Memory efficient principal component analysis for the dimensionality reduction of large mass spectrometry imaging data sets

eLIMS: Ensemble Learning-Based Spatial Segmentation of Mass Spectrometry Imaging to Explore Metabolic Heterogeneity

Spatial segmentation is an essential processing method for image analysis aiming to identify the characteristic suborgans or microregions from mass spectrometry imaging (MSI) data, which is critical for understanding the spatial heterogeneity of biological information and function and the underlying molecular signatures. Due to the intrinsic characteristics of MSI data including spectral nonlinearity, high-dimensionality, and large data size, the common segmentation methods lack the capability for capturing the accurate microregions associated with biological functions. Here we proposed an ensemble learning-based spatial segmentation strategy, named eLIMS, that combines a randomized unified manifold approximation and projection (r-UMAP) dimensionality reduction module for extracting significant features and an ensemble pixel clustering module for aggregating the clustering maps from r-UMAP. Three MSI datasets are used to evaluate the performance of eLIMS, including mouse fetus, human adenocarcinoma, and mouse brain. Experimental results demonstrate that the proposed method has potential in partitioning the heterogeneous tissues into several subregions associated with anatomical structure, i.e., the suborgans of the brain region in mouse fetus data are identified as dorsal pallium, midbrain, and brainstem. Furthermore, it effectively discovers critical microregions related to physiological and pathological variations offering new insight into metabolic heterogeneity.

The journey towards clinical adoption of MALDI-MS-based imaging proteomics: from current challenges to future expectations

Among the spatial omics techniques available, mass spectrometry imaging (MSI) represents one of the most promising owing to its capability to map the distribution of hundreds of peptides and proteins, as well as other classes of biomolecules, within a complex sample background in a multiplexed and relatively high-throughput manner. In particular, matrix-assisted laser desorption/ionisation (MALDI-MSI) has come to the fore and established itself as the most widely used technique in clinical research. However, the march of this technique towards clinical utility has been hindered by issues related to method reproducibility, appropriate biocomputational tools, and data storage. Notwithstanding these challenges, significant progress has been achieved in recent years regarding multiple facets of the technology and has rendered it more suitable for a possible clinical role. As such, there is now more robust and extensive evidence to suggest that the technology has the potential to support clinical decision-making processes under appropriate circumstances. In this review, we will discuss some of the recent developments that have facilitated this progress and outline some of the more promising clinical proteomics applications which have been developed with a clear goal towards implementation in mind.

Changes of Mass Spectra Patterns on a Brain Tissue Section Revealed by Deep Learning with Imaging Mass Spectrometry Data

The characteristic patterns of mass spectra in imaging mass spectrometry (IMS) strongly reflect the tissue environment. However, the boundaries formed where different tissue environments collide have not been visually assessed. In this study, IMS and convolutional neural network (CNN), one of the deep learning methods, were applied to the extraction of characteristic mass spectra patterns from training brain regions on rodents' brain sections. CNN produced classification models with high accuracy and low loss rate in any test data sets of mouse coronal sections measured by desorption electrospray ionization (DESI)-IMS and of mouse and rat sagittal sections by matrix-assisted laser desorption (MALDI)-IMS. On the basis of the extracted mass spectra pattern features, the histologically plausible segmentation and classification score imaging of the brain sections were obtained. The boundary imaging generated from classification scores showed the extreme changes of mass spectra patterns between the tissue environments, with no significant buffer zones for the intermediate state. The CNN-based analysis of IMS data is a useful tool for visually assessing the changes of mass spectra patterns on a tissue section, and it will contribute to a comprehensive view of the tissue environment.

Information Entropy-Based Strategy for the Quantitative Evaluation of Extensive Hyperspectral Images to Better Unveil Spatial Heterogeneity in Mass Spectrometry Imaging

Hyperspectral images can be generated from mass spectrometry imaging (MSI) data for the intuitive data visualization purpose. However, hundreds of HSIs can be generated by different dimensionality reduction methods, which poses great challenges in selecting the high-quality images with the best intuitive visualization results of the MSI data. Here, we presented a novel approach that objectively evaluates the image quality of the hyperspectral images. The applicability of this method was demonstrated by analyzing the MSI data acquired from human prostate cancer biopsy samples and mouse brain tissue section, which harbored an intrinsic tissue heterogeneity. Our method was based on the information entropy and contrast measured from image information content and image definition, respectively. The heterogeneity of the MSI data from high-dimensional space was reduced to three-dimensional embeddings and thoroughly evaluated to achieve satisfactory visualization results. The application of information entropy and contrast can be used to choose the optimized visualization results rapidly and objectively from an extensive number of hyperspectral images and be adopted to evaluate and optimize different dimensionality reduction algorithms and their hyperparameter combinations. In conclusion, the information entropy-based strategy could be a bridge between chemometrician and biologists.

