<LinkableSection id={"filter"} title={"Filtering"}>

Secondary analysis only includes transcripts that pass the default quality value (Q-Score) threshold of Q20 and are assigned to cells. Before running any secondary analysis steps, the XOA pipeline filters out cells with fewer than five transcripts. These cells are thus not included in the secondary analysis output files (`analysis/` and `analysis.zarr.zip`).

This filter is _not_ applied to other outputs, such as the cell-feature matrix and transcripts files. Filtered cells will be categorized as "Unassigned" when viewing K-means and graph-based cluster results in Xenium Explorer.

</LinkableSection>

<LinkableSection id={"dim-reduction"} title={"Dimensionality reduction"}>

</LinkableSection>

<LinkableSection id={"pca"} title={"PCA"} nested={true}>

In order to reduce the gene expression matrix to its most important features, XOA uses Principal Components Analysis (PCA) to change the dimensionality of the dataset from (cells x genes) to (cells x _M_) where _M_ is a number of principal components (10 PCs retained). The pipeline uses a python implementation of the IRLBA algorithm, (Baglama & Reichel, 2005), which we modified to reduce memory consumption. Note that if the data contains protein data, only the gene expression data will be used for PCA and subsequent analysis.

</LinkableSection>

<LinkableSection id={"umap"} title={"UMAP"} nested={true}>

XOA also supports visualization with UMAP (Uniform Manifold Approximation and Projection), which estimates a topology of the high dimensional data and uses this information to estimate a low dimensional embedding that preserves relationships present in the data. The pipeline uses the python implementation of this algorithm by McInnes et al (2018). XOA uses the following default parameter values for UMAP generation:

- `n_neighbors` = 30
- `min_dist` = 0.3
- `n_components` = 2
- `spread` = 1.0
- `metric` = "pearson"
- `n_epochs` = 200 or 500
- `learning_rate` = 1.0
- `random_seed` = 0

</LinkableSection>

<LinkableSection id={"clustering"} title={"Clustering"}>

XOA uses two methods for clustering cells based on expression similarity, both operating in the PCA space.

</LinkableSection>

<LinkableSection id={"graph"} title={"Graph-based"} nested={true}>

The graph-based clustering algorithm consists of building a sparse nearest-neighbor graph (where cells are linked if they among the k nearest Euclidean neighbors of one another), followed by Louvain Modularity Optimization (LMO; Blondel, Guillaume, Lambiotte, & Lefebvre, 2008), an algorithm which seeks to find highly-connected "modules" in the graph. The value of _k_, the number of nearest neighbors, is set to scale logarithmically with the number of cells.

Additionally, it performs a cluster-merging step using hierarchical clustering of cluster medoids in PCA space. This step merges pairs of sibling clusters if there are no differentially expressed genes or if present, protein features, between them (with a Benjamini-Hochberg adjusted p-value below 0.05). The hierarchical clustering and merging is repeated until there are no more cluster-pairs to merge. Protein feature information is only used to determine which clusters to merge, and not for the graph-based clustering output itself.

The use of LMO to cluster cells was inspired by a similar method in the R package [Seurat](http://satijalab.org/seurat).

</LinkableSection>

<LinkableSection id={"kmeans"} title={"K-means"} nested={true}>

XOA also performs traditional K-means clustering across a range of _K_ values, where _K_ is the preset number of clusters.

</LinkableSection>

<LinkableSection id={"diffexp"} title={"Differential gene expression"}>

To test for differences in mean expression between groups of cells, XOA uses the exact negative binomial test proposed by the authors of the sSeq method (Yu, Huber, & Vitek, 2013). When the counts become large, XOA switches to the fast asymptotic negative binomial test used in edgeR (Robinson & Smyth, 2007).

For each gene and each cluster _i_, XOA tests whether the mean expression in cluster _i_ differs from the mean expression across all other cells.

The mean expression of a feature for cluster _i_ is calculated as the total number of transcripts from that feature in cluster _i_ divided by the sum of the size factors for cells in cluster _i_. The size factor for each cell is the total transcript count in that cell divided by the median transcript count per cell (across all cells). The mean expression outside of cluster _i_ is calculated in the same way. The log2 fold-change of expression in cluster _i_ relative to other clusters is the log2 ratio of mean expression within cluster _i_ and outside of cluster _i_. When computing the log2 fold-change, a pseudocount of 1 is added to both the numerator and denominator of the mean expression.

