Unified classification of mouse retinal ganglion cells using function, morphology, and gene expression

Classification and characterization of neuronal types is critical for understanding their function and dysfunction. Neuronal classification schemes typically rely on measurements of electrophysiological, morphological and molecular features, but aligning these data sets has been challenging. Here, we present a unified classification of retinal ganglion cells (RGCs), the sole retinal output neurons. We used visually-evoked responses to classify 1777 mouse RGCs into 42 types. We also obtained morphological or transcriptomic data from subsets and used these measurements to align the functional classification to publicly available morphological and transcriptomic data sets. We created an online database that allows users to browse or download the data and to classify RGCs from their light responses using a machine-learning algorithm. This work provides a resource for studies of RGCs, their upstream circuits in the retina, and their projections in the brain, and establishes a framework for future efforts in neuronal classification and open data distribution.

Intrinsic electrophysiological properties predict variability in morphology and connectivity among striatal Parvalbumin-expressing Pthlh-cells

Determining the cellular content of the nervous system in terms of cell types and the rules of their connectivity represents a fundamental challenge to the neurosciences. The recent advent of high-throughput techniques, such as single-cell RNA-sequencing has allowed for greater resolution in the identification of cell types and/or states. Although most of the current neuronal classification schemes comprise discrete clusters, several recent studies have suggested that, perhaps especially, within the striatum, neuronal populations exist in continua, with regards to both their molecular and electrophysiological properties. Whether these continua are stable properties, established during development, or if they reflect acute differences in activity-dependent regulation of critical genes is currently unknown. We set out to determine whether gradient-like molecular differences in the recently described Pthlh-expressing inhibitory interneuron population, which contains the Pvalb-expressing cells, correlate with differences in morphological and connectivity properties. We show that morphology and long-range inputs correlate with a spatially organized molecular and electrophysiological gradient of Pthlh-interneurons, suggesting that the processing of different types of information (by distinct anatomical striatal regions) has different computational requirements.

A Deep Convolutional Neural Network for Image Malware Classification

Malware classification and detection is an important factor in computer system security. However, signature-based methods currently used cannot provide an accurate detection of zero-day attacks and polymorphic viruses. This is why there is a need for detection based on machine learning. The purpose of this work is to present a deep neuronal classification method using convolutional and recurrent network layers in order to obtain the best features for classification. The proposed model achieves 98.73% accuracy on the Microsoft malware dataset.