Three-dimensional quantitative mass spectrometry imaging in complex system: From subcellular to whole organism

Mass spectrometry imaging (MSI) has been applied for label-free three-dimensional (3D) imaging from position array across the whole organism, which provides high-dimensional quantitative data of inorganic or organic compounds that may play an important role in the regulation of cellular signaling, including metals, metabolites, lipids, drugs, peptides, and proteins. While MSI is suitable for investigation of the spatial distribution of molecules, it has a limitation with visualization and quantification of multiple molecules. 3D-MSI, however, can be applied toward exploring metabolic pathway as well as the interactions of lipid-protein, protein-protein, and metal-protein in complex systems from subcellular to the whole organism through an untargeted methodology. In this review, we highlight the methods and applications of MS-based 3D imaging to address the complexity of molecular interaction from nano- to micrometer lateral resolution, with particular focus on: (a) common and hybrid 3D-MSI techniques; (b) quantitative MSI methodology, including the methods using a stable isotope labeling internal standard (SILIS) and SILIS-free approaches with tissue extinction coefficient or virtual calibration; (c) reconstruction of the 3D organ; (d) application of 3D-MSI for biomarker screening and environmental toxicological research. 3D-MSI quantitative analysis provides accurate spatial information and quantitative variation of biomolecules, which may be valuable for the exploration of the molecular mechanism of the disease progresses and toxicological assessment of environmental pollutants in the whole organism. Additionally, we also discuss the challenges and perspectives on the future of 3D quantitative MSI.

Peak learning of mass spectrometry imaging data using artificial neural networks

Mass spectrometry imaging (MSI) is an emerging technology that holds potential for improving, biomarker discovery, metabolomics research, pharmaceutical applications and clinical diagnosis. Despite many solutions being developed, the large data size and high dimensional nature of MSI, especially 3D datasets, still pose computational and memory complexities that hinder accurate identification of biologically relevant molecular patterns. Moreover, the subjectivity in the selection of parameters for conventional pre-processing approaches can lead to bias. Therefore, we assess if a probabilistic generative model based on a fully connected variational autoencoder can be used for unsupervised analysis and peak learning of MSI data to uncover hidden structures. The resulting msiPL method learns and visualizes the underlying non-linear spectral manifold, revealing biologically relevant clusters of tissue anatomy in a mouse kidney and tumor heterogeneity in human prostatectomy tissue, colorectal carcinoma, and glioblastoma mouse model, with identification of underlying m/z peaks. The method is applied for the analysis of MSI datasets ranging from 3.3 to 78.9 GB, without prior pre-processing and peak picking, and acquired using different mass spectrometers at different centers.

A Critical and Concise Review of Mass Spectrometry Applied to Imaging in Drug Discovery

During the past decade, mass spectrometry imaging (MSI) has become a robust and versatile methodology to support modern pharmaceutical research and development. The technologies provide data on the biodistribution, metabolism, and delivery of drugs in tissues, while also providing molecular maps of endogenous metabolites, lipids, and proteins. This allows researchers to make both pharmacokinetic and pharmacodynamic measurements at cellular resolution in tissue sections or clinical biopsies. Despite drug imaging within samples now playing a vital role within research and development (R&D) in leading pharmaceutical companies, however, the challenges in turning compounds into medicines continue to evolve as rapidly as the technologies used to discover them. The increasing cost of development of new and emerging therapeutic modalities, along with the associated risks of late-stage program attrition, means there is still an unmet need in our ability to address an increasing array of challenging bioanalytical questions within drug discovery. We require new capabilities and strategies of integrated imaging to provide context for fundamental disease-related biological questions that can also offer insights into specific project challenges. Integrated molecular imaging and advanced image analysis have the opportunity to provide a world-class capability that can be deployed on projects in which we cannot answer the question with our battery of established assays. Therefore, here we will provide an updated concise review of the use of MSI for drug discovery; we will also critically consider what is required to embed MSI into a wider evolving R&D landscape and allow long-lasting impact in the industry.