Note that XOA's implementation differs slightly from the sSeq paper: the size factors are calculated using total transcript count instead of using DESeq's geometric mean-based definition of library size. As with sSeq, normalization is implicit in that the per-cell size factor parameter is incorporated as a factor in the exact and asymptotic probability calculations.

</LinkableSection>

<LinkableSection id={"refs"} title={"References"}>

- Baglama J and Reichel L. [Augmented implicitly restarted Lanczos bidiagonalization methods](https://epubs.siam.org/doi/abs/10.1137/04060593X). _SIAM Journal on Scientific Computing_ 27: 19–42, 2005.
- Blondel V, et al. [Fast unfolding of communities in large networks](https://iopscience.iop.org/article/10.1088/1742-5468/2008/10/P10008/meta). _Journal of Statistical Mechanics: Theory and Experiment_, 2008.
- McInnes L, Healy J, and Melville J. [UMAP: Uniform Manifold Approximation and Projection for dimension reduction](https://arxiv.org/abs/1802.03426). _arXiv_, 2018.
- Robinson M and Smyth G. [Small-sample estimation of negative binomial dispersion, with applications to SAGE data](https://academic.oup.com/biostatistics/article/9/2/321/353777). _Biostatistics_ 9: 321–332, 2007. [Link to edgeR source](https://bioconductor.org/packages/release/bioc/html/edgeR.html).
- Yu D, Huber W, and Vitek O. [Shrinkage estimation of dispersion in Negative Binomial models for RNA-seq experiments with small sample size](https://academic.oup.com/bioinformatics/article/29/10/1275/259212). _Bioinformatics_ 29: 1275–1282, 2013.

</LinkableSection>

<Flex style={{
 flexDirection: "column",
 gap: "18px",
 padding: "1rem 0 2rem 0"
}}>
<WorkflowCard link={"/support/software/xenium-onboard-analysis/latest/algorithms-overview/xoa-algorithms"} title={"Back to algorithms overview"} />
</Flex>

Gene Expression Secondary analysis

Filtering, dimensionality reduction, clustering, and differential expression algorithms for gene expression data.

Secondary Analysis of Xenium In Situ Gene Expression Data

Image processing begins by mapping known physical features on the glass slide surface (fiducials) to set the image coordinate system. The instrument then identifies the upper bound of the tissue section to find the location of the first Z-slice. These pieces of information collectively determine the range of Z-slices to image. This process has been robust to all tested sections and tissue types.

<LinkableSection id={"dapi"} title={"3D DAPI image"}>

The DAPI image process produces a complete 3D morphology image (the `morphology.ome.tif` output file) for each of the stained regions and determines the reference image volume that cells are segmented from and decoded transcripts are assigned to.

First, the lens distortion in internal sensor data is corrected. This is done computationally based on instrument calibration data, which are collected in order to characterize the optical system and are saved on-instrument. Next, the Z-stacks from internal sensor data (0.75 &micro;m step size) are further subsampled to a 3 &micro;m step size. This subsampling step size was determined empirically to be a useful resolution for [cell segmentation](/support/software/xenium-onboard-analysis/latest/algorithms-overview/segmentation#seg) quality.

Shared image features are then extracted from the regions where FOVs overlap. Feature matching is performed to estimate the offsets between adjoining FOVs. The offsets are used to ensure consistent alignment across the image (global alignment).

Finally, the 3D DAPI image volumes (Z-stacks) generated across FOVs are blended together to construct a stitched volume that is used for nucleus segmentation.

</LinkableSection>

<LinkableSection id={"focus-imgs"} title={"2D projection images"}>

The DAPI and multi-tissue stain image processes produce 2D projections for each image stain (output in the `morphology_focus/` directory). The DAPI focus image is based on the DAPI image captured in the cell segmentation cycle if the Cell Segmentation Staining workflow is used (described below), otherwise it is from cycle 1.

The steps described below also apply for the cell segmentation and protein images acquired following the Xenium In Situ Gene and Protein Expression with Cell Segmentation Staining assay. Additional Xenium Protein image processing algorithm steps are described on this [page](/support/software/xenium-onboard-analysis/latest/algorithms-overview/protein-algorithms).

