Exploiting integrative metabolomics to study host-parasite interactions in Plasmodium infections

Abstract

Despite years of research, malaria remains a significant global health burden, with poor diagnostic tests and increasing antimalarial drug resistance challenging diagnosis and treatment. While ‘single-omics’ based approaches have been instrumental in gaining insight into the biology and pathogenicity of the Plasmodium parasite and its interaction with the human host, a more comprehensive understanding of malaria pathogenesis can be achieved through ‘multi-omics’ approaches. Integrative methods, which combine metabolomics, lipidomics, transcriptomics and genomics datasets, offer a holistic systems biology approach to studying malaria. This review highlights recent advances, future directions, and challenges involved in employing integrative metabolomics approaches to interrogate the interactions between the Plasmodium parasite and the human host, paving the way towards targeted anti-malaria therapeutics and control intervention methods.

Global malaria burden and the advantage of integrative multi-omics

Malaria remains a significant global health burden, with an estimated 249 million cases in 2022 [1]. The emergence of resistance to nearly all frontline drugs presently used against the most virulent (Plasmodium falciparum)[2] and most widespread (P. vivax) [1] human malaria parasite species is a major concern for malaria control and eradication efforts. Various single-omics (see Glossary) based approaches including genomics, transcriptomics, and metabolomics (Figure 1A) have illuminated the etiology of host-parasite immune and metabolic interactions that shape malaria pathogenesis [3–7]. Metabolomics focuses on analyzing metabolites - low molecular weight molecules that are end products of cellular processes. Metabolites are highly reflective of environmental and physiological changes [8], thus providing a sensitive means to monitor metabolic changes in response to Plasmodium infection in vivo, and thereby offering a more accurate and immediate representation of the functional state of the parasite and host cells [8–11]. However, a single-omics approach cannot capture the full complexity of host-parasite interactions. This review highlights the advantages of integrative metabolomics, a multi-omics approach which combines analysis of metabolites with other omics datasets such as genomics (for genetic variation), transcriptomics (for gene expression at the level of transcription), and proteomics (for gene expression at the protein level) to facilitate a more comprehensive and holistic investigation of the molecular mechanisms underpinning host-parasite interactions. In addition, integrative metabolomics has the potential to identify novel therapeutic targets as well as diagnostic and prognostic biomarkers.

Figure 1. Overview of integrative metabolomics and multi-omics approaches for host and parasite.

A) Red panels indicate multi-omics data that can be generated from infected and uninfected human host erythrocytes, white blood cells, and host hepatocytes; yellow panels indicate data that can be generated from infected and uninfected human plasma or serum; blue panel indicates patient meta-data and histology findings, and the green triangle indicates the integration of all data types. B) Workflow of metabolomics data acquisition and processing. Abbreviations: single nucleotide polymorphisms (SNPs), copy number variations (CNVs), next generation sequencing (NGS), expression quantitative trait loci (eQTL), reverse transcription – polymerase chain reaction (RT-PCR), single-cell RNA sequencing (scRNA-seq) liquid chromatography/mass spectrometry (LC/MS), nuclear magnetic resonance (NMR), liquid chromatography tandem/mass spectrometry (LC/MS-MS), gas chromatography/mass spectrometry (GC/MS). Image created with BioRender.com.

Metabolomics and advances in malaria research

Several targeted and untargeted metabolomics studies have been published with the goal of answering questions about how the Plasmodium parasite alters its own metabolic processes, and those of its host, for proliferation, survival, and transmission [9–11], and including how these altered metabolites can be used for further diagnosis and treatment of Plasmodium infections. An in-depth discussion of these metabolomic studies is outside the scope of this review, however, recent representative papers of note include Cobbold et al. [12] who used untargeted mass spectrometry followed by targeted 13-C isotope-tracing metabolomics of sera to identify non-canonical metabolic pathways including four previously uncharacterized enzymes that play a role in isoprenoid biosynthesis, lipid homeostasis, and mitochondrial metabolism, essential for parasite development and proliferation [12]. And Na et al. [13] who attempted to bridge the gap in our understanding of metabolic perturbations induced by P. vivax and P. falciparum. The study revealed species-specific metabolic phenotypes and features, with P. falciparum-infected individuals exhibiting reduced levels of 2,3-diphosphoglycerate and glyceraldehyde-3phosphate, while P. vivax-infected individuals showed reduced levels of retinol and elevated levels of retinoic acid [13]. For all of these studies, a variety of broad-spectrum metabolomic profiling techniques have been developed (Figure 1B), each technique differing in its resolution and the type of molecules it targets.

