Improved Unsupervised Representation Learning of Spatial Transcriptomic Data with Sparse Filtering

We have developed representation learning methods, specifically to address the constraints and advantages of complex spatial data. Sparse filtering (SFt), uses principles of sparsity and mutual information to build representations from both global and local features from a minimal list of samples. Critically, the samples that comprise each representation are listed and ranked by informativeness. We used the Allen Mouse Brain Atlas gene expression data for prototyping and established performance metrics based on representation accuracy to labeled anatomy. SFt, implemented with the PyTorch machine learning libraries for Python, returned the most accurate reconstruction of anatomical ground truth of any method tested. SFt generated gene lists could be further compressed, retaining 95% of informativeness with only 580 genes. Finally, we build classifiers capable of parsing anatomy with >95% accuracy using only 10 derived genes. Sparse learning is a powerful, but underexplored means to derive biologically meaningful representations from complex datasets and a quantitative basis for compressed sensing of classifiable phenomena. SFt should be considered as an alternative to PCA or manifold learning for any high dimensional dataset and the basis for future spatial learning algorithms.

Diffusion Tensor Imaging Reveals Whole-Brain Microstructural Changes in the P301L Mouse Model of Tauopathy

<b><i>Introduction:</i></b> Increased expression of hyperphosphorylated tau and the formation of neurofibrillary tangles are associated with neuronal loss and white matter damage. Using high-resolution ex vivo diffusion tensor imaging (DTI), we investigated microstructural changes in the white and grey matter in the P301L mouse model of human tauopathy at 8.5 months of age. For unbiased computational analysis, we implemented a pipeline for voxel-based analysis (VBA) and atlas-based analysis (ABA) of DTI mouse brain data. <b><i>Methods:</i></b> Hemizygous and homozygous transgenic P301L mice and non-transgenic littermates were used. DTI data were acquired for generation of fractional anisotropy (FA), mean diffusivity (MD), radial diffusivity (RD), and axial diffusivity (AD) maps. VBA on the entire brain was performed using SPM8 and the SPM Mouse toolbox. Initially, all DTI maps were coregistered with the Allen mouse brain atlas to bring them to one common coordinate space. In VBA, coregistered DTI maps were normalized and smoothed in order to perform two-sample and unpaired <i>t</i> tests with false discovery rate correction to compare hemizygotes with non-transgenic littermates, homozygotes with non-transgenic littermates, and hemizygotes with homozygotes on each DTI parameter map. In ABA, the average values for selected regions of interests were computed with coregistered DTI maps and labels in Allen mouse brain atlas. Afterwards, a Kruskal-Wallis one-way ANOVA on ranks with a Tukey post hoc test was executed on the estimated average values. <b><i>Results:</i></b> With VBA, we found pronounced and brain-wide spread changes when comparing homozygous, P301L mice with non-transgenic littermates, which were not seen when comparing hemizygous P301L with non-transgenic animals. Statistical comparison of DTI metrics in selected brain regions by ABA corroborated findings from VBA. FA was found to be decreased in most brain regions, while MD, RD, and AD were increased in homozygotes compared to hemizygotes and non-transgenic littermates. <b><i>Discussion/Conclusion:</i></b> High-resolution ex vivo DTI demonstrated brain-wide microstructural and gene-dose-dependent changes in the P301L mouse model of human tauopathy. The DTI analysis pipeline may serve for the phenotyping of models of tauopathy and other brain diseases.

DTI reveals whole-brain microstructural changes in the P301L mouse model of tauopathy