</LinkableSection>

<LinkableSection id={"focus-multi-tissue"} title={"Cell segmentation stain image acquisition"} nested={true}>

_This algorithm step is performed for tissues prepared with the Multi-Tissue Stain Mix following the Cell Segmentation Staining workflow (see [CG000749](/support/in-situ-gene-expression/documentation/steps/assay/xenium-in-situ-gene-expression-with-morphology-based-cell-segmentation-staining-user-guide), [CG000760](/support/in-situ-gene-expression/documentation/steps/assay/xenium-prime-in-situ-gene-expression-with-optional-cell-segmentation-staining-user-guide)), or [CG000819](/support/in-situ-gene-expression/documentation/steps/assay/xenium-in-situ-gene-and-protein-expression-with-cell-segmentation-staining-user-guide)._

After RNA imaging cycles, two cycles are added to the on-instrument analysis workflow to image the cell membrane boundary stains and interior cell stains for multimodal cell segmentation.

- The first added cycle captures DAPI and the other stain channels in a blank state.
- The second added cycle captures DAPI and the other channels with their respective stains present.

The DAPI images from these additional two cycles are then aligned to the cycle 1 DAPI image so they are in the same coordinate system. Next, the background "blank" images are subtracted from the stain images for each of the multi-tissue stain channels, which helps to offset the effects of autofluorescence. Background subtraction is performed for the multi-tissue stain images, not the DAPI image.

In XOA v3.0, background thumbnail images are displayed in the `analysis_summary.html` Image QC gallery. Downsampled images are available in the `aux_outputs/` directory (i.e., `background_01_[channel].tiff`).

</LinkableSection>

<LinkableSection id={"focus-processing"} title={"Focus image processing"} nested={true}>

This algorithm step is based on multi-focus image fusion techniques\* to generate an all-in-focus stitched, 2D projection for each stain image (i.e., cell segmentation: boundary, interior RNA, and interior protein; DAPI; protein). It includes the following steps:

- In 16 x 16 pixel patches, the algorithm selects the best in-focus Z-slice from the aligned multi-channel stain image Z-stack, which has a 0.75 &micro;m step size and includes the DAPI, background, and cell segmentation stain channels. This process is applied to every FOV selected for analysis.
- The focus finding process then generates a global focus map based on all these channels, which is applied across every channel and cycle.

\*Review articles such as this [one](/support/software/xenium-onboard-analysis/latest/algorithms-overview/image-processing#references) provide some background information about multi-focus image fusion methods.

</LinkableSection>

<LinkableSection id={"deconv-crosstalk"} title={"Deconvolution, background subtraction, masking, and crosstalk correction"} nested={true}>

Seven Z-slices are sampled around the global focus map. Then, deconvolution and background removal are applied, and the middle slice of the volume is selected. This step improves resolution and contrast of images such that they appear sharper and have lower fluorescence intensity values (i.e., when visualized in Xenium Explorer).

Cell segmentation images undergo a masking step that uses information from the background images to identify and mask saturated pixels.

A spectral crosstalk correction is applied to remove signal bleedthrough from other channels for all cell segmentation stain images. Different spectral channels can interfere with each other. This step corrects that bleedthrough by removing the predicted contribution of each of the fluorophores from channels that are not intended to detect them.

Finally, the results are stitched together to generate a 2D projection image for each stain. These processed images are provided in the Xenium Onboard Analysis output bundle as multi-file OME-TIFF files, one per channel, in the `morphology_focus/` directory. The directory will contain the nuclei DAPI stain image, as well as any additional stain images for Xenium outputs generated with the multimodal cell segmentation and protein assay workflows. Outputs are described [here](/support/software/xenium-onboard-analysis/latest/analysis/xoa-output-understanding-outputs#images).

</LinkableSection>

<LinkableSection id={"rca"} title={"RCA product image processing"}>

The goal of RCA product image processing is to detect and filter puncta and correct distortion. Performed for every channel and cycle, the 3D image volumes (Z-stacks) obtained for each FOV are processed to detect the puncta in 3D space that correspond to labeled RCA products. Images are currently captured in four color channels and n cycles depending on the assay.

The RNA fluorescence image is scanned for punctum signals that stand out from the local background. The XYZ coordinates of each punctum are refined by examining local brightness. The signal intensity of the punctum is determined based on a fitted shape. Xenium Analyzer instrument-specific calibrations are made to adjust the optical system’s point spread function (PSF) model. This calibration enables increased detection and localization of puncta in transcript-dense regions and near the edges of FOVs, effectively minimizing FOV-related artifacts like roll-off.

Next, the pipeline filters out puncta that are unlikely to be true transcripts (non-punctate or low quality signals). Similar to DAPI images, curvature distortion is corrected.