Metabolomics too has shown potential in parsing apart metabolic signatures and discovering biomarkers based on infection status. Two such recent examples include Cordy et al. [14] who used untargeted high-resolution LC-MS to reveal a set of molecular features (specific amines, carnitines, and lipids) that differentiate acute and chronic infections in P. falciparum-infected humans and P. coatneyi-infected rhesus macaques [14]. And Abdrabou et al. who identified upregulation of steroidogenesis infection associated with coma in cerebral malaria [15]. These studies demonstrate how metabolomics can deliver a wide-ranging catalogue of the metabolic perturbations induced by antimalarial inhibitors and highlights its potential in identifying new lead compounds and understanding their modes of action.

Limitation of metabolomics and the promise of integrative metabolomics

Metabolomic studies have been instrumental in identifying metabolic requirements essential for Plasmodium proliferation and pathogenesis. However, they are not suited for capturing the full spectrum of host metabolic responses and changes, primarily due to variations in host genotype, immune and physiological state, and other environmental factors such as diet [16]. Another major challenge with most metabolomics studies is distinguishing the metabolic signatures of the host from those of the parasite, which is crucial for understanding the implications of metabolic changes in malaria pathogenesis and disease severity. The primary approach to tackle this challenge has been to rely on curated databases of host and parasite metabolic pathways [17]. However, our constantly evolving knowledge of these pathways may not always be reflected in the curated databases used for metabolomics analysis. Colvin et al. recently reported that only about half of the detectable metabolic features in untargeted metabolomics approaches of P. falciparum from in vitro cultures could be successfully identified and mapped to human and Plasmodium KEGG metabolic pathways [11]. Moreover, there is often an overlap between host and parasite metabolic pathways, as well as the metabolites involved in their respective eukaryotic systems [18]. Finally, although metabolites are considered the molecules most reflective of phenotypes in biological systems, handling metabolites can be challenging due to their highly dynamic nature [19]. If not managed properly, metabolites tend to convert to their by-products, and while easier to extract than proteins, their greater instability could potentially lead to false-negative results [20].

In recent years, integrating metabolomic profiling with host and parasite transcriptomic and genomic profiling has emerged as a new approach to address these challenges. This has proven powerful for exploring various aspects of malaria, including the identification of genetic variants responsible for resistance to frontline antimalarial drugs, novel markers for infection severity, and potential targets in the form of pathways, genes, proteins, and metabolites for the development of novel antimalarial therapeutics and interventions [21]. This integrated approach also shows promise in advancing malaria diagnostics and prognostics [22,23], although simple, cheap, and portable field-based tests are some way in the future. Ultimately, the gold standard in Plasmodium-host metabolomic research should involve the integration of multi-omics approaches, including genomics, transcriptomics, lipidomics, and proteomics (see Figure 1) collected as part of (at the very least) a well-designed case-control epidemiology study but also ideally a study design that establishes or advances towards causality (see Box 1).

Box 1. Robust study design for integrative ‘omics projects.

One of the most important first steps in undertaking an integrative metabolomics project to study host-parasite interactions in Plasmodium infections is generating a robust study design. When analyzing host and parasite metabolomic data several variables can influence metabolite abundance, which may or may not be related to disease status. These variables include confounding factors such as subject geographical location, ethnicity, sex, age, diet, complete blood count, and parasitemia. The collection and incorporation of patient meta-data and clinical data in an epidemiological ‘case-control’ study design can significantly mitigate potential biases caused by these confounding variables. Observational case-control studies are those that compare patients with a disease (cases) to those that do not have the disease (controls). If the objective of a study is to observe metabolic differences associated with an immune response elicited by a Plasmodium infection, it is imperative to compare uninfected individuals to infected individuals, and to measure blood cell types and abundance in the two groups since individuals vary in their base level of blood cells including subtypes of white blood cells. Thus, having well designed case-control studies is an important consideration when working with highly variable multi-omics studies, and the addition of meta-data is an important component of a good study design.

