Dimensionality reduction of longitudinal ’omics data using modern tensor factorizations

The structure is the message: Preserving experimental context through tensor decomposition

Recent biological studies have been revolutionized in scale and granularity by multiplex and high-throughput assays. Profiling cell responses across several experimental parameters, such as perturbations, time, and genetic contexts, leads to richer and more generalizable findings. However, these multidimensional datasets necessitate a reevaluation of the conventional methods for their representation and analysis. Traditionally, experimental parameters are merged to flatten the data into a two-dimensional matrix, sacrificing crucial experiment context reflected by the structure. As Marshall McLuhan famously stated, "the medium is the message." In this work, we propose that the experiment structure is the medium in which subsequent analysis is performed, and the optimal choice of data representation must reflect the experiment structure. We review how tensor-structured analyses and decompositions can preserve this information. We contend that tensor methods are poised to become integral to the biomedical data sciences toolkit.

Diagnosis and Risk Factors of Prediabetes and Diabetes in People Living With Human Immunodeficiency Virus: Evaluation of Clinical and Microbiome Parameters

Diabetes mellitus (DM) is more common among people living with human immunodeficiency virus (PLWH) compared with healthy individuals. In a prospective multicenter study (N = 248), we identified normoglycemic (48.7%), prediabetic (44.4%), and diabetic (6.9%) PLWH. Glycosylated hemoglobin (HbA1c) and fasting blood glucose (FBG) sensitivity in defining dysglycemia was 96.8%, while addition of oral glucose tolerance test led to reclassification of only 4 patients. Inclusion of 93 additional PLWH with known DM enabled identification of multiple independent predictors of dysglycemia or diabetes: older age, higher body mass index, Ethiopian origin, HIV duration, lower integrase inhibitor exposure, and advanced disease at diagnosis. Shotgun metagenomic microbiome analysis revealed 4 species that were significantly expanded with hyperglycemia/hyperinsulinemia, and 2 species that were differentially more prevalent in prediabetic/diabetic PLWH. Collectively, we uncover multiple potential host and microbiome predictors of altered glycemic status in PLWH, while demonstrating that FBG and HbA1c likely suffice for diabetes screening. These potential diabetic predictors merit future prospective validation.

Maternal antibiotic prophylaxis during cesarean section has a limited impact on the infant gut microbiome

Pregnant women undergoing a cesarean section (CS) typically receive antibiotics prior to skin incision to prevent infections. To investigate if the timing of antibiotics influences the infant gut microbiome, we conducted a randomized controlled trial (NCT06030713) in women delivering via a scheduled CS who received antibiotics either before skin incision or after umbilical cord clamping. We performed a longitudinal analysis on 172 samples from 28 infants at 8 post-birth time points and a cross-sectional analysis at 1 month in 79 infants from 3 cohorts. Although no significant associations with bacterial composition, metabolic pathways, short-chain fatty acids, and bile acids were found, we observed subtle differences between the groups at the bacterial strain level and in the load of antibiotic resistance genes. Rather, feeding mode was a predominant and defining factor impacting infant microbial composition. In conclusion, antibiotic administration during CS has only limited effects on the early-life gut microbiome.

Longitudinal single-cell data informs deterministic modelling of inflammatory bowel disease

Single-cell-based methods such as flow cytometry or single-cell mRNA sequencing (scRNA-seq) allow deep molecular and cellular profiling of immunological processes. Despite their high throughput, however, these measurements represent only a snapshot in time. Here, we explore how longitudinal single-cell-based datasets can be used for deterministic ordinary differential equation (ODE)-based modelling to mechanistically describe immune dynamics. We derived longitudinal changes in cell numbers of colonic cell types during inflammatory bowel disease (IBD) from flow cytometry and scRNA-seq data of murine colitis using ODE-based models. Our mathematical model generalised well across different protocols and experimental techniques, and we hypothesised that the estimated model parameters reflect biological processes. We validated this prediction of cellular turnover rates with KI-67 staining and with gene expression information from the scRNA-seq data not used for model fitting. Finally, we tested the translational relevance of the mathematical model by deconvolution of longitudinal bulk mRNA-sequencing data from a cohort of human IBD patients treated with olamkicept. We found that neutrophil depletion may contribute to IBD patients entering remission. The predictive power of IBD deterministic modelling highlights its potential to advance our understanding of immune dynamics in health and disease.

