Protocol to perform integrative analysis of high-dimensional single-cell multimodal data using an interpretable deep learning technique.

The advent of single-cell multi-omics sequencing technology makes it possible for researchers to leverage multiple modalities for individual cells. Here, we present a protocol to perform integrative analysis of high-dimensional single-cell multimodal data using an interpretable deep learning technique called moETM. We describe steps for data preprocessing, multi-omics integration, inclusion of prior pathway knowledge, and cross-omics imputation. As a demonstration, we used the single-cell multi-omics data collected from bone marrow mononuclear cells (GSE194122) as in our original study. For complete details on the use and execution of this protocol, please refer to Zhou et al.1.

Comparing exercise determinants between Black and White older adults with heart failure.

BACKGROUND: Heart Failure is a leading cause of mortality among older adults. Engaging in regular exercise at moderate-to-vigorous intensity has been shown to improve survival rates. Theory-informed methodologies have been recommended to promote exercise, but limited application of theoretical framework has been conducted for understanding racial disparities among older adults with heart failure. This study aimed to use the Health Belief Model to compare exercise behavior determinants between Black and White older adults diagnosed with heart failure. METHODS: The HF-ACTION Trial is a multi-site study designed to promote exercise among individuals with heart failure that randomized participants to an experimental (three months of group exercise sessions followed by home-based training) or control arm. The present study used structural equation modeling to test the change in Health Belief Model constructs and exercise behavior across 12 months among older adults. RESULTS: Participants (n = 671) were older adults, 72.28 (SD = 5.41) years old, (Black: n = 230; White, n = 441) diagnosed with heart failure and reduced ejection fraction. The model found perceived benefits, self-efficacy, perceived threats, and perceived barriers to predict exercise behavior among Black and White older adults. However, among these constructs, only perceived benefits and self-efficacy were facilitated via intervention for both races. Additionally, the intervention was effective for addressing perceived barriers to exercise only among White participants. Finally, the intervention did not result in a change of perceived threats for both races. CONCLUSIONS: Among health belief model constructs, perceived threats and barriers were not facilitated for both races in the experimental arm, and the intervention did not resolve barriers among Black older adults. Racial differences need to be considered when designing interventions for clinical populations as future studies are warranted to address barriers to exercise among Black older adults with heart failure.

Single-cell multi-omics topic embedding reveals cell-type-specific and COVID-19 severity-related immune signatures.

The advent of single-cell multi-omics sequencing technology makes it possible for researchers to leverage multiple modalities for individual cells and explore cell heterogeneity. However, the high-dimensional, discrete, and sparse nature of the data make the downstream analysis particularly challenging. Here, we propose an interpretable deep learning method called moETM to perform integrative analysis of high-dimensional single-cell multimodal data. moETM integrates multiple omics data via a product-of-experts in the encoder and employs multiple linear decoders to learn the multi-omics signatures. moETM demonstrates superior performance compared with six state-of-the-art methods on seven publicly available datasets. By applying moETM to the scRNA + scATAC data, we identified sequence motifs corresponding to the transcription factors regulating immune gene signatures. Applying moETM to CITE-seq data from the COVID-19 patients revealed not only known immune cell-type-specific signatures but also composite multi-omics biomarkers of critical conditions due to COVID-19, thus providing insights from both biological and clinical perspectives.

Single-cell multi-omic topic embedding reveals cell-type-specific and COVID-19 severity-related immune signatures.

The advent of single-cell multi-omics sequencing technology makes it possible for re-searchers to leverage multiple modalities for individual cells and explore cell heterogeneity. However, the high dimensional, discrete, and sparse nature of the data make the downstream analysis particularly challenging. Most of the existing computational methods for single-cell data analysis are either limited to single modality or lack flexibility and interpretability. In this study, we propose an interpretable deep learning method called multi-omic embedded topic model (moETM) to effectively perform integrative analysis of high-dimensional single-cell multimodal data. moETM integrates multiple omics data via a product-of-experts in the encoder for efficient variational inference and then employs multiple linear decoders to learn the multi-omic signatures of the gene regulatory programs. Through comprehensive experiments on public single-cell transcriptome and chromatin accessibility data (i.e., scRNA+scATAC), as well as scRNA and proteomic data (i.e., CITE-seq), moETM demonstrates superior performance compared with six state-of-the-art single-cell data analysis methods on seven publicly available datasets. By applying moETM to the scRNA+scATAC data in human bone marrow mononuclear cells (BMMCs), we identified sequence motifs corresponding to the transcription factors that regulate immune gene signatures. Applying moETM analysis to CITE-seq data from the COVID-19 patients revealed not only known immune cell-type-specific signatures but also composite multi-omic biomarkers of critical conditions due to COVID-19, thus providing insights from both biological and clinical perspectives.

