Multiomics Techniques Advancing Disease Research

New and advanced multiomics tools and technologies have opened new avenues of research and markedly enhanced various disciplines such as disease research and precision medicine¹. The practice of merging diverse data from various ‘omes increasingly provides a more holistic understanding of biological systems. As Maddison Masaeli, Co-Founder and CEO at Deepcell, aptly noted, “You can't explain biology in its complex form with one modality.”

A major leap in the field has been driven by improvements and integrations of various single-cell sequencing and imaging technologies. Baysoy et al’s recent review provides a thorough analysis of how these technologies have been particularly significant in areas such as cell lineage tracing, tissue-specific atlas production, cancer genetics, and more². Advanced computational methods are equally responsible for the surge of interest in multiomics approaches. Seurat v5, GLUE, and LIGER are prime examples of analytical tools developed to overcome the challenges of integrating diverse datasets^3,4,5.

In addition to these core technologies, many newer tools are driving significant progress in the field. They extend beyond the conventional domains typically associated with next-generation sequencing, such as genomics, transcriptomics, and epigenomics. This article examines two of these pivotal technologies and their roles in advancing multiomics research.

Morpholomics
Studying cellular appearance (i.e., morphology) under a microscope has been fundamental for determining cell types, states, and functions, as well as diagnosing diseases and tracking their progression. Many of the traditional methods to interpret cell imagery are either highly subjective, relying on expert interpretation, or limited in quantifiable parameters. Deepcell aims to bridge this gap with their REM-I platform, which integrates deep learning with cell biology to enable more objective assessments of cells based on their visual characteristics.

This new approach, Masaeli explained, is termed “morpholomics” and involves the high-throughput, quantitative, and deep analysis of cell morphology. Masaeli contends that morpholomics, as introduced by the REM-I platform, complements and expands on other aspects of multiomics by offering valuable insights into cell morphology.

Applications, AI, and Advancements
“Wherever you have cells in your workflow, I believe that Deepcell and its platforms can revolutionize the way that you do your work,” stated Masaeli. Since cells are integral components of most biological research, the REM-I’s range of applications is quite extensive. These processes span from basic cell culture practices to advanced diagnostic and staging techniques for various types of disease. In particular, it holds promising potential in fields such as oncology, drug discovery, and cell therapy. The minimal sample preparation required for REM-I’s workflows further increases the possible applications by eliminating the need for staining or labeling.

Another important aspect of the platform is its use of artificial intelligence (AI) and deep learning techniques. “We are taking a step toward bringing AI and deep learning into multiomic analysis,” noted Masaeli. She shared that REM-I offers a reliable, consistent “brain” that continuously analyzes cells and compares them against a vast database of images. Its deep morphological analysis and visual assessment capabilities are based on a model that has trained on millions of images and can identify minute differences that might elude even the most experienced scientists.

Recent advances in Deepcell products including REM-I are highlighted by a transition in the state-of-the-art deep learning technologies they use. The group's initial work relied heavily on convolutional neural networks (CNNs) and supervised learning models. These models required extensive data labeling and were tailored for specific applications that required a foundational understanding of deep learning from users. However, things have changed significantly over the past two years, with a move toward foundation models.

Unlike their predecessors, foundation models rely on a self-supervised learning approach that works to interpret a broad spectrum of data. Foundation models closely mirror human cognitive development, where learning occurs through exposure to diverse stimuli and allows for the recognition and categorization of new objects without the need for extensive training. This advancement enables users to ask different questions about cellular characteristics without needing deep learning expertise and marks an inflection point in Deepcell’s technology.

In addition to the novel morpholomic and multiomic analyses they provide, Masaeli believes that AI and deep learning will also be valuable outside of the life sciences, and is hopeful that they will deliver insights into our complex multimodal world.

Proteomics
The field of proteomics is focused on studying the structures, interactions, and roles of proteins within an organism. Brian Reed, Head of Research at Quantum-Si, explained that proteomics is a critical aspect of multiomics research as proteins provide real-time insights into the body. He noted that while many multiomic studies are reliant on RNA sequencing, proteins are the workhorses of cellular processes and offer a more accurate picture of biological activity.

