Biology subjects

Berens, P.

Publications and source records attributed to Berens, P..

9 recordsLinked to original sources

bioRxiv2018

The art of using t-SNE for single-cell transcriptomics

Single-cell transcriptomics yields ever growing data sets containing RNA expression levels for thousands of genes from up to millions of cells. Common data analysis pipelines include a dimensionality reduction step for visualising the data in two dimensions, most frequently performed using t-distributed stochastic neighbour embedding (t-SNE). It excels at revealing local structure in high-dimensional data, but naive applications often suffer from severe shortcomings, e.g. the global structure of the data is not represented accurately. Here we describe how to circumvent such pitfalls, and develop a protocol for creating more faithful t-SNE visualisations. It includes PCA initialisation, a high learning rate, and multi-scale similarity kernels; for very large data sets, we additionally use exaggeration and downsampling-based initialisation. We use published single-cell RNA-seq data sets to demonstrate that this protocol yields superior results compared to the naive application of t-SNE.

bioinformatics↗

bioRxiv2018

Bayesian Hypothesis Testing And Experimental Design For Two-Photon Imaging Data

Variability, stochastic or otherwise, is a central feature of neural circuits. Yet the means by which variation and uncertainty are derived from noisy observations of neural activity is often unprincipled, with too much weight placed on numerical convenience at the cost of statistical rigour. For two-photon imaging data, composed of fundamentally probabilistic streams of photon detections, the problem is particularly acute. Here, we present a complete statistical pipeline for the inference and analysis of neural activity using Gaussian Process Regression, applied to two-photon recordings of light-driven activity in ex vivo mouse retina. We demonstrate the flexibility and extensibility of these models, considering cases with non-stationary statistics, driven by complex parametric stimuli, in signal discrimination, hierarchical clustering and inference tasks. Sparse approximation methods allow these models to be fitted rapidly, permitting them to actively guiding the design of light stimulation in the midst of ongoing two-photon experiments.

neuroscience↗

bioRxiv2018

Human-level saccade detection performance using deep neural networks

Saccades are ballistic eye movements that rapidly shift gaze from one location of visual space to another. Detecting saccades in eye movement recordings is important not only for studying the neural mechanisms underlying sensory, motor, and cognitive processes, but also as a clinical and diagnostic tool. However, automatically detecting saccades can be difficult, particularly when such saccades are generated in coordination with other tracking eye movements, like smooth pursuits, or when the saccade amplitude is close to eye tracker noise levels, like with microsaccades. In such cases, labeling by human experts is required, but this is a tedious task prone to variability and error. We developed a convolutional neural network (CNN) to automatically detect saccades at human-level performance accuracy. Our algorithm surpasses state of the art according to common performance metrics, and will facilitate studies of neurophysiological processes underlying saccade generation and visual processing.

neuroscience↗

bioRxiv2018

Mouse dLGN receives input from a diverse population of retinal ganglion cells with limited convergence

In the mouse, the parallel output of more than 30 functional types of retinal ganglion cells (RGCs) serves as the basis for all further visual processing. Little is known about how the representation of visual information changes between the retina and the dorsolateral geniculate nucleus (dLGN) of the thalamus, the main relay station between the retina and cortex. Here, we functionally characterized responses of retrogradely labeled dLGN-projecting RGCs and dLGN neurons to the same set of visual stimuli. We found that many of the previously identified functional RGC types innervate the dLGN, which maintained a high degree of functional diversity. Using a linear model to assess functional connectivity between RGC types and dLGN neurons, we found that the responses of dLGN neurons could be predicted as a linear combination of inputs from on average five RGC types, but only two of those had the strongest functional impact. Thus, mouse dLGN receives input from a diverse population of RGCs with limited functional convergence.

neuroscience↗

bioRxiv2018

Sparse reduced-rank regression for exploratory visualization of single cell patch-seq recordings

In genomics, transcriptomics, and related biological fields (collectively known as omics), it is common to work with n << p datasets with the dimensionality much larger than the sample size. In recent years, combinations of experimental techniques began to yield multiple sets of features for the same set of biological replicates. One example is Patch-seq, a method combining single-cell RNA sequencing with electrophysiological recordings from the same cells. Here we present a framework based on sparse reduced-rank regression for obtaining an interpretable visualization of the relationship between the transcriptomic and the electrophysiological data. We use an elastic net regularization penalty that yields sparse solutions and allows for an efficient computational implementation. Using several publicly available Patch-seq datasets, we show that sparse reduced-rank regression outperforms both sparse full-rank regression and non-sparse reduced-rank regression in terms of predictive performance, and can outperform existing methods for sparse partial least squares and sparse canonical correlation analysis in terms of out-of-sample correlations. We introduce a bibiplot visualization in order to display the dominant factors determining the relationship between transcriptomic and electrophysiological properties of neurons. We believe that sparse reduced-rank regression can provide a valuable tool for the exploration and visualization of paired multivariate datasets, including Patch-seq.

bioinformatics↗

bioRxiv2017

Zebrafish differentially process colour across visual space to match natural scenes