Extracting electrophysiological correlates of functional magnetic resonance imaging data using the canonical polyadic decomposition.

The relation between electrophysiology and BOLD-fMRI requires further elucidation. One approach for studying this relation is to find time-frequency features from electrophysiology that explain the variance of BOLD time-series. Convolution of these features with a canonical hemodynamic response function (HRF) is often required to model neurovascular coupling mechanisms and thus account for time shifts between electrophysiological and BOLD-fMRI data. We propose a framework for extracting the spatial distribution of these time-frequency features while also estimating more flexible, region-specific HRFs. The core component of this method is the decomposition of a tensor containing impulse response functions using the Canonical Polyadic Decomposition. The outputs of this decomposition provide insight into the relation between electrophysiology and BOLD-fMRI and can be used to construct estimates of BOLD time-series. We demonstrated the performance of this method on simulated data while also examining the effects of simulated measurement noise and physiological confounds. Afterwards, we validated our method on publicly available task-based and resting-state EEG-fMRI data. We adjusted our method to accommodate the multisubject nature of these datasets, enabling the investigation of inter-subject variability with regards to EEG-to-BOLD neurovascular coupling mechanisms. We thus also demonstrate how EEG features for modelling the BOLD signal differ across subjects.

Modeling BOLD-fMRI Hemodynamics via Multidimensional Decomposition of Electrophysiology Data: A Simulation Study.

We propose a framework for studying the electrophysiological correlates of BOLD-fMRI. This framework relies on structured coupled matrix-tensor factorization (sCMTF), a joint multidimensional decomposition which reveals dynamical interactions between LFP/EEG oscillatory features and BOLD-fMRI data. We test whether LFP/EEG-BOLD co-fluctuations and regional hemodynamic response functions can be estimated by sCMTF using whole-brain modelling of restingstate activity. We produce permuted datasets to show that our framework extracts EEG/LFP temporal patterns that correlate significantly with BOLD signal fluctuations. Our framework is also capable of estimating HRFs that accurately embodies simulated hemodynamics, with a word of caution regarding initialization of the sCMTF algorithm.

The influence of tissue composition uncertainty on dose distributions in brachytherapy.

BACKGROUND AND PURPOSE: Model-based dose calculation algorithms (MBDCAs) have evolved from serving as a research tool into clinical practice in brachytherapy. This study investigates primary sources of tissue elemental compositions used as input to MBDCAs and the impact of their variability on MBDCA-based dosimetry. MATERIALS AND METHODS: Relevant studies were retrieved through PubMed. Minimum dose delivered to 90% of the target (D90), minimum dose delivered to the hottest specified volume for organs at risk (OAR) and mass energy-absorption coefficients (µen/ρ) generated by using EGSnrc "g" user-code were compared to assess the impact of compositional variability. RESULTS: Elemental composition for hydrogen, carbon, oxygen and nitrogen are derived from the gross contents of fats, proteins and carbohydrates for any given tissue, the compositions of which are taken from literature dating back to 1940-1950. Heavier elements are derived from studies performed in the 1950-1960. Variability in elemental composition impacts greatly D90 for target tissues and doses to OAR for brachytherapy with low energy sources and less for 192Ir-based brachytherapy. Discrepancies in µen/ρ are also indicative of dose differences. CONCLUSIONS: Updated elemental compositions are needed to optimize MBDCA-based dosimetry. Until then, tissue compositions based on gross simplifications in early studies will dominate the uncertainties in tissue heterogeneity.