Detecting active Lévy particles using differential dynamic microscopy

This paper extends differential dynamic microscopy to detect active Lévy particles by validating the method on synthetic data and demonstrating its application to experimental data, which reveals that *E. coli* does not exhibit Lévy walk signatures while *E. gracilis* does.

The Gist

Imagine you are trying to understand how tiny, single-celled organisms move through a drop of water. Some swim in a very predictable way: they go straight for a bit, stop, spin around randomly, and go straight again. Scientists call this "Run-and-Tumble." It's like a person walking down a hallway, stopping every few seconds to spin in a circle, and then picking a new direction.

But other organisms might move differently. They might go straight for a short time, then take a very long, straight path before turning. This is called a "Lévy walk." It's like a hiker who usually takes short steps but occasionally decides to sprint across a whole field without stopping. Detecting these rare, long "sprints" is incredibly hard because you have to watch the organism for a long time and over a large area to see the pattern.

This paper introduces a new, powerful way to spot these "sprints" without having to track every single cell individually. Here is the breakdown of their discovery:

The Problem: The "Needle in a Haystack"

To prove an organism is doing a Lévy walk, you need to see its movement patterns across many different sizes and times. If you only look at a tiny patch of water, you might miss the long sprints entirely. Traditional methods often require tracking individual cells one by one, which is slow and misses the big picture.

The Solution: A "Group Photo" Approach

The authors use a technique called Differential Dynamic Microscopy (DDM). Instead of tracking one cell, imagine taking a video of a crowded dance floor.

• Old way: You try to follow one specific dancer to see their steps.
• This paper's way: You look at the entire video and measure how much the "blur" of the crowd changes over time.

They analyze the "flicker" of the whole group. By looking at how the light patterns shift and blur, they can mathematically reconstruct the movement statistics of the entire crowd at once. It's like listening to the roar of a stadium crowd to figure out if the fans are cheering in short bursts or long, sustained waves, without needing to hear every individual voice.

The Discovery: Two Different Dancers

The team applied this method to two types of microorganisms:

1. E. coli (The Predictable Dancer):
   They looked at E. coli bacteria. Even though some theories suggested they might take long, random sprints (Lévy walks), the data showed they are actually very consistent. They run, tumble, and run again in a predictable rhythm. The "long sprints" were just an illusion caused by looking at the data in the wrong way. They are classic "Run-and-Tumble" walkers.

2. E. gracilis (The Sudden Sprinter):
   They then looked at a type of algae called Euglena gracilis. This one is different. The data clearly showed that these cells do take those rare, very long straight paths. They are true "Active Lévy Particles." The new method successfully caught the signature of these long sprints, proving they exist in this organism.

The Catch: Speed Variability

The paper also found a limitation. If the organisms change their speed too much (some swim fast, some slow, and they switch randomly), it becomes harder to spot the Lévy pattern. It's like trying to hear a specific rhythm in a song where everyone is playing at a different tempo; the pattern gets muddy. The method works best when the swimmers are relatively consistent in their speed.

The Bottom Line

This paper provides a new "high-throughput" (fast and efficient) tool for scientists. It allows them to distinguish between organisms that move in short, random bursts and those that take rare, long-distance sprints. By looking at the "blur" of the whole group rather than tracking individuals, they confirmed that E. coli is a steady, short-step walker, while E. gracilis is a master of the long, unpredictable sprint.

Technical Summary

Technical Summary: Detecting Active Lévy Particles Using Differential Dynamic Microscopy

Problem Statement
Detecting Lévy flights in biological systems, particularly among microorganisms, remains a significant experimental challenge. The core difficulty lies in the necessity of acquiring data across spatiotemporal scales spanning multiple orders of magnitude to reliably extract power-law scaling, a hallmark of Lévy walks. While Differential Dynamic Microscopy (DDM) has emerged as a high-throughput method capable of accessing scales across two orders of magnitude simultaneously, its application to self-propelled particles with algebraic run-time distributions (active Lévy particles, or ALPs) had not been fully established. Specifically, it was unclear whether DDM could robustly distinguish ALPs from standard run-and-tumble particles (RTPs), which exhibit exponentially distributed run times, given experimental constraints such as speed variability and finite observation windows.