While observational case-control studies that generate vast amounts of multi-omics data have a role to play in determining host-parasite associations, moving towards ascertaining cause and effect is the ultimate goal. For this, alternative study designs which establish causality are needed. For example, Mendelian Randomization (MR) [64,65] which makes inferences about causal effects based on data from observational studies. MR is also called ‘Mendelian deconfounding’ because it aims to provide an estimate of causality that is free from biases caused by confounding factors. Other alternative study designs include intervention or clinical trial studies[66] which aim to have prospectively assigned groups of participants that receive the intervention (exposure) compared to those who do not (control) and possible subsequent cross-over between the two groups. And finally experimental medicine studies, which involve an experimental intervention or challenge and are designed to explore disease mechanisms and establish proof of mechanism. For a recent review concerning the next generation of evidence-based medicine, see [67].

Integrative host and parasite metabolomics and transcriptomics

Human immune response

The human metabolomic and immune responses to infection are complexly interconnected, with metabolites such as mono- and polyunsaturated fatty acids and amino acid catabolism intermediates such as kynurenine modulating different compartment of host immune response and consequently determining the course of infection [24–29]. However, the current understanding of the molecular mechanisms underlying the metabolomic modulation of the host immune response during Plasmodium infection is still limited, and the mechanisms by which these changes impact gene expression of different compartments of the immune system are poorly understood. Integrating metabolomics with gene expression holds great promise in unraveling intricate networks of gene-metabolite-phenotype interactions that have the potential to unveil molecular signatures of host responses to Plasmodium infection in vivo [30,31].

Towards this goal, a key study by Abdrabou et al. [3] described multi-omics profiling encompassing metabolomics and transcriptomics to investigate metabolic perturbations associated with P. falciparum infection and its influence on the host adaptive immune response. By integrating metabolomics and blood transcriptomic profiling data derived from a cohort of paired P. falciparum infected and uninfected children in Burkina Faso, they demonstrated that the elevation of pregnenolone and androgen steroids plasma levels has a suppressive immunomodulatory effect on T-cells during infection. They also showed that this effect is mediated by changes in the expression of genes implicated in T-cell exhaustion and activation pathways [3]. This study exemplifies the potential of integrative metabolomics in providing insights into the factors contributing to variation in the human immune response to malaria parasite infection.

Clinical tolerance

Clinical tolerance towards Plasmodium infection is another factor that influences the course and outcome of human infection [32]. Clinical tolerance, a form of acquired immunity that prevents drastic metabolic disruptions and thus preserves life, includes both anti-Plasmodium mechanisms that facilitate parasite clearance, and immunoregulatory mechanisms that limit immunopathology [32]. Integrative metabolomics has proven invaluable in providing insights into the coordinated mechanisms underlying disease tolerance at the molecular and cellular levels [9]. Gardinassi et al. [33] used an integrative metabolomics-transcriptomics approach to identify coordinated gene expression and metabolic mechanisms that differentiate naïve and semi-immune human subjects in a controlled P. vivax study [33,34]. Transcriptomic profiling of whole blood demonstrated that previous exposure to P. vivax was associated with increased gene expression of myeloid cells, blood coagulation, and inflammatory responses, as well as downregulation of genes implicated in T-cell activation. Concurrently, plasma metabolomics showed differences in the host metabolic pathways of previously exposed subjects compared to those who were not infected; naïve individuals showed higher levels of immunomodulatory tryptophan catabolism metabolites such as kynurenine [33]. The integration of both omics datasets made it possible to link the transcriptional profile of myeloid cells and T lymphocytes, and the metabolic response during P. vivax infection. The study also underscored the significant roles of blood coagulation and platelet responses in disease tolerance to Plasmodium infection[33]. In agreement, an independent study by Vinhaes et al. [32] showed elevated levels of proinflammatory cytokines in symptomatic P. falciparum infections and greater perturbation in IL-10 and fibrinogen levels in asymptomatic infections.