Animal eyes evolve to process behaviourally important visual information, but how retinas deal with statistical asymmetries in visual space remains poorly understood. Using hyperspectral imaging in the field, in-vivo 2-photon imaging of retinal neurons and anatomy, here we show that larval zebrafish use a highly anisotropic retina to asymmetrically survey their natural visual world. First, different neurons dominate different parts of the eye, and are linked to a systematic shift in inner retinal function: Above the animal, there is little colour in nature and retinal circuits are largely achromatic. Conversely, the lower visual field and horizon are colour-rich, and are predominately surveyed by chromatic and colour-opponent circuits that are spectrally matched to the dominant chromatic axes in nature. Second, above the frontal horizon, a high-gain ultraviolet-system piggy-backs onto retinal circuits, likely to support prey-capture. Our results demonstrate high functional diversity among single genetically and morphologically defined types of neurons.

neuroscience↗

bioRxiv2017

Community-based benchmarking improves spike inference from two-photon calcium imaging data

In recent years, two-photon calcium imaging has become a standard tool to probe the function of neural circuits and to study computations in neuronal populations1, 2. However, the acquired signal is only an indirect measurement of neural activity due to the comparatively slow dynamics of fluorescent calcium indicators3. Different algorithms for estimating spike trains from noisy calcium measurements have been proposed in the past4-8, but it is an open question how far performance can be improved. Here, we report the results of the spikefinder challenge, launched to catalyze the development of new spike inference algorithms through crowd-sourcing. We present ten of the submitted algorithms which show improved performance compared to previously evaluated methods. Interestingly, the top-performing algorithms are based on a wide range of principles from deep neural networks to generative models, yet provide highly correlated estimates of the neural activity. The competition shows that benchmark challenges can drive algorithmic developments in neuroscience.

neuroscience↗

bioRxiv2017

Local Signal Processing In Mouse Horizontal Cell Dendrites

The mouse retina contains a single type of horizontal cell, a GABAergic interneuron that samples from all cone photoreceptors within reach and modulates their glutamatergic output via parallel feedback mechanisms. Because horizontal cells form an electrically-coupled network, they have been implicated in global signal processing, such as large scale contrast enhancement. Recently, it has been proposed that horizontal cells can also act locally at the level of individual cone photoreceptors. To test this possibility physiologically, we used two-photon microscopy to record light stimulus-evoked Ca2+ signals in cone axon terminals and horizontal cell dendrites as well as glutamate release in the outer plexiform layer. By selectively stimulating the two mouse cone opsins with green and UV light, we assessed whether signals from individual cones remain \"isolated\" within horizontal cell dendritic tips, or whether they spread across the dendritic arbour. Consistent with the mouse s opsin expression gradient, we found that the Ca2+ signals recorded from dendrites of dorsal horizontal cells were dominated by M- and those of ventral horizontal cells by S-opsin activation. The signals measured in neighbouring horizontal cell dendritic tips varied markedly in their chromatic preference, arguing against global processing. Rather, our experimental data and results from biophysically realistic modelling support the idea that horizontal cells can process cone input locally, extending the \"classical\" view of horizontal cells function. Pharmacologically removing horizontal cells from the circuitry reduced the sensitivity of the cone signal to low frequencies, suggesting that local horizontal cell feedback shapes the temporal properties of cone output.\n\nHighlightsO_LILight-evoked Ca2+ signals in horizontal cell dendrites reflect opsin gradient\nC_LIO_LIChromatic preferences in neighbouring dendritic tips vary markedly\nC_LIO_LIMouse horizontal cells process cone photoreceptor input locally\nC_LIO_LILocal horizontal cell feedback shapes the temporal properties of cone output\nC_LI\n\neTOC BlurbChapot et al. show that local light responses in mouse horizontal cell dendrites inherit properties, including chromatic preference, from the presynaptic cone photoreceptor, suggesting that their dendrites can provide \"private\" feedback to cones, for instance, to shape the temporal filtering properties of the cone synapse.

neuroscience↗

bioRxiv2016

Leveraging uncertainty information from deep neural networks for disease detection

Deep learning (DL) has revolutionized the field of computer vision and image processing. In medical imaging, algorithmic solutions based on DL have been shown to achieve high performance on tasks that previously required medical experts. However, DL-based solutions for disease detection have been proposed without methods to quantify and control their uncertainty in a decision. In contrast, a physician knows whether she is uncertain about a case and will consult more experienced colleagues if needed. Here we evaluate drop-out based Bayesian uncertainty measures for DL in diagnosing diabetic retinopathy (DR) from fundus images and show that it captures uncertainty better than straightforward alternatives. Furthermore, we show that uncertainty informed decision referral can improve diagnostic performance. Experiments across different networks, tasks and datasets show robust generalization. Depending on network capacity and task/dataset difficulty, we surpass 85% sensitivity and 80% specificity as recommended by the NHS when referring 0%-20% of the most uncertain decisions for further inspection. We analyse causes of uncertainty by relating intuitions from 2D visualizations to the high-dimensional image space. While uncertainty is sensitive to clinically relevant cases, sensitivity to unfamiliar data samples is task dependent, but can be rendered more robust.

bioinformatics↗