Unsupervised machine learning for exploratory data analysis in imaging mass spectrometry

Imaging mass spectrometry (IMS) is a rapidly advancing molecular imaging modality that can map the spatial distribution of molecules with high chemical specificity. IMS does not require prior tagging of molecular targets and is able to measure a large number of ions concurrently in a single experiment. While this makes it particularly suited for exploratory analysis, the large amount and high-dimensional nature of data generated by IMS techniques make automated computational analysis indispensable. Research into computational methods for IMS data has touched upon different aspects, including spectral preprocessing, data formats, dimensionality reduction, spatial registration, sample classification, differential analysis between IMS experiments, and data-driven fusion methods to extract patterns corroborated by both IMS and other imaging modalities. In this work, we review unsupervised machine learning methods for exploratory analysis of IMS data, with particular focus on (a) factorization, (b) clustering, and (c) manifold learning. To provide a view across the various IMS modalities, we have attempted to include examples from a range of approaches including matrix assisted laser desorption/ionization, desorption electrospray ionization, and secondary ion mass spectrometry-based IMS. This review aims to be an entry point for both (i) analytical chemists and mass spectrometry experts who want to explore computational techniques; and (ii) computer scientists and data mining specialists who want to enter the IMS field. © 2019 The Authors. Mass Spectrometry Reviews published by Wiley Periodicals, Inc. Mass SpecRev 00:1-47, 2019.

Chemometric Strategies for Sensitive Annotation and Validation of Anatomical Regions of Interest in Complex Imaging Mass Spectrometry Data

Imaging mass spectrometry (IMS) is a promising new chemical imaging modality that generates a large body of complex imaging data, which in turn can be approached using multivariate analysis approaches for image analysis and segmentation. Processing IMS raw data is critically important for proper data interpretation and has significant effects on the outcome of data analysis, in particular statistical modeling. Commonly, data processing methods are chosen based on rational motivations rather than comparative metrics, though no quantitative measures to assess and compare processing options have been suggested. We here present a data processing and analysis pipeline for IMS data interrogation, processing and ROI annotation, segmentation, and validation. This workflow includes (1) objective evaluation of processing methods for IMS datasets based on multivariate analysis using PCA. This was then followed by (2) ROI annotation and classification through region-based active contours (AC) segmentation based on the PCA component scores matrix. This provided class information for subsequent (3) OPLS-DA modeling to evaluate IMS data processing based on the quality metrics of their respective multivariate models and for robust quantification of ROI-specific signal localization. This workflow provides an unbiased strategy for sensitive annotation of anatomical regions of interest combined with quantitative comparison of processing procedures for multivariate analysis allowing robust ROI annotation and quantification of the associated molecular histology.

Deep data analysis via physically constrained linear unmixing: universal framework, domain examples, and a community-wide platform

Many spectral responses in materials science, physics, and chemistry experiments can be characterized as resulting from the superposition of a number of more basic individual spectra. In this context, unmixing is defined as the problem of determining the individual spectra, given measurements of multiple spectra that are spatially resolved across samples, as well as the determination of the corresponding abundance maps indicating the local weighting of each individual spectrum. Matrix factorization is a popular linear unmixing technique that considers that the mixture model between the individual spectra and the spatial maps is linear. Here, we present a tutorial paper targeted at domain scientists to introduce linear unmixing techniques, to facilitate greater understanding of spectroscopic imaging data. We detail a matrix factorization framework that can incorporate different domain information through various parameters of the matrix factorization method. We demonstrate many domain-specific examples to explain the expressivity of the matrix factorization framework and show how the appropriate use of domain-specific constraints such as non-negativity and sum-to-one abundance result in physically meaningful spectral decompositions that are more readily interpretable. Our aim is not only to explain the off-the-shelf available tools, but to add additional constraints when ready-made algorithms are unavailable for the task. All examples use the scalable open source implementation from https://github.com/ramkikannan/nmflibrary that can run from small laptops to supercomputers, creating a user-wide platform for rapid dissemination and adoption across scientific disciplines.

Modeling the effects of material properties on tablet compaction: A building block for controlling both batch and continuous pharmaceutical manufacturing processes

As the pharmaceutical industry modernizes its manufacturing practices and incorporates more efficient processing approaches, it is important to reevaluate which process design elements affect product quality and the means to study these systems. The purpose of this work is to provide insight on a methodology to correlate the effect of raw material properties to equipment and process performance using both data-driven and semi-empirical models. In this work, lubricated blends of pharmaceutically-relevant materials were made using varying levels of magnesium stearate, ranging from 0.25 to 1.5%. Materials characterization (e.g., compressibility, permeability, density, particle size) was performed for all materials and blends. The blends were compressed using a two by three experimental design, varying tablet fill cam depth and tablet thickness, respectively. Tablet properties (e.g., weight, tensile strength, and thickness) were collected for all tablets. Using the collected tablet property results, models coefficients for the semi-empirical Kuentz and Leuenberger equation, which relates the tablet tensile strength to changes in porosity, were regressed. Empirical models were then developed to correlate the values of the Kuentz and Leuenberger equation coefficients to the blend material properties. The empirical models were then used in conjunction with the Kuentz and Leuenberger equation to evaluate the compression design and operational space, accounting for material properties. This proof of concept work aimed at developing correlations between raw material properties and unit operation models can aid process development, especially in design space characterization and robustness analysis.

