discussion, as a convenient label for several mitochondria-focused frequency-based representations.

2. Forensic MitoFREQ as a mitogenome frequency estimator

In forensic genetics, MitoFREQ is a method for estimating the population frequency of a mitogenome when direct whole-sequence querying is not feasible but top-level haplogroup and SNV information are available (Andersen et al., 15 Jan 2026). The method uses two public resources, HelixMTdb and gnomAD, which harbour information from 195,983 and 56,406 mitogenomes, respectively. These resources do not support direct individual mitogenome frequency queries, but they provide TLHG distributions and SNV frequencies within each of 30 top-level haplogroups.

The core construction is explicitly conservative. After assigning a queried mitogenome to a TLHG, the method identifies the rarest observed SNV within that TLHG and weights that conditional SNV frequency by the TLHG frequency. If an SNV appears $x$ times among $n$ individuals in the TLHG, its estimated conditional frequency is $\hat p = x/n$. If both databases are pooled, the paper uses

$$\hat p_{\text{pooled}} = \frac{x_{\text{Helix}} + x_{\text{gnomAD}}}{n_{\text{Helix}} + n_{\text{gnomAD}}}.$$

The resulting mitogenome-frequency approximation is

$$P(\text{mitogenome}) \leq P(\text{TLHG})\,P(X_k \mid \text{TLHG}),$$

and the likelihood ratio is

$$LR = \frac{1}{P(\text{TLHG})\,P(X_k \mid \text{TLHG})}.$$

The paper further states that $2.87~\mathrm{GHz}$0 under its formulation, and proves that a refined haplogroup model cannot increase the estimated population frequency.

The workflow begins with TLHG inference. The paper derives a minimal panel of 227 specific positions from 39 haplogroup-defining motifs; these include practical collapses such as L4, L5, and L6 into L4-6, and R and B into R/B. Using only these 227 positions, rank 1 TLHG concordance was about 99.0% for GenBank, 99.5% for SWE, 99.8% for US2015, and 99.7% for US2020. Allowing rank 1 or rank 2 prediction raised concordance to 99.9% for GenBank and 100.0% for SWE, US2015, and US2020. In operational use, the paper computes predictions for both rank 1 and rank 2 and then uses the smallest LR of the two.

Validation was performed on 2,849 high-quality mitogenomes from forensic/reference datasets and on 61,295 scrutinized GenBank mitogenomes. Frequency estimates from HelixMTdb and gnomAD were highly consistent for the 4,762 comparable SNVs, with a Pearson correlation of $2.87~\mathrm{GHz}$1 on $2.87~\mathrm{GHz}$2-transformed frequencies. The reported LR range was approximately 100 to 100,000, which the paper presents as moderate to very strong evidential support depending on haplogroup and SNV rarity. The implementation is distributed as the open-source R package mitofreq, with a Shiny app that allows custom TLHG frequencies (Andersen et al., 15 Jan 2026).

The principal limitations are internal to the method’s assumptions. TLHG assignment must be correct; the chosen SNV must be informative for rarity; only homoplasmic positions are considered; indels are ignored; and database composition remains population-sensitive. The paper explicitly notes that geographic structure matters and that lineage-related maternal relatives are not appropriate “random” alternatives in evidential interpretation.

3. Triplet-frequency composition and mitochondrial phylogenomics

A precursor usage of the MitoFREQ idea appears in work on the triplet-frequency composition of mitochondrial genomes (Sadovsky, 2014). The object of analysis is a frequency dictionary of words of length $2.87~\mathrm{GHz}$3, with the paper focusing on triplets, $2.87~\mathrm{GHz}$4. For a word $2.87~\mathrm{GHz}$5, the frequency is given as $2.87~\mathrm{GHz}$6, where $2.87~\mathrm{GHz}$7 is the word count and $2.87~\mathrm{GHz}$8 is sequence length. Because there are $2.87~\mathrm{GHz}$9 triplets and their frequencies sum to 1, each mitogenome lies in a 63-dimensional simplex. The authors reduced the dimension to 63 by dropping GCG, the triplet with the smallest standard deviation, $W_3$0.

The dataset consisted of mitochondrial genomes from the EMBL-bank March 2011 release. Although that release contained about $W_3$1 entries, the final analysis used 1,132 animal mitochondrial genomes after restricting the database to orders or clades represented by at least five species. The resulting sample contained 988 Chordata and 144 Arthropoda entries. Reported chordate subdivisions included Batrachia 51, Chondrostei 5, Crocodylidae 7, Cryptodira 25, Dinosauria 94, Eutheria 193, Gymnophiona 16, Metatheria 18, Neopterygii 500, and Squamata 78. “Junk” symbols were removed by concatenating fragments separated by those symbols, and no further class-separability pretesting was performed.

