Background: Deep learning methods have outperformed previous techniques in most computer vision tasks, including image-based plant phenotyping. However, massive data collection of root traits and the development of associated artificial intelligence approaches have been hampered by the inaccessibility of the rhizosphere. Here we present ChronoRoot, a system that combines 3D-printed open-hardware with deep segmentation networks for high temporal resolution phenotyping of plant roots in agarized medium. Results: We developed a novel deep learning–based root extraction method that leverages the latest advances in convolutional neural networks for image segmentation and incorporates temporal consistency into the root system architecture reconstruction process. Automatic extraction of phenotypic parameters from sequences of images allowed a comprehensive characterization of the root system growth dynamics. Furthermore, novel time-associated parameters emerged from the analysis of spectral features derived from temporal signals. Conclusions: Our work shows that the combination of machine intelligence methods and a 3D-printed device expands the possibilities of root high-throughput phenotyping for genetics and natural variation studies, as well as the screening of clock-related mutants, revealing novel root traits.

Chromatin immunoprecipitation followed by sequencing (ChIP-seq) is widely used to identify factor binding to genomic DNA and chromatin modifications. ChIP-seq data analysis is affected by genomic regions that generate ultra-high artifactual signals. To remove these signals from ChIP-seq data, the Encyclopedia of DNA Elements (ENCODE) project developed comprehensive sets of regions defined by low mappability and ultra-high signals called blacklists for human, mouse (Mus musculus), nematode (Caenorhabditis elegans), and fruit fly (Drosophila melanogaster). However, blacklists are not currently available for many model and nonmodel species. Here, we describe an alternative approach for removing false-positive peaks called greenscreen. Greenscreen is easy to implement, requires few input samples, and uses analysis tools frequently employed for ChIP-seq. Greenscreen removes artifactual signals as effectively as blacklists in Arabidopsis thaliana and human ChIP-seq dataset while covering less of the genome and dramatically improves ChIP-seq peak calling and downstream analyses. Greenscreen filtering reveals true factor binding overlap and occupancy changes in different genetic backgrounds or tissues. Because it is effective with as few as two inputs, greenscreen is readily adaptable for use in any species or genome build. Although developed for ChIP-seq, greenscreen also identifies artifactual signals from other genomic datasets including Cleavage Under Targets and Release Using Nuclease. We present an improved ChIP-seq pipeline incorporating greenscreen that detects more true peaks than other methods.

Thousands of lncRNAs with ecotype-specific expression, including two that likely regulate primary root growth, are potentially linked to the evolution of regula

Dysregulated lipid homeostasis is emerging as a potential cause of neurodegenerative disorders. However, evidence of errors in lipid homeostasis as a pathogenic mechanism of neurodegeneration remains limited. Here, we show that cerebellar neurodegeneration caused by Sorting Nexin 14 (SNX14) deficiency is associated with lipid homeostasis defects. Recent studies indicate that SNX14 is an interorganelle lipid transfer protein that regulates lipid transport, lipid droplet (LD) biogenesis, and fatty acid desaturation, suggesting that human SNX14 deficiency belongs to an expanding class of cerebellar neurodegenerative disorders caused by altered cellular lipid homeostasis. To test this hypothesis, we generated a mouse model that recapitulates human SNX14 deficiency at a genetic and phenotypic level. We demonstrate that cerebellar Purkinje cells (PCs) are selectively vulnerable to SNX14 deficiency while forebrain regions preserve their neuronal content. Ultrastructure and lipidomic studies reveal widespread lipid storage and metabolism defects in SNX14-deficient mice. However, predegenerating SNX14-deficient cerebella show a unique accumulation of acylcarnitines and depletion of triglycerides. Furthermore, defects in LD content and telolysosome enlargement in predegenerating PCs suggest lipotoxicity as a pathogenic mechanism of SNX14 deficiency. Our work shows a selective cerebellar vulnerability to altered lipid homeostasis and provides a mouse model for future therapeutic studies.

Background: Deep learning methods have outperformed previous techniques in most computer vision tasks, including image-based plant phenotyping. However, massive data collection of root traits and the development of associated artificial intelligence approaches have been hampered by the inaccessibility of the rhizosphere. Here we present ChronoRoot, a system which combines 3D printed open-hardware with deep segmentation networks for high temporal resolution phenotyping of plant roots in agarized medium. Results: We developed a novel deep learning based root extraction method which leverages the latest advances in convolutional neural networks for image segmentation, and incorporates temporal consistency into the root system architecture reconstruction process. Automatic extraction of phenotypic parameters from sequences of images allowed a comprehensive characterization of the root system growth dynamics. Furthermore, novel time-associated parameters emerged from the analysis of spectral features derived from temporal signals. Conclusions: Altogether, our work shows that the combination of machine intelligence methods and a 3D-printed device expands the possibilities of root high-throughput phenotyping for genetics and natural variation studies as well as the screening of clock-related mutants, revealing novel root traits.

Metabolic dysregulation is one of the most common causes of pediatric neurodegenerative disorders. However, how the disruption of ubiquitous and essential metabolic pathways predominantly affect neural tissue remains unclear. Here we use mouse models of a childhood neurodegenerative disorder caused by AMPD2 deficiency to study cellular and molecular mechanisms that lead to selective neuronal vulnerability to purine metabolism imbalance. We show that mouse models of AMPD2 deficiency exhibit predominant degeneration of the hippocampal dentate gyrus, despite a general reduction of brain GTP levels. Neurodegeneration-resistant regions accumulate micron-sized filaments of IMPDH2, the rate limiting enzyme in GTP synthesis, while these filaments are barely detectable in the hippocampal dentate gyrus. Furthermore, we show that IMPDH2 filament disassembly reduces GTP levels and impairs growth of neural progenitor cells derived from individuals with human AMPD2 deficiency. Together, our findings suggest that IMPDH2 polymerization prevents detrimental GTP deprivation, opening the possibility of exploring the induction of IMPDH2 assembly as a therapy for neurodegeneration.