Methodology
The authors extend the DDM framework to characterize active Lévy particles. The methodology involves three primary components:

1. Theoretical Modeling: The study defines a paradigmatic model for ALPs where the run-time distribution follows a Lomax distribution (an algebraic tail truncated at small times), contrasting with the exponential distribution of RTPs. Using renewal theory, the authors derive the Intermediate Scattering Function (ISF), $f(k, \tau)$, for ALPs in both 2D and 3D. They analyze the asymptotic behavior of the ISF, specifically examining the decay rate $\lambda(k)$ and the scaling of the mean-squared displacement (MSD).
2. Simulation Validation: To validate the detection protocol, the authors generate synthetic imaging data for both ALPs and RTPs. They simulate particle trajectories with varying power-law exponents ($\mu$), persistence lengths, and speed variabilities ($\sigma_v$). The synthetic data is processed via DDM to calculate ISFs, which are then fitted against both ALP and RTP theoretical models to test the protocol's ability to recover ground-truth parameters and distinguish between the two models.
3. Experimental Application: The protocol is applied to existing and new experimental data:
   • Escherichia coli (using published data from Ref. [14]).
   • Euglena gracilis (using new experimental data under varying light intensities).
     The experimental ISFs are fitted simultaneously across multiple wavenumbers ($k$) using both the ALP and RTP models to determine which model best describes the motion.

Key Contributions and Results

• Asymptotic Signatures of ALPs: Theoretical analysis reveals that while RTPs and ALPs share similar behavior at small length scales (near the persistence length $\ell_p$), they diverge significantly at larger scales. For ALPs with $2 < \mu < 3$, the ISF exhibits persistent oscillations and decays according to a scaling law $k^{\mu-1}\tau$ at large length scales (up to $\sim 10\ell_p$). In contrast, RTPs converge to an exponential decay ($k^2\tau$) much more rapidly.
• Impact of Speed Variability: The study demonstrates that speed variability ($\sigma_v$) suppresses the oscillatory features in the ISF. Reliable distinction between ALPs and RTPs requires moderate speed variability ($\sigma_v \lesssim 0.2\bar{v}$). If variability is high, the differences between the models diminish, necessitating access to even larger length scales for detection.
• Finite Tumble Time: The analysis shows that a finite tumble time (whether exponentially or algebraically distributed) acts primarily as a time rescaling factor ($p_R$) on large scales, provided the mean tumble time is finite. It does not qualitatively alter the asymptotic scaling of the ISF.
• Simulation Validation: The protocol successfully recovers the power-law exponent $\mu$ and persistence length $\tau_R$ from synthetic data when speed variability is moderate. The ALP model provides a better fit for ALP data, while the RTP model fails to capture the long-range oscillations. Conversely, when fitting RTP data, the ALP model tends to overfit, yielding unstable or unphysically large $\mu$ values.
• Experimental Findings:
  • E. coli: The experimental ISFs are best described by the RTP model. While the ALP model can fit the data, it results in large estimation errors and unstable exponents (often $\mu \gg 3$), indicating overfitting. This supports the conclusion that E. coli exhibits diffusive behavior on the observed scales rather than Lévy walks.
  • E. gracilis: The ISFs of E. gracilis exhibit strong oscillations at small $k$ that are well-captured by the ALP model but poorly fitted by the RTP model. The fitted exponents are stable and consistently less than 3 ($\mu \approx 2.05 - 2.30$), confirming that E. gracilis behaves as active Lévy particles on length scales up to $10^3$ µm.

Significance and Claims
The paper claims to provide a robust, high-throughput protocol for detecting active Lévy particles using DDM. The authors emphasize that the detection of Lévy walks relies critically on accessing length scales an order of magnitude larger than the persistence length ($\sim 10\ell_p$) to observe the distinct asymptotic scaling of the ISF. The study validates that DDM is a versatile tool for this purpose, provided experimental conditions (specifically speed variability) allow for the resolution of the characteristic oscillatory decay. The application to E. coli and E. gracilis serves as a concrete demonstration, resolving previous ambiguities by showing that E. coli follows RTP dynamics on the measured scales, while E. gracilis exhibits signatures of active Lévy walks. The work underscores the importance of multi-scale analysis in distinguishing between different modes of active matter transport.