henhaobio
Sign in

bioRxivBiophysicsPreprint

QuantiTrack: A unified software to study protein dynamics in living cells

David A BalliD, Kaustubh WaghiD, Diana A. StavrevaiD, Le Hoang, R. Louis SchiltziD, Raj ChariiD, Razi Raziuddin, Davide MazzaiD, Arpita UpadhyayaiD, Gordon L. HageriD, Tatiana KarpovaiD

PDFDOICodeData
In-depth full-text digest available · Jump to digest ↓

Abstract

Linking the spatiotemporal dynamics of proteins in live cells to biological function is a fundamental challenge in biology. Single molecule tracking (SMT) has emerged as a powerful technique to investigate protein dynamics at the single molecule level. However, SMT analysis often requires expertise in biophysical modeling and programming, and integrating results from different analyses can be challenging. To address these barriers, we developed QuantiTrack: a MATLAB-based SMT analysis software with a simple graphical user interface. This provides a much-needed end-to-end solution where a user can load a movie, detect and track single molecules, and perform complementary downstream analyses within a standardized workflow. QuantiTrack includes quantitative metrics for selecting detection and tracking parameters and troubleshooting experimental design, and includes a detailed step-by-step User Guide. We used simulations to demonstrate how signal intensity, labeling density, and motion blur affect detection and tracking fidelity. Using multi-state simulations, we further benchmarked complementary methods to identify distinct mobility states from heterogeneous trajectory populations. Finally, we applied QuantiTrack to real experimental data where we address how the glucocorticoid receptor (GR), a hormone-regulated transcription factor, responds to treatment and washout of its cognate hormone. Hormone washout results in rapid (in minutes) downregulation of GR target genes to basal levels. By integrating complementary analyses within QuantiTrack, we showed that hormone washout substantially reduced the bound fraction of GR, its occupancy in the mobility state associated with GR activation, and dwell times. Together, these analyses showcase QuantiTrack as an integrated platform for extracting biologically meaningful measurements from single molecule trajectories.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 3 (15 Sept 2026) · CC0 1.0

QuantiTrack, a MATLAB GUI, unifies single-molecule detection, tracking, quality control and mobility analyses; for glucocorticoid receptor, hormone washout cut bound fraction, dwell times and low-mobility occupancy.

Single molecule tracking (SMT) analysis usually means passing the same movies through several separate programs, demanding programming and biophysical modeling skills and making results hard to compare across tools. QuantiTrack offers one documented GUI with quantitative quality-control metrics for picking detection and tracking parameters and a shared track table feeding Spot-On, Richardson-Lucy, pEMv2, MSD, turning-angle and dwell-time analyses.

Key results

  1. SuggestedWithout ground truth, the Kolmogorov-Smirnov (KS) separation between spot and local-background intensity distributions plateaus near the detection threshold that maximizes detection fidelity. This gives a practical rule for choosing the threshold.

    sptPALMsim Brownian motion (D = 1 μm^2/s) at intensities 25, 75, 150: Jaccard similarity coefficient (JSC) rose with intensity and peaked for intensity 150 at threshold 75 with hpass = 4 px; the KS-vs-threshold curve reaches its inflection where JSC peaks (TP/FP/FN counts had error bars over 10 movies). Applied to GR movies, the rule gave Th = 200. · Using the quality control metrics within…; Glucocorticoid receptor target genes and GR…; Detection and tracking fidelity; Figure legends

  2. SuggestedMotion blur and particle crowding limit what can be detected and tracked. At 10 ms exposure and 12 ms intervals, fast diffusers lose detection fidelity, and their tracks break into fragments rather than linking to the wrong particles.

    Motion blur degraded localization precision from 24 nm to 41 nm, JSC from 0.72 to 0.13 and inflection KS from 0.97 to 0.7; JSC halved at D = 7.5 μm^2/s. With MJ = 16 px, D ≤ 5 μm^2/s was recovered accurately. At D = 5 μm^2/s, going from 10 to 20 particles/frame cut track-level JSC; D = 1 μm^2/s tolerated 30 particles/frame. · The effect of motion blur on detection fidelity; Optimizing tracking fidelity; Detection and tracking fidelity

  3. SuggestedOn simulated two-state mixtures, ensemble MSD could not separate the states. Spot-On, Richardson-Lucy (RL) MSD-distribution analysis and pEMv2 each recovered both states; Spot-On underestimated Dfast and pEMv2 overestimated the slow-state fraction, while RL agreed with ground truth.

    Dslow = 0.1 and Dfast = 2.5 μm^2/s at 20/80, 50/50 and 80/20 slow/fast; 120 movies per mixture. Spot-On estimates agreed between QuantiTrack and ground-truth tracks, though Dfast was underestimated in both; RL gave bimodal MSD distributions; pEMv2 found two states with correct D but overestimated the slow fraction in all three mixtures. · Deciphering mobility states in multicomponent…; Complementary analyses for resolving mobility…; Optical-dynamic simulations; Figure legends

  4. ShownIn MCF-7 cells, hormone washout returned two GR target genes to baseline. Fast SMT in an MCF-7 line with endogenously HaloTagged GR also showed a clear drop in the GR bound fraction.

    PER1 and SGK1 were back to baseline at 60 min after washout of 1.2 μM Hcort but higher at 60 min under continued treatment. Three-state (and two-state) Spot-On showed a significant bound-fraction reduction; cells/tracks 130/93,836 (treatment) vs 133/108,696 (washout). Mouse ESC references: 0.90 (H2B) to 0.16 (Halo-3xNLS), two-state model. · Glucocorticoid receptor target genes and GR…; RNA-qPCR experiments; Figure legends

  5. ShownSlow SMT (200 ms intervals) with H2B-based photobleaching correction showed that washout shortens GR dwell times. It also depletes the lowest-mobility pEMv2 state, which is linked to transcriptional activity.