Stable tensor neural networks for efficient deep learning

Learning from complex, multidimensional data has become central to computational mathematics, and among the most successful high-dimensional function approximators are deep neural networks (DNNs). Training DNNs is posed as an optimization problem to learn network weights or parameters that well-approximate a mapping from input to target data. Multiway data or tensors arise naturally in myriad ways in deep learning, in particular as input data and as high-dimensional weights and features extracted by the network, with the latter often being a bottleneck in terms of speed and memory. In this work, we leverage tensor representations and processing to efficiently parameterize DNNs when learning from high-dimensional data. We propose tensor neural networks (t-NNs), a natural extension of traditional fully-connected networks, that can be trained efficiently in a reduced, yet more powerful parameter space. Our t-NNs are built upon matrix-mimetic tensor-tensor products, which retain algebraic properties of matrix multiplication while capturing high-dimensional correlations. Mimeticity enables t-NNs to inherit desirable properties of modern DNN architectures. We exemplify this by extending recent work on stable neural networks, which interpret DNNs as discretizations of differential equations, to our multidimensional framework. We provide empirical evidence of the parametric advantages of t-NNs on dimensionality reduction using autoencoders and classification using fully-connected and stable variants on benchmark imaging datasets MNIST and CIFAR-10.

The promise of explainable deep learning for omics data analysis: Adding new discovery tools to AI

Deep learning has already revolutionised the way a wide range of data is processed in many areas of daily life. The ability to learn abstractions and relationships from heterogeneous data has provided impressively accurate prediction and classification tools to handle increasingly big datasets. This has a significant impact on the growing wealth of omics datasets, with the unprecedented opportunity for a better understanding of the complexity of living organisms. While this revolution is transforming the way these data are analyzed, explainable deep learning is emerging as an additional tool with the potential to change the way biological data is interpreted. Explainability addresses critical issues such as transparency, so important when computational tools are introduced especially in clinical environments. Moreover, it empowers artificial intelligence with the capability to provide new insights into the input data, thus adding an element of discovery to these already powerful resources. In this review, we provide an overview of the transformative effects explainable deep learning is having on multiple sectors, ranging from genome engineering and genomics, from radiomics to drug design and clinical trials. We offer a perspective to life scientists, to better understand the potential of these tools, and a motivation to implement them in their research, by suggesting learning resources they can use to move their first steps in this field.

A primer on correlation-based dimension reduction methods for multi-omics analysis

The continuing advances of omic technologies mean that it is now more tangible to measure the numerous features collectively reflecting the molecular properties of a sample. When multiple omic methods are used, statistical and computational approaches can exploit these large, connected profiles. Multi-omics is the integration of different omic data sources from the same biological sample. In this review, we focus on correlation-based dimension reduction approaches for single omic datasets, followed by methods for pairs of omics datasets, before detailing further techniques for three or more omic datasets. We also briefly detail network methods when three or more omic datasets are available and which complement correlation-oriented tools. To aid readers new to this area, these are all linked to relevant R packages that can implement these procedures. Finally, we discuss scenarios of experimental design and present road maps that simplify the selection of appropriate analysis methods. This review will help researchers navigate emerging methods for multi-omics and integrating diverse omic datasets appropriately. This raises the opportunity of implementing population multi-omics with large sample sizes as omics technologies and our understanding improve.

Moving beyond DNA: towards functional analysis of the vaginal microbiome by non-sequencing-based methods

Over the last two decades, sequencing-based methods have revolutionised our understanding of niche-specific microbial complexity. In the lower female reproductive tract, these approaches have enabled identification of bacterial compositional structures associated with health and disease. Application of metagenomics and metatranscriptomics strategies have provided insight into the putative function of these communities but it is increasingly clear that direct measures of microbial and host cell function are required to understand the contribution of microbe-host interactions to pathophysiology. Here we explore and discuss current methods and approaches, many of which rely upon mass-spectrometry, being used to capture functional insight into the vaginal mucosal interface. In addition to improving mechanistic understanding, these methods offer innovative solutions for the development of diagnostic and therapeutic strategies designed to improve women's health.

Personalized microbiome-driven effects of non-nutritive sweeteners on human glucose tolerance

Non-nutritive sweeteners (NNS) are commonly integrated into human diet and presumed to be inert; however, animal studies suggest that they may impact the microbiome and downstream glycemic responses. We causally assessed NNS impacts in humans and their microbiomes in a randomized-controlled trial encompassing 120 healthy adults, administered saccharin, sucralose, aspartame, and stevia sachets for 2 weeks in doses lower than the acceptable daily intake, compared with controls receiving sachet-contained vehicle glucose or no supplement. As groups, each administered NNS distinctly altered stool and oral microbiome and plasma metabolome, whereas saccharin and sucralose significantly impaired glycemic responses. Importantly, gnotobiotic mice conventionalized with microbiomes from multiple top and bottom responders of each of the four NNS-supplemented groups featured glycemic responses largely reflecting those noted in respective human donors, which were preempted by distinct microbial signals, as exemplified by sucralose. Collectively, human NNS consumption may induce person-specific, microbiome-dependent glycemic alterations, necessitating future assessment of clinical implications.