Dual transcriptomic profiling of host and parasite

In blood stage malaria

The interconnectedness of host and Plasmodium metabolisms in vivo makes it particularly challenging to distinguish profiles associated with either the host or the parasite. Integrating transcriptomic data is therefore fundamental for determining how human and Plasmodium gene expression impact one another and how this ultimately impacts the metabolome and vice versa during infection (see Figure 2). Yamagishi et al. [35] first described the transcriptomic interaction of the human host and P. falciparum gene expression using a dual Tag RNA sequencing method on peripheral blood from infected subjects. Increased expression of CD163 (a haptoglobin receptor found on the surface of macrophages that plays a role in clearance and endocytosis of hemoglobin complexes) was found in patients with a greater number of P. falciparum RNA Tags, and expression of genes coding for nucleolin, lactotransferrin, RNF34 (ring finger protein 34, E3 ubiquitin protein ligase), and SAMD1 was found to be associated with the severity of infection. There was also increased expression of the P. falciparum pyruvate kinase gene PFF1300w in patients < 20 years of age, correlated with higher disease severity [35].

Figure 2. Summary of data types that can be obtained from vacutainers of whole blood from an individual infected with Plasmodium.

Top panel shows vacutainer with separated plasma (or serum from clotted blood), infected red blood cells, host white blood cells, and peripheral blood mononuclocytes (Buffy coat layer). Plasma or serum can undergo metabolomic or lipidomic profiling, with asexual and sexual life cycle stages of the Plasmodium parasite indicated. Separated host white blood cells can be used for genomic DNA data; histology data can be derived from white and red blood cells; exogenous factors that can influence metabolite and lipidomic abundances are shown in dashed box. Bottom panel indicates separate vacutainer for RNA data with homogenized whole blood, and examples of combining host and parasite metabolomic data with transcript data.. Several examples of metabolites that are up or down-regulated supported by Plasmodium gene expression patterns and the human host response are shown in green, with upwards pointing arrows indicating up regulation/increased response, and downwards pointing red arrows indicating down regulation/suppressed response. Image created with BioRender.com.

More recent studies have gone beyond Tag RNA sequencing and employed whole transcriptome RNA sequencing of host and malaria parasite (dual RNA sequencing). Integrated dual RNA-seq of peripheral blood from 46 Gambian children infected with P. falciparum identified human genes differentially expressed between severe and uncomplicated malaria cases [36], with severe malaria individuals having elevated expression of genes associated with coma, hyperlactatemia, thrombocytopenia, and high expression of neutrophil granule-related genes [36]. Increased gene expression of genes involved in cytoadhesion to vascular endothelium, rigidity of infected erythrocytes, and parasite growth rate, were seen in the P. falciparum transcriptome. Dual RNA-seq also allows analysis of co-expressed genes. For example, P. falciparum infection was shown to increase expression of granulopoiesis and interferon-γ-related genes, which in turn was correlated with decreased gene expression for signaling proteins released by host cells in response to infection [36]. Analysis of transcriptional profiles of P. falciparum isolated from patients suggests that in vivo parasites might exist in distinct physiological states corresponding to stress or starvation, which have not been observed in ex vivo cultures [16]. Similarly, a study by Caamano-Gutierrez et al. [37] identified significant metabolic differences between the in vivo and in vitro life cycles of P. falciparum, such as reduction in growth rates and lengthening of the life cycle.

