Mobile Menu

The State Of: Multi-Omics

This feature was written by Taylor Fulton-Ward.

The advent of multi-omics, or the combination of different omics for analysis, has created immense potential for biological discovery. Over recent years, there has been significant expansion in the technologies and models available for use in multi-omics. Importantly, these methodologies have, and may further contribute to, substantial developments in all aspects of life sciences.  

In this feature, we explore the history and current state of multi-omics. We highlight the newest developments within the field and discuss potential future advances.  

The past

Development of different omic approaches

Multi-omics describes the combination of different omic approaches within one set of analyses. These include genomics, transcriptomics, proteomics, epigenomics and metabolomics, among others. Thus, to understand the history of multi-omics let’s first take a look at how each individual omic approach came to be developed. 

A selection of the multi-omics approaches that are currently available to researchers. Image credit: Roychowdhury, et al., 2023.  

Genomics

In 1871, Friedrich Miescher identified the presence of nuclein, now known to be DNA, in the cell nucleus. Roughly 30 years later, Sutton and Boveri defined the Chromosome Theory of Heredity, and Kossel was awarded the Nobel Prize in Physiology and Medicine for the discovery of the five nucleotide bases: A, C, G, T and U. In 1950, Erwin Chargaff expanded on this knowledge to define the pairing of nucleotide bases.

A few years later, Hershey and Chase demonstrated that DNA, and not protein, is responsible for carrying genetic information, and in 1953, Watson and Crick discovered the double helix structure of DNA. Marshall Nirenberg, Har Gobind Khorana, and their research team were awarded a Nobel Prize in 1968 for their contributions to the understanding of the “codon” of DNA, alongside Holley for sequencing transfer RNA (tRNA).  

This baseline knowledge led to Frederick Sanger sequencing the first complete genome of phiX174 virus in 1977. Alongside Walter Gilbert and Paul Berg, he was awarded a Nobel Prize in 1980 for laying the foundations of DNA sequencing technology.  

Over the next 10 years, various sequencing methodologies were developed and in 1990 the Human Genome Project was launched. In 2003, the first sequence of the human genome was generated and opened the doors to enormous potential for biological discovery. 

Transcriptomics

Transcriptomics, the study of RNA transcripts, was first established in the 1990s with the use of Sanger-sequencing techniques. Throughout the 1990s and early 2000s, micro-arrays and RNA-sequencing were developed, the latter of which allowed for extensive profiling of the human transcriptome. Modern day applications of transcriptomics are being deployed using single-cell and spatial methods, capturing thousands of RNA transcripts across thousands of individual cells.  

Proteomics

The study of the interaction, function, structure and composition of proteins was defined as ‘proteomics’ by Marc Wilkins in 1995. Various proteomic techniques have been developed since, for example, the advancement of gel-based arrays. However, for high-throughput expression analysis, mass spectrometry is generally utilised. The use of mass spectrometry for proteomics was established by Andersen and Mann in 2000 and is now widely employed as a sensitive tool for protein identification.  

Epigenomics

In 1942, Conrad Waddington used the term “epigenetic landscape” to describe how genes interact with their environment to cause heritable changes in gene expression, without impacting the underlying DNA sequence. The field of epigenomics was later established to study the effects of chromatin structure, histone modifications and DNA methylation on gene regulation.  

Early studies used next generation sequencing technologies, which have been developed and revolutionised in recent years, leading to large-scale epigenetic studies at single-cell and even spatial resolution.  

Metabolomics

The concept of biological fluids depicting the health of an individual has been reported throughout history. For example, 3,000 years ago, doctors in ancient China used ants to assess the glucose levels of urine. During the 1940s, Roger Williams reported a metabolic pattern when studying biological fluids. In 1971, Horning and colleagues used gas chromatography-mass spectrometry to measure the metabolic profile of urine and tissue extracts.  

Metabolomics, the study of metabolites, was first introduced as a term in 1998. Since the initial use of metabolomics in the plant research community, techniques for metabolomics have significantly increased, including the use of mass spectrometry and chromatography, and more recently, nuclear magnetic resonance spectroscopy. In 2005, the METLIN database was developed to characterise over 10,000 metabolites by mass spectrometry. The Human Metabolome Project was concluded in 2007 to provide a comprehensive source of metabolic data.  

Combining individual omic datasets to develop multi-omics

Each single omic technology provides a wealth of information for biological research discovery. However, since each omic is often complementary to one another, combining these data – termed multi-omics -provides a greater depth to analysis. As an example, early multi-omics research included the combination of RNA- and ChiP-sequencing data of head and neck squamous cell carcinoma cell lines to identify cancer-specific histone marks associated with driver genes. Additionally, proteomics data was combined with genomic and transcriptomic data to identify driver genes in rectal and colon cancer. These studies demonstrated the importance of combining different omic approaches and began the drastic expansion in the multi-omics field.  