Clustering was unsupervised $W_3$2-means in Euclidean distance on the 63-dimensional triplet-frequency vectors, implemented in ViDaExpert. The methods mention 2-class, 3-class, and 4-class clustering, although the reported results focus on 2 and 3 classes. In the 2-class case, the solution was very stable across $W_3$3 runs, with almost all runs producing classes of size 154 and 978; only 13 runs produced a degenerate singleton class. The dominant pattern was that 142 Arthropoda genomes always formed one class, with only two exceptions: Reticulitermes flavipes (EF206314) and Gampsocleis gratiosa (EU527333). The paper also notes that turtles and fossils fall in the same class, and that Mammalia and Neoptera are unexpectedly grouped together in the same class.

The 3-class case was similarly stable across $W_3$4 runs. Only three abundance patterns occurred: $W_3$5 entries in 18 cases, $W_3$6 in 854 cases, and $W_3$7 in 136 cases. No genomes changed cluster assignment across runs. The clade-to-cluster counts were strongly structured: Actinopterygii 464 / 46 / 0, Amphibia 40 / 17 / 8, Archosauria and Lepidosauria 1 / 176 / 0, Mammalia 0 / 1 / 211, Neoptera 0 / 4 / 139, and Testudines 0 / 25 / 0. Amphibia was the most heterogeneous clade, occupying all three clusters; within it, Anura placed all 13 genomes in the second class, whereas Caudata placed 7 in the third class and 2 in the second.

The paper does not report a formal correlation coefficient or $W_3$8-value. Instead, regularity is quantified through cluster stability across $W_3$9 runs, low numbers of escaping genomes, and the extreme bias of clade counts across clusters. In the two-class case, the escaped-genome rate for Mammalia and Neoptera was said to be $T_1$0 and $T_1$1, respectively. For the chordate subset, 976 genomes formed one class with only 12 escaping. The conclusion is that mitogenomes are not randomly distributed in triplet-frequency space; rather, proximity in mitochondrial frequency space mirrors morphological or taxonomic proximity, which the authors interpret as evidence of strong co-evolution of mitochondrial and somatic genomes (Sadovsky, 2014).

4. Oscillatory oxidative dynamics as a frequency-based mitochondrial regime

A different but conceptually related usage of MitoFREQ arises from single-mitochondrion photostimulation studies that expose repeated oxidative and electrical events during reversible morphology changes (Wang et al., 2015). The experimental system used a confocal microscope coupled to a homemade Ti:Sapphire femtosecond laser and a Yb-doped fiber laser. The principal stimulation condition was a Ti:Sapphire femtosecond laser at 810 nm, 75 fs pulse width, 80 MHz repetition rate, typically 6 mW at the cell, and often 0.1 s duration. The beam was focused by a $T_1$2 water-immersion objective with $T_1$3, producing a focal spot of less than $T_1$4 in diameter and allowing single-mitochondrion targeting.

Under 810 nm, 6 mW, 0.1 s stimulation in HeLa cells, most targeted mitochondria fragmented rapidly and then recovered their tubular morphology. Specifically, 56 of 59 fragmented and 3 of 59 swelled. Fragmentation became significant in about 10 s after photostimulation, and recovery occurred in about 40 s on average; 21 of 25 fragmented mitochondria recovered within 100 s. Neighboring mitochondria were largely unaffected. Structural discontinuity was verified with mitochondrial matrix-targeted photoactivatable GFP: in fragmented mitochondria, activated PAGFP did not diffuse to the dark portion, whereas control mitochondria showed normal diffusion, and diffusion returned after recovery. The same mitochondrion could be stimulated again after recovery, with a second fragmentation-recovery cycle resembling the first.

The mechanistic centerpiece was the appearance of mROS flashes. The reported sequence was that mROS appeared first at the laser focal spot, spread through the whole mitochondrion, and was eventually scavenged. In one representative event, mROS was detected around 8.8 s, spread through the mitochondrion by 12.1 s, and was scavenged by 42 s. The paper repeatedly refers to series or repeated mROS flashes within the 100 s post-stimulation window, although it does not assign a formal frequency in Hz. NAC at 5 mM reduced mROS and strongly inhibited fragmentation, whereas TBHP at $T_1$5 increased oxidative stress and increased the fraction of mitochondria showing fragmentation or swelling.