<Image src={"https://cdn.10xgenomics.com/image/upload/v1692908783/software-support/Xenium-onboard-analysis/RNAimage.png"} />

</LinkableSection>

<LinkableSection id={"references"} title={"References"}>

- Zafar R, Farid MS, and Khan MH. [Multi-Focus Image Fusion: Algorithms, Evaluation, and a Library](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8321074). J Imaging. 6:60, 2020.

</LinkableSection>

<Flex style={{
 flexDirection: "column",
 gap: "18px",
 padding: "1rem 0 2rem 0"
}}>
<WorkflowCard link={"/support/software/xenium-onboard-analysis/latest/algorithms-overview/xoa-algorithms"} title={"Back to algorithms overview"} />
</Flex>

Image Processing

3D DAPI, 2D projection, and RCA product image processing algorithms.

Image Processing Algorithms

<LinkableSection id={"overview"} title={"Workflow overview"}>

The Xenium In Situ Gene and Protein Expression with Cell Segmentation Staining assay ([CG000819](/support/in-situ-gene-expression/documentation/steps/assay/xenium-in-situ-gene-and-protein-expression-with-cell-segmentation-staining-user-guide)) assays RNA and protein at the subcellular level in human formalin fixed & paraffin embedded (FFPE) tissue sections (referred to as Xenium Protein assay hereafter). The data acquisition steps and algorithms for processing transcript data to generate Xenium Onboard Analysis (XOA) outputs are described on the main Algorithm Overview [page](/support/software/xenium-onboard-analysis/latest/algorithms-overview/xoa-algorithms).

On this page, we describe the data acquisition steps and algorithms for processing protein data to generate XOA outputs. Like the RNA images, the internal image sensor captures data across multiple Z-planes for protein images with a 0.75 µm step size across the entire tissue thickness for every field of view (FOV) in the user-selected region (see Region Selection Guidelines in the [Xenium Analyzer instrument user guide](/support/in-situ-gene-expression/documentation/instruments/xenium-analyzer/xenium-analyzer-user-guide)). Image data are captured for multiple fluorescence channels in every cycle, which are processed and stitched to build a spatial map of the protein expression across the selected tissue section.

<Flex style={{
 justifyContent: "center"
}}>
<Image src={"https://cdn.10xgenomics.com/image/upload/v1752255117/software-support/Xenium-onboard-analysis/v4.0/XeniumProtein_HighLevelWorkflow.jpg"} width={"600px"} length={"600px"} />
</Flex>
 

After the RNA decoding and cell segmentation cycles, the Xenium Protein assay includes up to seven additional cycles to image protein subpanel markers. A second background image is acquired for performing background subtraction on the protein image data.

<Image src={"https://cdn.10xgenomics.com/image/upload/v1752255030/software-support/Xenium-onboard-analysis/v4.0/XeniumProtein_XAcycles.png"} />
 

</LinkableSection>

<LinkableSection id={"xoa"} title={"Onboard analysis pipeline"}>

The XOA v4.0 pipeline introduces support for analysis of protein data. The high-level pipeline steps for the Xenium In Situ Gene and Protein Expression with Cell Segmentation Staining assay (CG000819) are shown in the diagram below:

<Flex style={{
 justifyContent: "center"
}}>
<Image src={"https://cdn.10xgenomics.com/image/upload/v1752255031/software-support/Xenium-onboard-analysis/v4.0/XeniumProteinSoftwareAlgoWorkflow.png"} width={"450px"} length={"450px"} />
</Flex>
 

The following steps are the _same_ as those described for gene expression only datasets (see main Algorithm Overview [page](/support/software/xenium-onboard-analysis/latest/algorithms-overview/xoa-algorithms)): **image processing** (DAPI, cell segmentation images), **nucleus and cell segmentation**, **RCA product image registration**, **decoding**, **deduplication**, and **gene expression secondary analysis**.

XOA v4.0 introduces algorithms to process protein data. The **_protein_ image processing**, **mean intensity**, and **secondary analysis** algorithms are described in the sections below.

</LinkableSection>

<LinkableSection id={"protein-img"} title={"Protein image processing"}>

The protein image process algorithms produce a 2D projection for each image stain (output in the `morphology_focus/` directory).

The DAPI images from these protein cycles are aligned to the cycle 1 DAPI image, so they are in the same coordinate system. Protein images are sampled based on a global focus map, derived from the DAPI, background, and cell segmentation stain Z-stacks. A deconvolution algorithm is then applied and the middle slice of the deconvolved image volume is selected for both background and protein cycles. Image features will appear sharper and fluorescence intensity values lower as a result of image deconvolution algorithm improvements. These are the same steps described for DAPI and cell segmentation images [here](/support/software/xenium-onboard-analysis/latest/algorithms-overview/image-processing#focus-imgs).