Several animal model systems have been used for investigating the differences between the gene expression of the host and Plasmodium parasite. For example, the non-human primate Macaca mullatta / monkey malaria parasite species P. cynomolgi model system is used because of the similarities in symptoms to the P. vivax-human host including anemia and thrombocytopenia, and mimics the inter-individual differences of disease severity exhibited in the latter [38]. A multi-omics (transcriptomics, untargeted metabolomics, and targeted lipidomics) of peripheral blood and bone marrow of four infected monkeys collected at consecutive time points over 100 days was able to distinguish between gene signatures of acute and relapsing infections, as well as gene signatures of uninfected monkeys, in an ‘unsupervised’ (i.e., without guidance) manner that was not possible on the separate data types. Furthermore, the transcriptional data were used to determine sub-networks to further illuminate host-parasite interactions [39].

Another model system, the laboratory mouse Mus musculus / rodent malaria parasite species P. berghei, has also been used for investigating the differences between the gene expression of the host and Plasmodium parasite. Previous studies have shown that shown that glucose-6-phosphate (G6PD) deficient humans are somewhat protected against uncomplicated P. vivax infection, and G6PD deficiency reduces disease severity [40]. By using G6PD-deficient mice infected with P. berghei, Yi et al. [41] showed that they had less severe cerebral malaria and milder liver injury compared to wild type. Dual RNA-seq data showed this could be due to genes related to the proinflammatory response [41].

Information can also be gleaned from extant parasite and host public transcriptomic data sets. In a meta-analysis Mukherjee et al. [42] mined studies with single RNA-seq or dual RNA-seq from human and monkey hosts infected with their respective malaria parasite species. From this they derived core networks of co-expressed genes from different tissue and species types. This demonstrated the utility of re-analyzing pre-existing data in order to identify core genes that are differentially regulated during malaria parasite infection [42].

In liver stage malaria

The liver stage of infection is an integral part of the Plasmodium life-cycle, yet it is clinically silent making dual host and parasite RNA-seq a vital tool for discerning gene expression and metabolic needs obligatory for parasite survival and replication. When paired with other detection assays it can be a powerful tool to show localization of gene expression. To illustrate, LaMonte et al. [43] demonstrated that expression of the human mucosal protein Mucin-13 gene is strongly upregulated during Plasmodium infection in the liver stage. Validating these dual-RNA seq results with immunofluorescence assays they showed localization patterns of Mucin-13 protein in hepatic cells that were species independent, revealing that the human response to Plasmodium infection may be conserved regardless of parasite species [43].

Additionally, utilization of RNA-seq date combined with immunofluorescence and protein localization assays have been used to elucidate crucial genes involved in the liver stage. For example, Posfai et al. [44] showed that expression of aquaporin-3 (AQP3) was induced in infected vs. uninfected human liver cells, with human AQP3 localizing to the parasitophorous vacuole membrane during infection rather than the host cell membrane, and indicating that AQP3 is vital for parasite liver stage development [44]. However, while bulk RNA-seq methods characterize both parasite and host liver stage transcriptomes, single-cell transcriptomic resolution is needed to address spaciotemporal differences of the liver during infection. Afriat et al. [45] created a single-cell atlas of the Plasmodium liver stage using the P. berghei model system in which single-cell RNA-seq was combined with single-molecule transcript imaging to characterize expression patterns. Parts of the liver exhibited significantly different patterns of parasite gene expression even at later time points when there was much parasite abundance. Parts of the liver with decreased parasite transcript expression had enrichment of host-related immune recruitment genes (i.e., those related to IFNγ response and increased activation of Notch, p53 and MYC). These host liver cells showed a distinct phenotype in which the PVM was disintegrated, ultimately showing that these cells were distinct from other infected liver cells [45].

Limitations of multi-omics integrative data collection and analysis

As the ability to generate large omics datasets becomes easier, challenges in multi-omics integrative data collection and analysis are magnified. While untargeted, single-omics approaches are suitable for data exploration, such datasets have their drawbacks, ranging from technical limitations (reproducibility and lack of biological replicates) to batch effects when profiling samples at different times. To maximize the robustness of such studies, the minimum sample size required for the biological question or hypothesis under study at a particular statistical power should be determined in advance. Several reviews delve into sample size and power analysis, along with compiled online resources for calculating the minimum effect size for a study question [46,47]. Additionally, there are R packages that can perform simulation-based power analysis [48]. Multiple hypothesis testing too is required since the larger the dataset, the greater the probability of identifying a positive correlation or significant finding by chance[49–51].