IntroductionIncreased expression of hyperphosphorylated tau and the formation of neurofibrillary tangles are associated with neuronal loss and white matter damage. Using high resolution ex vivo diffusion tensor imaging (DTI), we investigated microstructural changes in the white and grey matter in the P301L mouse model of human tauopathy at 8.5 months-of-age. For unbiased computational analysis, we implemented a pipeline for voxel-based analysis (VBA) and atlas-based analysis (ABA) of DTI mouse brain data.MethodsHemizygous and homozygous transgenic P301L mice and non-transgenic littermates were used. DTI data were acquired for generation of fractional anisotropy (FA), mean diffusivity (MD), radial diffusivity (RD), axial diffusivity (AD) maps. VBA on the entire brain were performed using SPM8 and SPM Mouse toolbox. Initially, all DTI maps were co-registered with Allen mouse brain atlas to bring them to one common coordinate space. In VBA, co-registered DTI maps were normalized and smoothed in order to perform two-sample t-tests to compare hemizygotes with non-transgenic littermates, homozygotes with non-transgenic littermates, hemizygotes with homozygotes on each DTI parameter map. In ABA, the average values for selected regions-of-interests were computed with co-registered DTI maps and labels in Allen mouse brain atlas. After, the same two-sample t-tests were executed on the estimated average values.ResultsWe made reconstructed DTI data and VBA and ABA pipeline publicly available. With VBA, we found microstructural changes in the white matter in hemizygous P301L mice compared to non-transgenic littermates. In contrast, more pronounced and brain-wide spread changes were observed in VBA when comparing homozygous P301L mice with non-transgenic littermates. Statistical comparison of DTI metrics in selected brain regions by ABA corroborated findings from VBA. FA was found to be decreased in most brain regions, while MD, RD and AD were increased compared to hemizygotes and non-transgenic littermates.Discussion/ConclusionHigh resolution ex vivo DTI demonstrated brain-wide microstructural changes in the P301L mouse model of human tauopathy. The comparison between hemizygous and homozygous P301L mice revealed a gene-dose dependent effect on DTI metrics. The publicly available computational data analysis pipeline can provide a platform for future mechanistic and longitudinal studies.

SpaGE: Spatial Gene Enhancement using scRNA-seq

Single-cell technologies are emerging fast due to their ability to unravel the heterogeneity of biological systems. While scRNA-seq is a powerful tool that measures whole-transcriptome expression of single cells, it lacks their spatial localization. Novel spatial transcriptomics methods do retain cells spatial information but some methods can only measure tens to hundreds of transcripts. To resolve this discrepancy, we developed SpaGE, a method that integrates spatial and scRNA-seq datasets to predict whole-transcriptome expressions in their spatial configuration. Using five dataset-pairs, SpaGE outperformed previously published methods and showed scalability to large datasets. Moreover, SpaGE predicted new spatial gene patterns that are confirmed independently using in situ hybridization data from the Allen Mouse Brain Atlas.

Identification of genetic markers for cortical areas using a Random Forest classification routine and the Allen Mouse Brain Atlas

The mammalian neocortex is subdivided into a series of ‘cortical areas’ that are functionally and anatomically distinct, and are often distinguished in brain sections using histochemical stains and other markers of protein expression. We searched the Allen Mouse Brain Atlas, a database of gene expression, for novel markers of cortical areas. We employed a random forest algorithm to screen for genes that change expression at area borders. We found novel genetic markers for 19 of 39 areas and provide code that quickly and efficiently searches the Allen Mouse Brain Atlas.

A tool for analyzing electrode tracks from slice histology

It is now possible to record from hundreds of neurons across multiple brain regions in a single electrophysiology experiment. An essential step in the ensuing data analysis is to assign recorded neurons to the correct brain regions. Brain regions are typically identified after the recordings by comparing images of brain slices to a reference atlas by eye. This introduces error, in particular when slices are not cut at a perfectly coronal angle or when electrode tracks span multiple slices. Here we introduce SHARP-Track, a tool to localize regions of interest and plot the brain regions they pass through. SHARP-Track offers a MATLAB user interface to explore the Allen Mouse Brain Atlas, register asymmetric slice images to the atlas using manual input, and interactively analyze electrode tracks. We find that it reduces error compared to localizing electrodes in a reference atlas by eye. See github.com/cortex-lab/allenCCF for the software and wiki.

Comparing the Expression of Genes Related to Serotonin (5-HT) in C57BL/6J Mice and Humans Based on Data Available at the Allen Mouse Brain Atlas and Allen Human Brain Atlas

Brain atlases are tools based on comprehensive studies used to locate biological characteristics (structures, connections, proteins, and gene expression) in different regions of the brain. These atlases have been disseminated to the point where tools have been created to store, manage, and share the information they contain. This study used the data published by the Allen Mouse Brain Atlas (2004) for mice (C57BL/6J) and Allen Human Brain Atlas (2010) for humans (6 donors) to compare the expression of serotonin-related genes. Genes of interest were searched for manually in each case (in situ hybridization for mice and microarrays for humans), normalized expression data (z-scores) were extracted, and the results were graphed. Despite the differences in methodology, quantification, and subjects used in the process, a high degree of similarity was found between expression data. Here we compare expression in a way that allows the use of translational research methods to infer and validate knowledge. This type of study allows part of the relationship between structures and functions to be identified, by examining expression patterns and comparing levels of expression in different states, anatomical correlations, and phenotypes between different species. The study concludes by discussing the importance of knowing, managing, and disseminating comprehensive, open-access studies in neuroscience.