Membrane-potential dynamics were monitored with TMRM and mROS with DihydroRhodamine 123. Each MMP oscillation corresponded exactly to an mROS flash, and the paper states that 20 independent experiments supported this relationship. Transient opening of the mitochondrial permeability transition pore was assayed with calcein/CoCl$T_1$6. At 6 mW there was no fluorescence quenching, indicating no permanent mPTP opening during restorable fragmentation, whereas at 20 mW calcein quenching indicated permanent or long-time mPTP opening and unrecoverable damage. Cyclosporin A at $T_1$7 increased the percentage of fragmented or swollen mitochondria after photostimulation and significantly slowed recovery. Under CsA, very weak stimulation of 0.5 mW for 30 s could slowly accumulate mROS to a level comparable to 3 mW for 0.2 s without CsA; without CsA, ultra-weak stimulation up to 1 mW for more than 60 s elicited no response.

The paper’s interpretation is that tightly focused femtosecond-laser photostimulation induces localized mROS generation, the rising mROS promotes fragmentation, transient mPTP openings help scavenge or release excess mROS, and the organelle then recovers once oxidative burden is reduced. At higher powers, the regime changes: 18 mW produced irreversible fragmentation, 30 mW caused severe disruption, and 20 mW was associated with Bax accumulation, cytochrome $T_1$8 release, and non-restorable damage. This supports a distinction between a low-power homeostatic regime of repeated oxidative events and a high-power destructive regime (Wang et al., 2015). A plausible implication is that MitoFREQ, in this usage, refers not to sequence frequencies but to a temporally patterned mitochondrial response mode.

5. Spectral and photothermal mitochondrial fingerprinting

Another frequency-like extension of the MitoFREQ idea appears in two-photon photothermal microscopy, where mitochondrial state is inferred from pump-wavelength-dependent intrinsic chromophore signatures rather than fluorescence labels (Hai et al., 15 Apr 2026). The technique measures optical absorption of intracellular chromophores such as NADH and FAD by detecting localized thermal transients generated through two-photon absorption. The pump is a tunable NIR beam spanning roughly 680–960 nm, the probe is generated by frequency doubling a 1045 nm beam to 522.5 nm, and the beams are scanned with galvanometric mirrors and focused with a $T_1$9, 1.2 NA water-immersion objective. Transmitted probe light is collected by a 1.4 NA condenser, and the signal is recorded with a photodetector and lock-in amplifier. The pump is modulated by an acousto-optic modulator at 125 kHz and 5% duty cycle.

The imaging principle is a three-step process: two-photon absorption by the chromophore, nonradiative relaxation producing localized heat, and transient probe modulation through the thermal lens. The simulation section uses the heat equation

$x$0

with $x$1 as temperature, $x$2 as thermal diffusivity, $x$3 as absorbed energy density, $x$4 as density, and $x$5 as heat capacity. Because NADH and FAD have fluorescence yields of about 2–4% and 8–10%, respectively, most absorbed energy becomes heat. The paper states that photothermal sensing can provide roughly 10–20× signal boost relative to autofluorescence in these low-efficiency biomolecules.

Using the standard limit-of-detection formula

$x$6

the reported 2PPT LODs were $x$7 for NADH and $x$8 for FAD, compared with 17.5 and $x$9, respectively, for two-photon autofluorescence. Thus 2PPT was about 20× more sensitive for NADH and about 10× more sensitive for FAD. In live SK-OV-3 cell mitochondria, pump-wavelength spectroscopy showed a dip around 750 nm and a relatively constant signal from 770–820 nm. Phasor analysis placed mitochondrial spectra closest to FAD, then NADH, and least similar to cytochrome $n$0, supporting the claim that the mitochondrial contrast arises mainly from metabolic coenzymes rather than heme proteins.

The paper also links spectral signatures to morphology and perturbation. It distinguishes tubular or rod-shaped mitochondria from oval or blob-shaped mitochondria, noting that the former are associated with healthier, energy-producing states, whereas the latter are associated with stress, aging, disease, or cell death. In SK-OV-3 cells, 2PPT visualized both shapes with a signal-to-background ratio of 14.27 dB, compared with about 3.36 dB for autofluorescence. The reported SBR formula was

$n$1

NAC treatment reduced NADH by about 33% and FAD by about 37%, and NAC-treated cells showed fewer rod-shaped mitochondria and more oval structures. Under starvation, tubular mitochondria showed increased NADH at 24 h and decline at 48 h, while oval mitochondria showed significant NADH decrease at both time points and a similar downward trend in FAD. In ovarian cancer spheroids grown to about $n$2 diameter and imaged through about $n$3 depth, cisplatin treatment significantly decreased NADH and significantly increased FAD, which the authors interpret as a shift toward a more oxidative metabolic phenotype (Hai et al., 15 Apr 2026).