    Photobleaching rate taken from a triple-exponential fit to H2B survival (42/33,173 cells/tracks). pEMv2 found 3 GR states; after bound-fraction normalization, washout gave a ∼75% reduction in states 1 and 2 (sub-tracks 18,811 treatment vs 3,971 washout). RL analysis showed 3 to 4 groups at lag 0.8 s in both conditions. · Hormone washout results in shorter GR dwell…; Photobleaching correction; Figure legends

Methods, data and reproducibility

MATLAB GUI that localizes spots by fixed-width Gaussian PSF fitting, links them by nearest-neighbor or u-track LAP tracking, reports QC metrics (KS statistic, SNR, jump and second-neighbor histograms), then runs Spot-On, Richardson-Lucy, pEMv2, MSD, turning-angle and photobleaching-corrected dwell-time analyses; benchmarked on sptPALMsim simulations.

  • sptPALMsim single-state simulations (Brownian and fractional Brownian motion) varying signal intensity, D (1 to 20 μm^2/s), motion blur and labeling density · n = 10 movies per parameter set
  • sptPALMsim two-state slow/fast mixtures (Dslow 0.1, Dfast 2.5 μm^2/s; 20/80, 50/50, 80/20 proportions) · n = 120 movies per mixture
  • Fast SMT (12 ms frames) of endogenous Halo-GR in a clonal MCF-7 knock-in line, hydrocortisone treatment vs washout · human (MCF-7 cells) · n = 130/93,836 (treatment), 133/108,696 (washout) cells/tracks · 10.5281/zenodo.22756685 (representative movies and trackTables)
  • Slow SMT (200 ms interval, 10 ms exposure) of Halo-GR in MCF-7, treatment vs washout · human (MCF-7 cells) · n = 96/58,239 (treatment), 70/32,183 (washout) cells/tracks · 10.5281/zenodo.22756685 (representative movies and trackTables)
  • Slow SMT of endogenous H2B-Halo in MCF-7, used as the photobleaching reference · human (MCF-7 cells) · n = 42/33,173 cells/tracks
  • Public fast SMT (7.5 ms frames) of H2B-Halo, Halo-CTCF, Halo-SOX2 and Halo-3xNLS provided by Hansen and colleagues · mouse (embryonic stem cells)
  • RNA-qPCR of PER1 and SGK1 in MCF-7 after 1.2 μM hydrocortisone treatment or washout · human (MCF-7 cells)
Methods & tools
QuantiTrack (MATLAB GUI, built on the TrackRecord framework), sptPALMsim optical-dynamic SMT simulator, Nearest-neighbor tracker with maximum-jump constraint, u-track Linear Assignment Problem (LAP) tracker, Spot-On (multi-lag jump-distance state fitting), Richardson-Lucy (RL) MSD-distribution analysis, pEMv2 (perturbation expectation-maximization v2), Kinetic model of Mazza et al., Kolmogorov-Smirnov statistic and SNR QC metrics, CRISPR-Cas9 HaloTag knock-in (lentiCRISPR v2), FACS, Western blot, HILO microscopy with JF549 and JFX650 HaloTag ligands, RNA-qPCR (PER1, SGK1)
Software & environment
MATLAB-based GUI (MATLAB version not stated); simulations generated with the sptPALMsim Python package; Spot-On parameters listed (e.g., TimeGap = 12, dZ = 0.7).
Reproducibility gaps
Requires proprietary MATLAB; Zenodo holds only representative GR movies and trackTables, not all raw data; no links for simulation scripts or the reused mouse ESC dataset; MATLAB version not given.
Limitations (6)
  • Built in proprietary MATLAB, so users without a licence cannot run it; vbSPT, TrackIt, state-array SPT and other complementary analyses are not included and need QuantiTrack trajectories to be reformatted. · stated by the authors
  • SMT cannot resolve site-specific binding, so 'bound' operationally means a stopped spot; bound fractions and dwell times are not equivalent to specific chromatin-site occupancy. · stated by the authors
  • Estimates depend on acquisition settings, analysis parameters and model choice; absolute values may change, and simulation-derived limits (e.g., D up to 5 μm^2/s recoverable, tolerable labeling densities) apply only to the simulated imaging conditions. · stated by the authors
  • Simulations are idealized and omit the kinetic complexity of cellular data, and turning-angle anisotropy is flagged as the most exploratory module, needing further modeling before quantitative conclusions. · stated by the authors
  • GR results come from one validated clone per knock-in MCF-7 line, and the text gives no named statistical test or p-values for the reported 'significant' drops in bound fraction and dwell time. · noted in this summary
  • In simulations pEMv2 overestimated slow-state fractions and Spot-On underestimated Dfast, so absolute GR state fractions (such as the ∼75% decrease in states 1 and 2) may carry method-specific bias; relative comparisons are safer. · noted in this summary

Read this if you run live-cell single molecule tracking and want one GUI to set detection and tracking parameters and compare bound fraction, mobility state and dwell time analyses.

Citations
0

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISSimon '26Kuhn '20McSwiggen '23Simon '22Heckert '21Ito '23
63% semantic matchBiophysics
François Simon, Chris H. Wiggins, Lucien E. Weiss·bioRxiv·Jan 24, 2026