Skip to content

ScaiDigest Volume 4: Machine Learning Helps Deliver on the Promise of Spatial Omics

ScaiDigest Volume 4

ScaiDigest Volume 4

ScaiDigest Volume 4

In our latest ScaiDigest Volume 4, Sukalp Muzumdar, one of our brilliant minds at Scailyte, takes us on a journey through the evolving landscape of single-cell multiomics, with a special focus on how machine learning is helping us understand  cellular spatial context.


Unlocking New Dimensions: Emerging technologies like spatial ‘omics, CODEX, and imaging mass cytometry (IMC) are filling this gap, enabling researchers to gain unbiased systems-level insights into tissue development and disease.


Machine Learning Unleashed: Sukalp highlights recent papers where machine learning, and particularly transformer-based models, are making a significant impact on cell segmentation, phenotyping, and associating specific cell types with clinical outcomes.


Connecting the Dots: Sukalp spotlights a novel supervised spatial enrichment approach (S3-CIMA), driven by convolutional neural networks. This technique directly associates tumor microenvironment composition with clinical outcomes, opening doors to clinically-relevant insights.


Stay tuned for more insightful updates and a deeper dive into the world of single-cell multiomics.

“As a reader of this newsletter, I am certain that you are aware of how single-cell multiomics has brought about a paradigm shift in our ability to understand and characterize complex biological systems. Nonetheless, conventional single-cell analyses in my opinion have a severe shortcoming, which is their inability to account for cellular spatial context. Cell-cell communication (both direct contact and paracrine signaling) play crucial roles in various disease contexts, with several immunotherapies being targeted at disrupting this class of interaction (e.g. anti-CTLA-4 and anti-PD-1/PD-L1 checkpoint inhibitors) in cancer.

 

While spatial ‘omics and other multiplexed imaging techniques such as Co-Detection by indEXing (CODEX) and imaging mass cytometry (IMC) are bridging this gap, enabling researchers to gain an unbiased systems-level understanding of tissue development and disease, there are several areas of active research and development in the field.

 

A typical image analysis pipeline requires scientists to first demarcate cells from images (segmentation), next to assign these segmented cells to a specific type or phenotype (e.g. to differentiate between tumor and responder immune cells), and finally to understand the association of the measured data with a disease or treatment outcome of interest (such as response to immunotherapy). Machine learning is being exploited in all steps of this analysis pipeline, and here I present a few recent papers which caught my eye in the field.

 

Constituting the first step of this pipeline, cell segmentation was recently the subject of a NeurIPS challenge, with a summary paper being published recently. In general, all approaches used deep learning, with transformer-based models performing the best. Since its introduction in 2017, the transformer architecture has taken the ML world by storm (powering Dall-E and ChatGPT amongst others), and transformer-based models demonstrated amongst the best performance in this challenge, including the best-performing algorithm (T1-osilab/MEDIAR). While this challenge evaluated the broad applicability of such algorithms across diverse image types, in the context of subcellular spatial transcriptomics (as enabled by Stereo-seq and Seq-scope), we typically also have access to the underlying measured ‘omics data (e.g. transcriptomic data), and transformer algorithms such as SCS which incorporate this data are also emerging as important players in the field, and in my opinion are likely to out-perform generalist segmentation algorithms due to their ability to exploit biological knowledge and relationships.

Once cells have been segmented, phenotyping and cell-type assignment is an important task for further downstream analyses, followed by an association of specific cell types with an endpoint of interest.

Use of S3-CIMA to discover cells associated with an outcome of interest.
Use of S3-CIMA to discover cells associated with an outcome of interest.

As described in a recent paper from our scientific co-founder Prof. Manfred Claassen, association with an endpoint of interest has heretofore been typically carried out in an unsupervised approach where proportions of cells assigned to pre-determined cell types can be compared across an outcome of interest (e.g. responders vs. non-responders to a particular immunotherapy). The paper presents a novel supervised spatial enrichment approach (S3-CIMA), where a convolutional neural network is able to directly associate tumor microenvironment composition (either globally, or within the proximity of anchors defined either by cell type or a functional marker) with a clinical outcome of interest. The technique described was used with great success on CODEX images to unravel the complex interaction between MAIT cells and PD-L1+ tumor-associated macrophages in the context of hepatocellular carcinoma, which further helped highlight them as potential drivers of PD–1/PD-L1-directed immune checkpoint blockade. I find this a great showcase of the power of such machine learning techniques to draw clinically-relevant insights into disease processes.