Interactive Visual Exploration of 3D Mass Spectrometry Imaging Data Using Hierarchical Stochastic Neighbor Embedding Reveals Spatiomolecular Structures at Full Data Resolution

Technological advances in mass spectrometry imaging (MSI) have contributed to growing interest in 3D MSI. However, the large size of 3D MSI data sets has made their efficient analysis and visualization and the identification of informative molecular patterns computationally challenging. Hierarchical stochastic neighbor embedding (HSNE), a nonlinear dimensionality reduction technique that aims at finding hierarchical and multiscale representations of large data sets, is a recent development that enables the analysis of millions of data points, with manageable time and memory complexities. We demonstrate that HSNE can be used to analyze large 3D MSI data sets at full mass spectral and spatial resolution. To benchmark the technique as well as demonstrate its broad applicability, we have analyzed a number of publicly available 3D MSI data sets, recorded from various biological systems and spanning different mass-spectrometry ionization techniques. We demonstrate that HSNE is able to rapidly identify regions of interest within these large high-dimensionality data sets as well as aid the identification of molecular ions that characterize these regions of interest; furthermore, through clearly separating measurement artifacts, the HSNE analysis exhibits a degree of robustness to measurement batch effects, spatially correlated noise, and mass spectral misalignment.

Label-free molecular imaging of the kidney

In this review, we will highlight technologies that enable scientists to study the molecular characteristics of tissues and/or cells without the need for antibodies or other labeling techniques. Specifically, we will focus on matrix-assisted laser desorption/ionization imaging mass spectrometry, infrared spectroscopy, and Raman spectroscopy.

MALDI Imaging of Liquid Extraction Surface Analysis Sampled Tissue

Combined mass spectrometry imaging methods in which two different techniques are executed on the same sample have recently been reported for a number of sample types. Such an approach can be used to examine the sampling effects of the first technique with a second, higher resolution method and also combines the advantages of each technique for a more complete analysis. In this work matrix-assisted laser desorption ionization mass spectrometry imaging (MALDI MSI) was used to study the effects of liquid extraction surface analysis (LESA) sampling on mouse brain tissue. Complementary multivariate analysis techniques including principal component analysis, non-negative matrix factorization, and t-distributed stochastic neighbor embedding were applied to MALDI MS images acquired from tissue which had been sampled by LESA to gain a better understanding of localized tissue washing in LESA sampling. It was found that MALDI MS images could be used to visualize regions sampled by LESA. The variability in sampling area, spatial precision, and delocalization of analytes in tissue induced by LESA were assessed using both single-ion images and images provided by multivariate analysis.

Mass Spectrometry in Plant-omics

Plant-omics is rapidly becoming an important field of study in the scientific community due to the urgent need to address many of the most important questions facing humanity today with regard to agriculture, medicine, biofuels, environmental decontamination, ecological sustainability, etc. High-performance mass spectrometry is a dominant tool for interrogating the metabolomes, peptidomes, and proteomes of a diversity of plant species under various conditions, revealing key insights into the functions and mechanisms of plant biochemistry.

Nonresonant, femtosecond laser vaporization and electrospray post-ionization mass spectrometry as a tool for biological tissue imaging

An ambient mass spectrometry imaging (MSI) source is demonstrated with both high spatial and mass resolution that enables measurement of the compositional heterogeneity within a biological tissue sample. The source is based on nonresonant, femtosecond laser electrospray mass spectrometry (LEMS) coupled to a quadrupole time-of-flight (QTOF) mass analyzer. No matrix deposition and minimal sample preparation is necessary for the source. The laser, translation stage, and mass spectrometer are synchronized and controlled using a customized user interface. Single or multiple laser shots may be applied to each pixel. A scanning rate of 2.0s per pixel is achieved. Measurement of a patterned ink film indicates the potential of LEMS for ambient imaging with a lateral resolution of ∼60μm. Metabolites including sugar, anthocyanins and other small metabolites were successfully mapped from plant samples without oversampling using a spot size of 60×70μm(2). Molecular identification of the detected analytes from the tissue was enabled by accurate mass measurement in conjunction with tandem mass spectrometry. Statistical analysis, non-negative matrix factorization and principle component analysis, were applied to the imaging data to extract regions with distinct and/or correlated spectral profiles.