Ultimately, some of the most insightful findings can be gleaned by having well-outlined biological questions to begin with, and an understanding of the limitations of the tools employed to analyze them (Box 2). This process begins before any data generation, starting with the study design. For instance, if the aim is to identify a specific metabolite or set of known genes implicated in a pathway, a cost-effective study could use a targeted profiling method, such as hybridization-based target enrichment or targeted NMR. Subsequently employing a one-size-fits-all approach in terms of analysis quantification or data analysis is not necessarily a good approach. However, devising and implementing standard methodologies for consistent data collection and processing of samples, as well as standardization of databases used as a reference for quantification [52] can be highly useful. Such standardization also enables comparison of data sets between regions and study types, as has been successfully implemented in other scientific areas, such as multi-omics and metabolite studies in microbial ecology [53].

Box 2. Bioinformatics tools for integrative metabolomics.

With generation of extensive amounts of omics data there is a demand for bioinformatics tools that can integrate multi-omics data types (recently reviewed in Vahabi et al. [68] and Agamah et al.[69]). Notably, bioinformatics tools such as MetaboAnalyst [70,71], XCMS Online [72], MetExplore [73], OmicsDI [74,75], MetCyc [76], BioCyc [77], MetaboFlow [78], MetaboLights [79], MetExplore [80] and MetaboNetworks [81] offer diverse capabilities for integrating metabolomic data with genomics, transcriptomics, and proteomics datasets to better understand and analyze the complex interactions between metabolites and other biological entities. While these tools offer many advantages in terms of accessibility and functionality, it is still necessary to understand the challenges associated with data integration and analysis due to data heterogeneity, standardization, and algorithmic biases. Additionally, the compatibility and interoperability of these tools with various data formats and platforms can present obstacles in achieving seamless integration. Addressing these limitations will be crucial for refining the accuracy and reliability of integrative metabolomics analyses and harnessing their full potential in unraveling complex biological relationships between the host and parasite.

Another issue that arises as large amounts of multi-omics data are generated is the potential detection of thousands of biological entities (metabolites, proteins, gene transcripts), and hundreds of individual pathways implicated in infection. Additionally, certain omics approaches, such as proteomic and transcriptomic profiling, may exhibit bias towards the detection of only highly abundant proteins and transcripts, leading to biological components with lower expression levels being missed. While there is no single solution for this, pathway and gene set enrichment in addition to the generation of interaction networks can help greatly in identifying and focusing on relevant biological relevant pathways and networks. In addition, the integration of at least two data types, such as metabolomics and RNA-seq data where highly abundant metabolites can be correlated with gene expression patterns to create a more comprehensive view of molecular networks, is a powerful approach. The current surge in machine learning techniques used in malaria research (for example classifying Plasmodium infection status based on blood parameters [54]) can be implemented to refine such searches. None-the-less, this should not replace experimental validation of biomarkers and metabolites, as this remains one of the most reliable strategies for confirming individual molecules of interest.

Concluding remarks and future perspectives

In recent years, the integration of metabolomics into malaria research has offered valuable insights into poorly understood phenomena. Concurrently, our knowledge of Plasmodium genome variation has significantly advanced, particularly with the release of the MalariaGEN P. falciparum genome variation datasets, culminating most recently with the Pf7 dataset [55]. This dataset encompasses genotypic variation data from 20,864 samples collected in 82 studies conducted across 33 countries in Africa, Asia, South America and Oceania. With the continuous evolution of genotypic sequencing techniques, a promising future application is the expansion of quantitative trait loci (QTL) association mapping studies of metabolites (e.g. Lewis et al. [56]) to genome-wide association studies using both host and parasite genome-wide genotypic data (see Outstanding Questions). Employing this unbiased approach can enable the discovery of novel genetic variants impacting metabolic pathways and their associations with disease susceptibility, severity, and treatment outcomes. This, in turn, can help explain the well-documented inter-ethnic variation in host response to infection [57–63]. Additionally, this approach can also provide insights into the molecular mechanisms that underlie how genetic predispositions alter the course of infection. Moreover, the identification of specific genetic variants influencing metabolic profiles could reveal potential targets for personalized anti-malarial interventions and tailored treatment strategies.