The paper does not explicitly define this platform as MitoFREQ. Nevertheless, its emphasis on reproducible spectral fingerprints, morphology-sensitive metabolic contrast, and single-organelle response profiling makes it a close conceptual analogue of a MitoFREQ-like mitochondrial state readout.

6. Frequency-sensitive relaxometry and organelle-resolved tracking

A further extension is provided by fluorescent nanodiamond relaxometry, which couples spatial tracking to a mitochondrial frequency-sensitive measurement of local magnetic-noise fluctuations (Su et al., 10 Dec 2025). FNDs containing NV centers act as nanoscale sensors whose longitudinal spin relaxation time $n$4 changes in the presence of paramagnetic species such as reactive oxygen species and some metal ions. The paper models the relaxometry signal as

$n$5

where $n$6 is normalized fluorescence after dark time $n$7, $n$8 is fluorescence contrast, $n$9 is equilibrium fluorescence intensity, and $\hat p = x/n$0 is the relaxation time. The relaxation rate $\hat p = x/n$1 is proportional to the spectral density of magnetic noise near the NV resonance frequency of about $\hat p = x/n$2; shorter $\hat p = x/n$3 indicates stronger local paramagnetic noise.

The instrumentation is a home-built multimode microscope supporting confocal imaging, wide-field imaging, and $\hat p = x/n$4 relaxometry. It uses 532 nm excitation for FND/NV optical pumping and readout, 488 nm for MitoTracker Green, and 365 nm for DAPI. Wide-field relaxometry uses an APD/PD, confocal relaxometry uses a SPAD, and multichannel imaging uses an sCMOS camera. To integrate tracking with sensing, the authors localize the FND before each $\hat p = x/n$5 measurement, detect defocus by intensity thresholding, and perform automated triaxial line scans in $\hat p = x/n$6, $\hat p = x/n$7, and $\hat p = x/n$8 with Gaussian fitting to recover coordinates. An example trajectory covered $\hat p = x/n$9 over about 140 min while $$\hat p_{\text{pooled}} = \frac{x_{\text{Helix}} + x_{\text{gnomAD}}}{n_{\text{Helix}} + n_{\text{gnomAD}}}.$$0 was measured concurrently.

For mitochondrial targeting, the paper conjugates a mitochondria-targeting sequence,

$$\hat p_{\text{pooled}} = \frac{x_{\text{Helix}} + x_{\text{gnomAD}}}{n_{\text{Helix}} + n_{\text{gnomAD}}}.$$1

to generate FND-MTS. DLS shows a zeta-potential shift from negative to positive and an increase in hydrodynamic diameter, and confocal imaging shows substantially improved colocalization with MitoTracker Green. The mitochondrial targeting result is quantified by Manders coefficients, with $$\hat p_{\text{pooled}} = \frac{x_{\text{Helix}} + x_{\text{gnomAD}}}{n_{\text{Helix}} + n_{\text{gnomAD}}}.$$2 for FND-MTS compared with $$\hat p_{\text{pooled}} = \frac{x_{\text{Helix}} + x_{\text{gnomAD}}}{n_{\text{Helix}} + n_{\text{gnomAD}}}.$$3 for bare FNDs. For nuclear targeting, the authors use TAT $$\hat p_{\text{pooled}} = \frac{x_{\text{Helix}} + x_{\text{gnomAD}}}{n_{\text{Helix}} + n_{\text{gnomAD}}}.$$4 and NLS $$\hat p_{\text{pooled}} = \frac{x_{\text{Helix}} + x_{\text{gnomAD}}}{n_{\text{Helix}} + n_{\text{gnomAD}}}.$$5, verifying localization by z-stacks, 3D projections, and TEM.