The present

Single cell vs. spatial multi-omics 

Of note, single cell multi-omics refers to analysis of individual cells in isolation, whilst spatial multi-omics additionally preserves the spatial context of cells within tissues. Last year, we developed a feature that discusses the differences between single cell and spatial multi-omics in further detail, which can be found here.  

Integration of multi-omics

Integration of multi-omics refers to the combination of omic data to relate each level of information to one another. Depending on the samples and data sets available, the method for integration differs. You can find a detailed overview of integration strategies for multi-omics here.  

In short, the integration strategy used will differ depending on whether the data is matched (recorded from the same cell) or unmatched (recorded from different cells). Additionally, depending on which omic datasets exist for each sample, integration can be classified as horizontal, vertical, diagonal or mosaic.  

Recent developments 

Over recent years, substantial developments in both single cell and spatial multi-omics approaches have been observed, as well as in the construction of various integration techniques. Let’s take a look at some recent developments in multi-omics over the last 12 months: 

HyperGCN: an effective deep representation learning framework for the integrative analysis of spatial transcriptomics data (Ma et al., 2024.) – This study establishes an unsupervised method based on hypergraph induced graph convolutional network, HyperGCN, as a model for the integrative analysis of spatial transcriptomic data.  

Exploration of mitochondrial autophagy related genes in the diagnosis model construction and molecular marker mining of Alzheimer’s disease based on multi-omics integration (Wang et al., 2024.) – The researchers use integrative analysis to combine DNA methylation and transcriptomic data of Alzheimer’s disease. They used this analysis to produce a diagnostic model and experimentally validate five diagnostic genes for Alzheimer’s disease. 

Single-cell multi-omics and spatial multi-omics data integration via dual-path graph attention auto-encoder (Lv et al., 2024. Preprint) – This study establishes a data integration technique for single-cell and spatial multi-omics data. The model is based on dual-path graph attention auto-encoder (SSGATE).  

Multi-omic profiling reveals potential biomarkers of hepatocellular carcinoma prognosis and therapy response among mitochondria-associated cell death genes in the context of 3P medicine (Hu et al., 2024.) – In this paper, the researchers conducted a comprehensive multi-omic analysis employing single-cell, spatial and bulk transcriptomics in hepatocellular carcinoma. They built a novel predictive model for hepatocellular carcinoma, defined as the mitochondrial cell death index.  

A multi-omic single-cell landscape of cellular diversification in the developing human cerebral cortex (Tian et al., 2024.) – This study conducted a multi-omic analysis on gene expression and open chromatin data from the same cell. The study aimed to build an understanding of the landscape of the developing human neocortex.  

Deep pan-cancer analysis and multi-omics evidence reveal that ALG3 inhibits CD8+ T cell infiltration by suppressing chemokine secretion and is associated with 5-fluorouracil sensitivity (Wu et al., 2024.) – Multi-omic analysis across different cancer types revealed ALG3 to be a diagnostic and predictive biomarker. ALG3 was also shown to regulate immune infiltration and sensitivity to 5-fluorouracil.  

Visual Analysis of Multi-Omics Data (Swart et al., 2024. Preprint) – Swart et al. establish a novel tool to visually analyse up to four types of omic data simultaneously. The tool enables the user to visualise the metabolic reactions, pathways and metabolites of a single organism. 

mastR: Marker Automated Screening Tool for multi-omics data (Chen et al., 2024. Preprint) – This paper presents a new tool, mastR, which can be used for accurate marker identification from omics data. 

From FLG

In January, we released our Multi-omics Playbook, an annual intelligence report gathering insights from experts from across the multi-omics space. It’s still available for download here.  

Last month, we also hosted a 3-part webinar series discussing the recent developments in multi-omics. These webinars included:  

Webinar 1: Multi-Omics Applications in Human Health and Disease 

Webinar 2: The Single-Cell and Spatial Multi-Omics Toolbox 

Webinar 3: The Latest Multi-Omics Data Integration Strategies 

You can sign up to watch these webinars on demand here.   

The future

Despite the significant developments made to date, multi-omic technologies can be expected to continue to advance in several aspects. These include an improvement in throughput and a reduction in cost, which subsequently will improve accessibility. The provision of omic datasets and multi-omic analysis on online archives will further aid data access. It is likely researchers will be able to combine a greater number of omic modalities to contribute to a deeper understanding of biology.  

Technologies are expected to improve in their sensitivity and specificity in the characterisation of each omic modality. For example, due to antibody-based techniques, proteomic analysis is restricted in terms of the number of proteins that can be detected simultaneously. Whilst unbiased mass spectrometry can remove this limitation, it is currently limited in its ability to be combined with other omic modalities. 

The future of spatial multi-omics is looking particularly exciting. A paper in Science last year predicts the advent of three-dimensional spatial omics techniques, for example, on whole organs or organisms, and the development of spatial-temporal omics. Others suggest the ability to capture ancestral cellular states, alongside transient phenotypes, will also enable significant developments in the field.  

The fast-growing field of multi-omics is likely to continue to add to our knowledge of biology. As technologies improve and accessibility grows, the advent of multi-omics may revolutionise scientific research as we know it.