Biology subjects

Evans, S. D.

Publications and source records attributed to Evans, S. D..

3 recordsLinked to original sources

bioRxiv2026

afMLevel: Deep learning U-Net Models for Levelling Atomic Force Microscopy Images and Movies

AFM experiments can now produce hundreds of frames per session and this rises to thousands in the case of video-rate high-speed AFM, but every image must still be levelled to remove tilt, bowing and line noise before it can be measured. Current automatic routines rely on user-chosen masks and parameters that work for some sample types but fail on others, so levelling often remains a manual, subjective bottleneck. Here, we present two U-Net models developed using machine learning for levelling AFM images without altering the sample topography. MLMask segments features so that they are excluded from conventional plane and line background fits, and MLBackground predicts the noise and artifact background directly, which is then subtracted. Both were trained on 2,001 experimental images from multiple instruments and sample types, using manually levelled images as ground truth. On 164 unseen images, MLMask routines brought 70-80% of images below a mean squared error of 0.1 relative to manual levelling, compared with 60-70% for the best multi-plane method, with a lower mean error (0.1-0.15 vs 0.15-0.3), and both models significantly outperformed fourteen classical routines (p < 0.01). Applied frame by frame, the models process 512x512 pixel HS-AFM movies at 4.7 frames.s-1. We provide afMLevel as an openly-available package for the use of other researchers. By removing the need to choose levelling parameters image by image, this package together with its models and paired training data, provides a foundation for automated AFM analysis pipelines.

biophysics↗

bioRxiv2024

Evidence for a transfer-to-trap mechanism of fluorophore concentration quenching in lipid bilayers

It is important to understand the behaviours of fluorescent molecules because, firstly, they are often utilized as probes in biophysical experiments and, secondly, they are crucial cofactors in biological processes such as photosynthesis. A phenomenon called fluorescence quenching occurs when fluorophores are present at high concentrations but the mechanisms for quenching are debated. Here, we used a technique called in-membrane electrophoresis to generate concentration gradients of fluorophores within a supported lipid bilayer (SLB), across which quenching was expected to occur. Fluorescence lifetime imaging microscopy (FLIM) provides images where the fluorescence intensity in each pixel is correlated to fluorescence lifetime: the intensity provides information about the location and concentration of fluorophores and the lifetime reveals the occurrence of energy-dissipative processes. FLIM was used to compare the quenching behaviour of three commonly-used fluorophores: Texas Red (TR), nitrobenzoaxadiazole (NBD) and 4,4-difluoro-4-bora-3a,4a-diaza-s-indacene (BODIPY). FLIM images provided evidence of quenching in regions where the fluorophores accumulated but the degree of quenching varied between the different fluorophores. The relationship between quenching and concentration was quantified and the critical radius for trap formation, representing the relative quenching strength, was calculated as 2.70, 2.02 and 1.14 nm, for BODIPY, TR and NBD, respectively. The experimental data supports the theory that quenching takes place via a transfer-to-trap mechanism which proposes, firstly, that excitation energy is transferred between fluorophores and may reach a trap site resulting in immediate energy dissipation and, secondly, that trap sites are formed in a concentration-dependent manner. Some previous work suggested that quenching occurs only when fluorophores aggregate, or form long-lived dimers, but our data and this theory argues that traps may be statistical pairs of fluorophores that exist only transiently. Our findings should inspire future work to assess whether these traps can be charge-transfer states, excited state dimers or something else.

biophysics↗

bioRxiv2020

Understanding the photophysics and structural organization of photosynthetic proteins using model lipid membranes assembled from natural plant thylakoids

The light-harvesting (LH) biomembranes from photosynthetic organisms perform solar energy absorption and transfer with high efficiency. There is great interest in the nanoscale biophysics of photosynthesis, however, natural membranes are complex and highly curved so can be challenging to study. Here we present model photosynthetic "hybrid membranes" assembled from a combination of natural LH membranes and synthetic lipids deposited into a patterned polymerized lipid template on glass. This arrangement offers many advantages over previous model systems including: a sufficiently complex mixture of natural proteins to mimic the biological processes, a modular self-assembly mechanism, and a stabilizing template promoting the formation of supported lipid bilayers from complex natural membranes with high protein content (that would not otherwise form). These hybrid membranes can be used as a platform to delineate the complex relationship between LH energy pathways and membrane organization. Atomic force microscopy and fluorescence lifetime microscopy revealed that hybrid membranes have an elongated fluorescence lifetime ([~]4 ns) compared to native membranes ([~]0.5 ns), a direct consequence of reduced protein density and an uncoupling of protein-protein interactions. We observed the real time self-assembly and migration of LH proteins from natural membrane extracts into the hybrid membranes and monitored the photophysical state of the membranes at each stage. Finally, experiments utilizing our hybrid membranes suggest that assays currently used in the photosynthesis community to test the electron transfer activity of Photosystem II may have non-specific interactions with other proteins, implying that new methods are needed for reliable quantification of electron transfers in photosynthesis.

biophysics↗