Biology subjects

Mulla, Y.

Publications and source records attributed to Mulla, Y..

9 recordsLinked to original sources

bioRxiv2026

Degree bias predicts, outperforms and improves machine learning phage-host interaction predictions

Machine learning (ML) models predicting phage-host interactions are widely used to prioritize phages for phage therapy. The Area Under the Receiver Operating Characteristic (AUROC) metric is standard for model evaluation, with reported high scores (0.7 - 0.95) suggesting strong predictive power. However, because phage-bacterium networks are highly structured, these scores can be inflated by "degree bias", the baseline likelihood of infection driven purely by average phage infectivity and bacterial susceptibility. Here, we show that an interaction-naive model based solely on network degrees achieves higher global AUROC scores than five published ML models across ten datasets, indicating that global AUROC primarily reflects network structure rather than predicted interactions. Using randomized degree-conserved networks and degree-stratified evaluation, we isolate true interaction predictions and demonstrate that current ML model performance is far more modest than global AUROC suggests. Finally, we demonstrate that regressing out degree bias from training data improves interaction predictability and boosts global and stratified performance. Our work provides a critical framework for model evaluation and a straightforward route to improve ML-guided phage selection.

microbiology↗

bioRxiv2026

Phage phenotyping by measuring plaque expansion dynamics

Bacteriophages exhibit staggering genomic diversity, yet phenotypic characterization remains bottlenecked by assays that generally yield simple binary infection outcomes. Quantitative traits can be extracted from standard liquid cultures, but many phages fail to produce a measurable effect in these settings, necessitating an improved quantitative approach on semi-solid media. While end-point plaque size is commonly used for quantitative phenotyping, it only provides a static snapshot that fails to disentangle phage performance across changing bacterial growth phases. Here, we report a cost-effective, scalable method to quantify the bacterial killing rates of diverse phages by continuously tracking plaque development using a consumer-grade flatbed scanner and custom open-source computer vision software. By fitting a minimal, five-parameter phenomenological model to the diverse BASEL collection of Escherichia coli phages, we accurately captured plaque expansion dynamics across all bacterial growth phases. We discovered that plaque expansion parameters vary significantly across individual phages and taxonomic families, with the initial expansion rate demonstrating an inverse correlation with viral genome size. Notably, these plaque expansion dynamics did not correlate with bacterial collapse times in liquid culture, indicating that they represent distinct physiological and kinetic processes. Finally, we demonstrated that phage taxonomic family can be predicted with 72.1% accuracy solely from dynamic plaque parameters using a random forest classifier, whereas conventional endpoint plaque sizes held no predictive power. This time-lapse phenotyping approach yields robust kinetic data superior to end-point measurements, establishing a scalable platform to accelerate phage discovery and genotype-phenotype mapping.

microbiology↗

bioRxiv2025

Fine-Tuning of Material Properties via Catch Bonds

Semiflexible polymer networks are ubiquitous in biological systems, including a scaffolding structure within cells called the actin cytoskeleton. The polymers in these networks are often interconnected by transient bonds. For example, actin filaments in the cytoskeleton are physically connected via cross-linkers. The mechanical and kinetic properties of the cross-linkers significantly affect the rheological properties of the actin cytoskeleton. Here, we employed an agent-based model to elucidate how the force-dependent behaviors of the cross-linkers determine the material properties of passive networks without molecular motors and the force generation of active networks with molecular motors. The cross-linkers are assumed to behave as a slip bond whose dissociation rate is proportional to forces or as a catch-slip bond whose dissociation rate is inversely proportional to forces at low force level but proportional to forces at high force level. We found that catch-slip-bond cross-linkers can increase both stress and strain at a yield point by forming force-bearing elements that turn over continuously, which is impossible to achieve without the catch-slip bonds. In addition, we demonstrated that the catch-slip-bond cross-linkers help myosin motors generate greater internal contractile forces by reinforcing the force-bearing parts of the active network.

biophysics↗

bioRxiv2024

Rapid resistance evolution against phage cocktails

When bacteria are treated with multiple antibiotics simultaneously, resistance is highly unlikely to evolve. In stark contrast, resistance against multiple phages frequently arises during therapy. Why does resistance against multi-phage cocktails evolve so easily? Using a mathematical model, we show how the bacterial evolutionary dynamics and phage replicative dynamics uniquely intertwine, facilitating the rapid evolution of multi-phage resistance. As different phages replicate and become inhibitory at varying time points, bacteria can sequentially acquire resistance rather than simultaneously - increasing the chance of multi-resistance by orders of magnitude. Additionally, we predict and experimentally verify a regime where multi-phage resistance is robustly prevented. Our findings provide a framework for the rational design of phage cocktails to curtail resistance development. Resistance can be minimized by reducing the dose of the most potent phages or by using phages with longer latent periods, as this helps synchronize multi-phage selection.

microbiology↗

bioRxiv2024

Extreme diversity of phage amplification rates and phage-antibiotic interactions revealed by PHORCE