Associations between host gene expression, the mucosal microbiome, and clinical outcome in the pelvic pouch of patients with inflammatory bowel disease

BACKGROUND

Pouchitis is common after ileal pouch-anal anastomosis (IPAA) surgery for ulcerative colitis (UC). Similar to inflammatory bowel disease (IBD), both host genetics and the microbiota are implicated in its pathogenesis. We use the IPAA model of IBD to associate mucosal host gene expression with mucosal microbiomes and clinical outcomes. We analyze host transcriptomic data and 16S rRNA gene sequencing data from paired biopsies from IPAA patients with UC and familial adenomatous polyposis. To achieve power for a genome-wide microbiome-transcriptome association study, we use principal component analysis for transcript and clade reduction, and identify significant co-variation between clades and transcripts.

RESULTS

Host transcripts co-vary primarily with biopsy location and inflammation, while microbes co-vary primarily with antibiotic use. Transcript-microbe associations are surprisingly modest, but the most strongly microbially-associated host transcript pattern is enriched for complement cascade genes and for the interleukin-12 pathway. Activation of these host processes is inversely correlated with Sutterella, Akkermansia, Bifidobacteria, and Roseburia abundance, and positively correlated with Escherichia abundance.

CONCLUSIONS

This study quantifies the effects of inflammation, antibiotic use, and biopsy location upon the microbiome and host transcriptome during pouchitis. Understanding these effects is essential for basic biological insights as well as for well-designed and adequately-powered studies. Additionally, our study provides a method for profiling host-microbe interactions with appropriate statistical power using high-throughput sequencing, and suggests that cross-sectional changes in gut epithelial transcription are not a major component of the host-microbiome regulatory interface during pouchitis.

The use of random projections for the analysis of mass spectrometry imaging data

The 'curse of dimensionality' imposes fundamental limits on the analysis of the large, information rich datasets that are produced by mass spectrometry imaging. Additionally, such datasets are often too large to be analyzed as a whole and so dimensionality reduction is required before further analysis can be performed. We investigate the use of simple random projections for the dimensionality reduction of mass spectrometry imaging data and examine how they enable efficient and fast segmentation using k-means clustering. The method is computationally efficient and can be implemented such that only one spectrum is needed in memory at any time. We use this technique to reveal histologically significant regions within MALDI images of diseased human liver. Segmentation results achieved following a reduction in the dimensionality of the data by more than 99% (without peak picking) showed that histologic changes due to disease can be automatically visualized from molecular images.

Potential use of multivariate curve resolution for the analysis of mass spectrometry images

The application of MCR-ALS to mass spectrometry imaging data provides spatial distribution and MS spectra of components, allowing compound identification.

Standardized processing of MALDI imaging raw data for enhancement of weak analyte signals in mouse models of gastric cancer and Alzheimer's disease

Conventional mass spectrometry image preprocessing methods used for denoising, such as the Savitzky-Golay smoothing or discrete wavelet transformation, typically do not only remove noise but also weak signals. Recently, memory-efficient principal component analysis (PCA) in conjunction with random projections (RP) has been proposed for reversible compression and analysis of large mass spectrometry imaging datasets. It considers single-pixel spectra in their local context and consequently offers the prospect of using information from the spectra of adjacent pixels for denoising or signal enhancement. However, little systematic analysis of key RP-PCA parameters has been reported so far, and the utility and validity of this method for context-dependent enhancement of known medically or pharmacologically relevant weak analyte signals in linear-mode matrix-assisted laser desorption/ionization (MALDI) mass spectra has not been explored yet. Here, we investigate MALDI imaging datasets from mouse models of Alzheimer's disease and gastric cancer to systematically assess the importance of selecting the right number of random projections k and of principal components (PCs) L for reconstructing reproducibly denoised images after compression. We provide detailed quantitative data for comparison of RP-PCA-denoising with the Savitzky-Golay and wavelet-based denoising in these mouse models as a resource for the mass spectrometry imaging community. Most importantly, we demonstrate that RP-PCA preprocessing can enhance signals of low-intensity amyloid-β peptide isoforms such as Aβ1-26 even in sparsely distributed Alzheimer's β-amyloid plaques and that it enables enhanced imaging of multiply acetylated histone H4 isoforms in response to pharmacological histone deacetylase inhibition in vivo. We conclude that RP-PCA denoising may be a useful preprocessing step in biomarker discovery workflows.