Comparison between 10X Genomics Xenium and NanoString CosMx

Data published by the Martelotto lab

There has recently been some noise in the Spatial Biology world, given that there is a legal dispute in the field, specifically involving 10X Genomics, NanoString, and other companies. This dispute is related to allegations of patent infringement, where 10X Genomics claims that NanoString and other companies have used its patented technology or methods without authorization or proper licensing.

Spatial Biology is a scientific field that focuses on studying the biological and molecular aspects of cells and tissues within their spatial context. It involves techniques and tools that enable researchers to examine the organization and interactions of molecules within a tissue sample.

10x Genomics is a leading company in the field of spatial biology, which is a rapidly growing field that allows scientists to study the location and expression of genes in tissues and organs. NanoString is another major player in the spatial biology market.

In recent months, 10x Genomics has filed a series of patent infringement lawsuits against NanoString and other companies, alleging that their spatial biology products infringe on 10x Genomics' patents. Many of the alleged infringements involve the Xenium and CosMx product lines of the two companies.

Xenium is 10x Genomics' spatial gene expression platform, while CosMx is NanoString's high-resolution spatial biology platform. Both platforms allow scientists to map the expression of hundreds or even thousands of genes at once in a single tissue sample.

10x Genomics has accused NanoString of infringing on its patents related to a variety of spatial biology technologies, including:

• Spatial barcoding: This technology allows scientists to identify the location of individual cells in a tissue sample.

• Sequencing by synthesis: This technology is used to sequence the DNA or RNA from a tissue sample.

• Image analysis: This technology is used to analyze the images produced by spatial biology instruments.

Side-by-side comparison

The Martelotto lab recently published a side-by-side comparison of the Xenium and CosMx spatial biology technologies, concluding that Xenium outperforms CosMx in all tested metrics.

https://www.martelottolab.org/s/Martelotto-et-al-CosMx-Xenium-Comparison-v2.pdf

While CosMx had an initial advantage in the number of genes assayed (1000 vs. 377 for Xenium), Xenium was found to be superior in the following areas:

• Sensitivity: Xenium was able to detect more genes than CosMx, even at low expression levels.

• Accuracy: Xenium produced more accurate gene expression measurements than CosMx.

• Reproducibility: Xenium data was more reproducible than CosMx data.

• Spatial resolution: Xenium images had higher spatial resolution than CosMx images.

The Martelotto lab's study also found that the Xenium workflow was significantly faster and less labor-intensive on the data analysis side than the CosMx workflow.

The Xenium workflow took a total of 63-75 hours, while the CosMx workflow took 214-240 hours. This difference is due to the fact that the CosMx workflow requires longer scanning and data processing steps, including:

• Tissue post-sectioning prep: The CosMx workflow requires tissue overnight bake at 60 degrees, which is long process.

• Scanning time: The CosMx workflow requires 96 hours of scan time, which compares badly with the 24 hours for Xenium.

• Data processing time: The CosMx workflow requires multiple data processing steps, which take 20 plus 48 hours, compared to the minimal post-processing steps in the Xenium.

The Xenium workflow is thus a much faster and simpler process.

Another important aspect is that Xenium has a significantly lower false discovery rate (FDR) than CosMx. This means that Xenium is more specific than CosMx in finding the correct genes or transcripts in the sample.

In other words, Xenium is less likely to produce false positive results. A false positive result is a result that indicates that a gene or transcript is expressed when it is actually not expressed.

This is an important advantage because it means that researchers can be more confident in the results that they obtain from Xenium. With CosMx, there is a greater risk of obtaining false positive results, which could lead to inaccurate conclusions.

The lower FDR of Xenium is likely due to a number of factors, including the higher sensitivity and accuracy of the Xenium platform.

Xenium more closely recapitulates the biological understanding of the sample than CosMx. This means that Xenium produces results that are more consistent with what researchers expect to see based on their biological knowledge.

In the image below, we can see that the Xenium data points are more tightly correlated to scRNAseq data than the CosMx data points. This indicates that the Xenium data is more reproducible in scRNAseq results and that the Xenium platform is better at capturing the spatial organization of gene expression in the sample.

In conclusion, Xenium produces results that are more consistent with biological expectations. This is a major advantage of Xenium over CosMx, as it means that researchers can be more confident in the results that they obtain.

Xenium has superior specificity of marker expression across the T-cells in the sample. This means that Xenium is better at distinguishing between different types of T-cells than CosMx.

In the image, we can see that the Xenium data shows clear separation between the different T-cell populations. The CosMx data, on the other hand, shows more overlap between the different T-cell populations. This indicates that Xenium is more specific in its detection of T-cell markers.

Another issue is that CosMx suffers from a data analysis inconsistency, which is that the FOVs are not stitched together. This can cause cell duplications with data inconsistencies at the boundaries between one FOV and the next one.

This is a significant issue because it can lead to inaccurate results. For example, if a cell is located at the edge of two FOVs and the FOVs are not stitched together, the cell may be counted twice in the data, but without fully overlapping transcripts. This can lead to overestimates of cell numbers and inaccurate gene expression levels.

The data inconsistencies at the boundaries of the FOVs can also make it difficult to interpret the data. For example, if a cell is missing transcripts from one FOV but not the other, it is difficult to know whether the cell is actually missing those transcripts or whether the inconsistency is due to a problem with the data stitching.

The image below shows a clear example of the data analysis issue caused by FOV stitching in CosMx. In the image, you can see arrows pointing to missing transcripts across duplicate cells. This indicates that CosMx has failed to stitch together the FOVs correctly, resulting in duplicate cells with incomplete data.

Conclusions and forward-looking estimations

Below I’ll describe the conclusions about this side-by-side comparison between 10X Genomics Xenium TXG 0.00%↑ and NanoString CosMx NSTG 0.00%↑ .