Outstanding Questions.

• What strategies can be used to annotate uncharacterized metabolites?

• How can we improve the detection of metabolites, transcripts, and proteins that are less abundant in a biological sample?

• How can we advance beyond the technical limitations of proteomics datasets (which often lack reproducibility) and transcriptomics datasets (which are susceptible to RNA degradation) and both of which are often lacking the appropriate number of biological and technical replicates?

• How do we select the metabolites, genes, proteins, or pathways, for further study, when we find many that are altered by a Plasmodium infection?

• How can we overcome the problem that adding more single ‘omic data types increases cost, and often comes at the expense of decreasing sample size?

In summary, integrative metabolomics offers a comprehensive view of the molecular mechanisms underlying malaria pathogenesis. This approach can reveal how changes in gene expression or protein levels affect metabolic pathways, and vice versa, which can provide deeper insights into the regulation of metabolic pathways essential for Plasmodium parasite survival and growth. Moreover, this approach can shed light on novel host-parasite interactions and metabolic adaptations both organisms undergo to survive and proliferate, enabling the development of strategies to disrupt these interactions and hinder disease progression. Additionally, it can facilitate the identification of Plasmodium species-specific features and differences such as metabolic signatures, drug resistance mechanisms, and host immune responses, leading to the development of tailored anti-malarial interventions for different Plasmodium infections (P. falciparum and P. vivax). However, future studies will require a more standardized pipeline approach for the integration of multi-omics data. Finally, integrative metabolomics improves the accuracy and reliability of findings by corroborating results across different omics data types, reducing false positives and increasing confidence in the identified targets or biomarkers.

Highlights.

Approaches using single omics and targeted metabolomics have highlighted important biological mechanisms of Plasmodium proliferation and pathogenesis. However, such approaches are limited in their capacity to capture the full spectrum of host metabolic responses and changes.

Recent advances in global metabolomic and lipidomic profiling coupled together with other ‘omics approaches, such as transcriptomics and genomics and integration of phenotypic data, allows for greater insights into the specific perturbations associated with host-parasite interaction.

A promising future application is the expansion of QTL association mapping studies of metabolites to genome-wide association studies using both host and parasite genotypic data for the discovery of novel genetic variants impacting metabolic pathways and their associations with disease susceptibility and treatment outcomes.

Ultimately robust study designs that advance beyond association studies towards causality are needed.

Glossary

Biomarker

an indicator of some biological state or condition that can be measured accurately and reproducibly

Bulk RNA sequencing

sequencing of all RNA within a single cell or organism

Dual RNA sequencing

simultaneous sequencing of host and parasite RNA with subsequent mapping and alignment to the respective host and parasite genomes

Genomics

the study of all the genes in a genome, their function, interactions, and evolution

Integrative metabolomics

combining multiple types of data from single omics approaches such as transcriptomics, genomics, etc with metabolomics

Lipidomics

systematic study of changing lipids in a system

Metabolite

single small chemical molecule that makes up parts of greater metabolic pathways

Metabolomics

systematic study of changing chemicals and metabolites in a system. The collection of metabolomic data can be targeted, i.e., measuring defined groups of characterized and biochemically annotated metabolites, or untargeted, i.e., a more broad analysis of all the measurable metabolites in a sample including unknowns

Multi-omics

data generated from more than one omics approach such as genomics, transcriptomics, proteomics, metabolomics, etc

Proteomics

the study of the interactions, function, composition, and structures of proteins and their cellular activities

Single-cell RNA sequencing

sequencing of expressed genes at the individual cell level

Single-omics

data generated from one ‘omic’ approach

Systems biology

a holistic approach to biology through the generation of high dimensional datasets and their integration and analysis using computational power

Transcriptomics

the study of all expressed genes in a single cell or organism