The mitochondrial metabolic perturbation experiment uses CCCP at $$\hat p_{\text{pooled}} = \frac{x_{\text{Helix}} + x_{\text{gnomAD}}}{n_{\text{Helix}} + n_{\text{gnomAD}}}.$$6, an uncoupler of oxidative phosphorylation. After CCCP addition, $$\hat p_{\text{pooled}} = \frac{x_{\text{Helix}} + x_{\text{gnomAD}}}{n_{\text{Helix}} + n_{\text{gnomAD}}}.$$7 of mitochondrial-targeted FND-MTS decreases significantly, consistent with increased paramagnetic noise under oxidative stress. Cell-free CCCP controls showed no significant change, arguing against a direct chemical artifact. The study also compared $$\hat p_{\text{pooled}} = \frac{x_{\text{Helix}} + x_{\text{gnomAD}}}{n_{\text{Helix}} + n_{\text{gnomAD}}}.$$8 across intracellular compartments: nucleus $$\hat p_{\text{pooled}} = \frac{x_{\text{Helix}} + x_{\text{gnomAD}}}{n_{\text{Helix}} + n_{\text{gnomAD}}}.$$9, cytoplasm $$P(\text{mitogenome}) \leq P(\text{TLHG})\,P(X_k \mid \text{TLHG}),$$0, and complete media $$P(\text{mitogenome}) \leq P(\text{TLHG})\,P(X_k \mid \text{TLHG}),$$1. No significant difference was found between nucleus and cytoplasm at baseline, but both were much shorter than in complete media, indicating a substantially noisier intracellular magnetic environment.

The authors emphasize stability and reversibility. $$P(\text{mitogenome}) \leq P(\text{TLHG})\,P(X_k \mid \text{TLHG}),$$2 remained essentially stable in water, MES buffer, PBS, and 1 M NaCl; in fixed HeLa cells over 1 h there was no significant monotonic drift; and switching between GdCl$$P(\text{mitogenome}) \leq P(\text{TLHG})\,P(X_k \mid \text{TLHG}),$$3 and DTPA yielded a fully reversible response over 10 cycles. The stated limitations include intrinsic heterogeneity of FNDs, imperfect targeting efficiency, the relatively large 40 nm particle size, brightness limitations, and incomplete specificity because many paramagnetic species can alter $$P(\text{mitogenome}) \leq P(\text{TLHG})\,P(X_k \mid \text{TLHG}),$$4. In conceptual terms, this work supports a MitoFREQ interpretation in which mitochondrial state is read out through frequency-sensitive relaxometry rather than through sequence composition or optical spectra.

7. Synthesis, significance, and boundaries

Across the supplied literature, MitoFREQ does not identify a single canonical object. Instead, it spans at least four analytically distinct regimes: conditioned lineage frequencies in forensic mtDNA interpretation, triplet-frequency composition in mitogenome phylogenomics, oscillatory oxidative dynamics in single-organelle perturbation experiments, and spectral or relaxometric mitochondrial state readouts in label-free or nanosensor imaging (Andersen et al., 15 Jan 2026, Sadovsky, 2014, Wang et al., 2015, Hai et al., 15 Apr 2026, Su et al., 10 Dec 2025). The common denominator is not the experimental platform but the use of frequency-structured information to represent mitochondrial identity, rarity, or state.

The significance of the forensic formulation is methodological and evidential. It converts TLHG frequencies and within-lineage SNV frequencies from HelixMTdb and gnomAD into conservative mitogenome frequency estimates and operational likelihood ratios, while remaining usable for partial or degraded profiles. The significance of the triplet-frequency formulation is evolutionary: a purely sequence-compositional embedding of mitogenomes in 63-dimensional space shows strong concordance with morphological taxonomy. The significance of the photostimulation, 2PPT, and relaxometry formulations is functional: each treats mitochondrial state as something that can be read out through repeated oxidative events, chromophore-specific spectral structure, or gigahertz-scale magnetic-noise sensitivity.

Several boundaries are equally clear. The forensic MitoFREQ method is a defined statistical procedure; the other usages are conceptual analogues rather than formal implementations. The triplet-frequency study does not report a formal correlation coefficient or $$P(\text{mitogenome}) \leq P(\text{TLHG})\,P(X_k \mid \text{TLHG}),$$5-value. The photostimulation study does not provide a formal oscillation frequency in Hz. The 2PPT study explicitly notes that the method is not inherently molecule-specific and therefore relies on spectroscopy and phasor analysis to identify chromophore sources. The relaxometry study explicitly notes that $$P(\text{mitogenome}) \leq P(\text{TLHG})\,P(X_k \mid \text{TLHG}),$$6 is affected by many paramagnetic species, not just ROS. These limitations matter because they mark the point at which “frequency” is a powerful descriptor but not a complete mechanistic explanation.

A final misconception is that MitoFREQ necessarily implies one modality, such as mitochondrial DNA analysis or optical frequency-domain imaging. The literature here indicates otherwise. A more precise statement is that MitoFREQ, in current usage, refers either to a specific forensic estimator for mitogenome population frequencies or, more broadly, to mitochondrial analytic schemes in which frequency statistics, spectral signatures, oscillatory dynamics, or frequency-sensitive physical observables provide the primary coordinate system for inference.