 

In conclusion, I see multiplexed imaging and spatial ‘omics representing a natural progression from single-cell techniques enabling a closer characterization of all-too-crucial cellular context, helping us to further understand cell-cell communication and spatial organization in development and disease.”

About Scailyte

Scailyte is an ETH Zürich spin-off with a best-in-class artificial intelligence platform for the discovery of complex disease patterns from single-cell data. Our solution provides unprecedented insight into the disease and patients’ biology and enables the discovery of new clinically-relevant biomarker signatures by uncovering human’s hidden “single-cell” secrets. 

Scailyte’s proprietary best-in-class data analysis platform ScaiVision™ associates multimodal single-cell datasets (RNA-/TCR-/BCR-seq, proteomics, etc.) with clinical endpoints, such as disease diagnosis, progression, severity, treatment response, and toxicity response to identify ultra-sensitive biomarker signatures and cell functionality states. The performance and clinically-relevant applications of Scailyte’s platform ScaiVision have been demonstrated in well established CAR-T cell therapies and various clinical projects in Oncology and Immunology.

For more information, visit www.scailyte.com and connect on social media @LinkedIn and @Twitter.

ScailyteTM and ScaiVisionTM are registered trademarks proprietary to Scailyte AG.

Recent News

Scailyte at BIO International Convention 2024

We are thrilled to attend the Bio International Convention 2024 in San Diego on June 3-6 where busin...

Read more

Recent News

Inside Scailyte - Episode 10: Meet Dennis Göhlsdorf

In our latest episode of "Inside Scailyte," we're excited to introduce you to Dennis Göhlsdorf, our...

Read more

Recent News

Precision Med TRI-CON 2024

Join us for our presentation on "Enabling Precision Medicine by Combining Single-Cell Multi-Omics Da...

Read more

Recent News

Inside Scailyte - Episode 9: Meet Shaoline Sheppard

In the latest episode of "Inside Scailyte," we're thrilled to introduce you to Shaoline Sheppard, ou...

Read more

Recent News

NextGen Omics US 2024

Scailyte is headed to NextGen Omics US 2024 in Boston, MA on March 21-22. Join us as we showcase our...

Read more

Recent News

AI-driven single-cell data analysis identifies a cell signature predictive of neurotoxicity and clinical response in CAR-T cell therapy of DLBCL

In our latest white paper, "AI-driven single-cell data analysis identifies a cell signature predicti...

Read more

Recent News

Inside Scailyte - Episode 8: Meet Sajid Maqbool

Meet Sajid Maqbool, our Information Technology Engineer at Scailyte AG. In our latest episode of "In...

Read more

Recent News

Inside Scailyte – Episode 7: Meet Anna Dimitrova

Meet Anna Dimitrova, our Chief Financial Officer at Scailyte AG. In our latest episode of "Inside Sc...

Read more

Recent News

01 /04

Scailyte at BIO International Convention 2024

We are thrilled to attend the Bio International Convention 2024 in San Diego on June 3-6 where busin...

Read more

Inside Scailyte - Episode 10: Meet Dennis Göhlsdorf

In our latest episode of "Inside Scailyte," we're excited to introduce you to Dennis Göhlsdorf, our...

Read more

Recent News

02 /04

Precision Med TRI-CON 2024

Join us for our presentation on "Enabling Precision Medicine by Combining Single-Cell Multi-Omics Da...

Read more

Inside Scailyte - Episode 9: Meet Shaoline Sheppard

In the latest episode of "Inside Scailyte," we're thrilled to introduce you to Shaoline Sheppard, ou...

Read more

Recent News

03 /04

NextGen Omics US 2024

Scailyte is headed to NextGen Omics US 2024 in Boston, MA on March 21-22. Join us as we showcase our...

Read more

AI-driven single-cell data analysis identifies a cell signature predictive of neurotoxicity and clinical response in CAR-T cell therapy of DLBCL

In our latest white paper, "AI-driven single-cell data analysis identifies a cell signature predicti...

Read more

Recent News

04 /04

Inside Scailyte - Episode 8: Meet Sajid Maqbool

Meet Sajid Maqbool, our Information Technology Engineer at Scailyte AG. In our latest episode of "In...

Read more

Inside Scailyte – Episode 7: Meet Anna Dimitrova

Meet Anna Dimitrova, our Chief Financial Officer at Scailyte AG. In our latest episode of "Inside Sc...

Read more

Recent News