However, proteomics has historically been a challenging research area. Traditional methods for protein analysis were restricted to technologies like fixed panels that suffer from limited detection capabilities or mass spectrometry, which is costly and requires technical expertise. Understanding the need for improved protein analysis tools in the field, Quantum-Si set out to revolutionize proteomics research similar to how next-generation sequencing transformed genomics.

Their efforts culminated in the release of Platinum®, the Next-Generation Protein Sequencer™. This benchtop instrument simplifies complex protein analyses by sequencing proteins in real-time through the employment of fluorescently labeled amino acid recognizers and aminopeptidases on a semiconductor chip⁶. Reed emphasized that protein sequencing overcomes many of the limitations presented by other proteomic methods and promises to become a vital component of multiomics research workflows.

Integrations, Applications, and Beyond
Following Platinum's launch, the team has observed growing interest in protein sequencing from genomics groups and other researchers not traditionally aligned with the field. “They want to do more than stop at RNA sequencing,” stated Jeff Hawkins, President and CEO of Quantum-Si. This desire comes as no surprise to the Quantum-Si founding team, all experts in multiomics, with many members having prior experience in leading genomics companies.

Protein sequencing has applications in various research settings. For example, genomics labs or those lacking specialized equipment may utilize protein sequencing for protein identification or characterizing complex samples. In large proteomics core labs, protein sequencing can be employed for tasks that are challenging or inefficient with existing instruments. Researchers in these settings might use protein sequencing to investigate amino acid variants, identify specific alterations at certain positions, or detect post-translational modifications. Pharmaceutical companies, biotechnology firms, and contract manufacturers have also shown an interest in protein sequencing for quality assurance and quality control (QA/QC) processes, such as antibody verification.

With these newer, more accessible technologies, multiomics research is becoming mainstream. While smaller groups focused solely on proteomics or genomics may persist due to factors like cost and scalability, Hawkins noted that all the leading research institutions will inevitably gravitate toward multiomics research. Hawkins also shared that they are seeing a growing embrace of new spatial analysis tools and genomic sequencing technologies (both long- and short-read sequencing) to address specific research needs that were, until recently, more limited.

Despite the previous accessibility challenges in proteomics, Reed and Hawkins believe that protein sequencing has removed numerous barriers and soon it will be uncommon to see studies that include RNA sequencing without integrating proteomic data or other multiomic dimensions. This shift is not just about adopting new technologies, but about the holistic understanding they enable. As Hawkins emphasized, “The lab of the future is multiomic.”

References

1. Babu M, Snyder M. Multi-Omics Profiling for Health. Mol Cell Proteomics. 2023;22(6):100561. doi:10.1016/j.mcpro.2023.100561
2. Baysoy A, Bai Z, Satija R, Fan R. The technological landscape and applications of single-cell multi-omics. Nat Rev Mol Cell Biol. 2023;24(10):695-713. doi:10.1038/s41580-023-00615-w
3. Hao Y, Stuart T, Kowalski MH, et al. Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol. Published online May 25, 2023. doi:10.1038/s41587-023-01767-y
4. Cao ZJ, Gao G. Multi-omics single-cell data integration and regulatory inference with graph-linked embedding. Nat Biotechnol. 2022;40(10):1458-1466. doi:10.1038/s41587-022-01284-4
5. Welch JD, Kozareva V, Ferreira A, Vanderburg C, Martin C, Macosko EZ. Single-Cell Multi-omic Integration Compares and Contrasts Features of Brain Cell Identity. Cell. 2019;177(7):1873-1887.e17. doi:10.1016/j.cell.2019.05.006
6. Reed BD, Meyer MJ, Abramzon V, et al. Real-time dynamic single-molecule protein sequencing on an integrated semiconductor device. Science. 2022;378(6616):186-192. doi:10.1126/science.abo7651