Growth rate plays a fundamental role in microbiology and serves as an important proxy for fitness in evolution. While high-throughput measurements of bacterial growth rates are easily performed in any microbiology lab, similar methods are lacking for bacteriophages. This gap hinders systematic comparisons of important phage phenotypes, such as their amplification rate in bacterial populations and their bactericidal effect, across different phages and environmental conditions. Here, we show that the amplification rate of lytic phages can be quantified by analyzing bacterial population growth and collapse dynamics under phage predation using a parsimonious mathematical model - an approach termed PHORCE (Phage-Host Observation for Rate estimation from Collapse Events). We found that the resulting phage amplification rate captures the bactericidal effect independent of initial phage and bacterial population sizes and across different growth conditions. Using high-throughput PHORCE, we found that the amplification rates of Escherichia coli phages vary widely by more than three orders of magnitude. Furthermore, our approach suggests that phage-antibiotic interactions are predominantly determined by the antibiotic, and not by the phage. In particular, the ribosome-inhibiting antibiotic doxycycline generally showed antagonism with phage amplification, whereas the DNA-damaging antibiotic nitrofurantoin was synergistic. This framework provides a means to quantitatively characterize phage phenotypes and may facilitate future high-throughput phage screens for antibacterial applications.

microbiology↗

bioRxiv2022

Near-critical tuning of cooperativity revealed by spontaneous switching in a protein signalling array

Dynamic properties of allosteric complexes are crucial for cellular information processing. However, direct observations of allosteric switches have been limited to compact molecular assemblies. Here, we report in vivo FRET measurements of spontaneous discrete-level fluctuations in the activity of the Escherichia coli chemosensory array -- an extensive membrane-associated assembly comprising thousands of molecules. Finite-size scaling analysis of the temporal statistics by a two-dimensional conformational spread model revealed nearest-neighbor coupling strengths within 3% of the Ising second-order phase transition, indicating that chemosensory arrays are poised at criticality. Our analysis yields estimates for the intrinsic timescale of conformational changes (~ 10 ms) of allosteric units, and identifies near-critical tuning as a design principle for balancing the inherent tradeoff between response amplitude and response speed in higher-order signaling assemblies. One-setence summaryIn vivo measurements of protein signaling array fluctuations reveal an allosteric system poised at criticality.

biophysics↗

bioRxiv2020

Weak catch bonds make strong networks

Molecular catch bonds are ubiquitous in biology and well-studied in the context of leukocyte extravasion1, cellular mechanosensing2,3, and urinary tract infection4. Unlike normal (slip) bonds, catch bonds strengthen under tension. The current paradigm is that this remarkable ability enables cells to increase their adhesion in fast fluid flows1,4, and hence provides strength-on-demand. Recently, cytoskeletal crosslinkers have been discovered that also display catch bonding5-8. It has been suggested that they strengthen cells, following the strength-on-demand paradigm9,10. However, catch bonds tend to be weaker compared to regular (slip) bonds because they have cryptic binding sites that are often inactive11-13. Therefore, the role of catch bonding in the cytoskeleton remains unclear. Here we reconstitute cytoskeletal actin networks to show that catch bonds render them both stronger and more deformable than slip bonds, even though the bonds themselves are weaker. We develop a model to show that weak binding allows the catch bonds to mitigate crack initiation by moving from low- to high-tension areas in response to mechanical loading. By contrast, slip bonds remain trapped in stress-free areas. We therefore propose that the mechanism of catch bonding is typified by dissociation-on-demand rather than strength-on-demand. Dissociation-on-demand can explain how both cytolinkers5-8,10,14,15 and adhesins1,2,4,12,16-20 exploit continuous redistribution to combine mechanical strength with the adaptability required for movement and proliferation21. Our findings provide a new perspective on diseases where catch bonding is compromised11,12 such as kidney focal segmental glomerulosclerosis22,23, caused by the -actinin-4 mutant studied here. Moreover, catch bonds provide a route towards creating life-like materials that combine strength with deformability24.

biophysics↗

bioRxiv2018

Frustrated binding of biopolymer crosslinkers

Transiently crosslinked actin filament networks allow cells to combine elastic rigidity with the ability to deform viscoelastically. Theoretical models of semiflexible polymer networks predict that the crosslinker unbinding rate governs the timescale beyond which viscoelastic flow occurs. However a direct comparison between network and crosslinker dynamics is lacking. Here we measure the networks stress relaxation timescale using rheology and the lifetime of bound crosslinkers using fluorescence recovery after photobleaching. lntruigingly, we observe that the crosslinker unbinding rate measured by FRAP is more than an order of magnitude slower than the rate measured by rheology. We rationalize this difference with a three-state model where crosslinkers are bound to either 0, 1 or 2 filaments, which allows us to extract crosslinker transition rates that are otherwise difficult to access. We find that the unbinding rate of singly bound crosslinkers is nearly two orders of magnitude slower than for double bound ones. We attribute the increased unbinding rate of doubly bound crosslinkers to the high stiffness of biopolymers, which frustrates crosslinker binding.

biophysics↗

bioRxiv2018

Unravelling the response of the actin cortex to electric pulses in vesicles

We study the role of a biomimetic actin cortex during the application of electric pulses that induce electroporation or electropermeabilization, using giant unilamellar vesicles (GUVs) as a model system. The actin cortex, a subjacently attached interconnected network of actin filaments, regulates the shape and mechanical properties of the plasma membrane of mammalian cells, and is a major factor influencing the mechanical response of the cell to external physical cues. We demonstrate that the presence of an actin shell inhibits the formation of macropores in the electroporated GUVs. Additionally, experiments on the uptake of dye molecules after electroporation show that the actin network slows down the resealing process of the permeabilized membrane. We further analyze the stability of the actin network inside the GUVs exposed to high electric pulses. We find disruption of the actin layer that is likely due to the electrophoretic forces acting on the actin filaments during the permeabilization of the GUVs. Our findings on the GUVs containing a biomimetic cortex provide a step towards understanding the discrepancies between the electroporation mechanism of a living cell and its simplified model of the empty GUV.

biophysics↗