</LinkableSection>

<LinkableSection id={"bg-sub"} title={"Background subtraction"} nested={true}>

Next, the 1st and 2nd background cycle images are used to estimate background, which is subtracted from each deconvolved protein image to help offset the effects of autofluorescence. The estimated background image is locally adjusted for uneven brightness patterns by scaling the background in 32 x 32 pixel patches to create a subtracted image with uniform background intensity.

During background subtraction, there may be cases where the background image pixel is brighter than the protein image pixel, resulting in negative values. However, downstream analysis tools may not be able to read signed integer data. To maintain compatibility with other analysis tools and preserve the information that would have been lost in protein images due to clipping negative values to zero, XOA v4.0 instead adds a fixed offset value of 100 units to each pixel. This offset is not applied to cell segmentation images.

An example is shown below. For simplicity, we illustrate pixel values for one protein marker in a 3 x 3 pixel "cell" with an offset of 100 units. The yellow pixel's negative intensity value is retained by applying the offset, instead of clipping it to zero.

<Image src={"https://cdn.10xgenomics.com/image/upload/v1752254635/software-support/Xenium-onboard-analysis/v4.0/BGsubtraction.png"} /> 

For guidance on accounting for this offset value in outputs, read the section [below](/support/software/xenium-onboard-analysis/latest/algorithms-overview/protein-algorithms#offset).

</LinkableSection>

<LinkableSection id={"qc-mask"} title={"Saturation masking and crosstalk correction"} nested={true}>

Protein images then undergo a masking step using saturation QC mask images (output in `morphology_focus_qc_masks/`).

- **Adjust for exposure time differences**: In the image processing pipeline, XOA uses the background images to determine which pixels to mask. Since each protein image is captured with different exposure times compared to the background images, the intensities of the background images are adjusted accordingly.
- **Create QC masks for each protein image**: We identify pixels in the adjusted background images used for subtraction and/or of a different acquisition channel that were saturated. This step is critical because the crosstalk correction algorithm requires valid pixels across all channels for a given cycle. In the binary morphology focus QC mask, pixels have two possible values: saturated pixels = 255 (white) and unsaturated pixels = 0 (black).
- **Use QC masks to mask protein images**: Pixels with saturated autofluorescence are masked in the protein images (set to zero) because they violate the assumptions of the background subtraction and crosstalk correction algorithms. In some cases, when comparing protein and QC saturation masks, you may notice some zero pixel values in protein images are non-zero due to JPEG2000 image compression. Saturated protein image pixels are not masked in order to retain potential positive protein signal.

Finally, a spectral crosstalk correction is applied to remove signal bleedthrough from other channels for protein images. Different spectral channels can interfere with each other. This step corrects that bleedthrough by removing the predicted contribution of each of the fluorophores from channels that are not intended to detect them.

These processed images are provided in the Xenium Onboard Analysis output bundle as multi-file OME-TIFF files, one per channel, in the `morphology_focus/` directory. Outputs are detailed [here](/support/software/xenium-onboard-analysis/latest/analysis/protein-data).

</LinkableSection>

<LinkableSection id={"offset"} title={"Xenium software suite accounts for offset"}>

The Xenium software suite automatically accounts for the [offset applied after background subtraction](/support/software/xenium-onboard-analysis/latest/algorithms-overview/protein-algorithms#bg-sub) in the protein data outputs and when visualizing the protein images.

</LinkableSection>

<LinkableSection id={"mean-intensity"} title={"Protein mean fluorescence intensity"} nested={true}>

The cell-feature matrix files contain protein expression information quantified by a per-cell mean fluorescence intensity statistic. In each cell, every protein marker feature is quantified by calculating the **mean fluorescence intensity** using pixels in the 2D cell segmentation mask.

- This statistic does not include intensity values measured outside cell boundaries and it is not a count value.
- XOA automatically excludes pixels that do not yield reliable protein signal. These invalid pixels stem from: 1) saturation in the scaled background image (recorded in the saturation QC mask), or 2) production of non-existent pixel regions from image distortion correction or stitching at tissue ROI edges (recorded as 0s or 1s in morphology focus image). The presence of invalid pixels primarily affects reanalysis of protein image data. Xenium Ranger also excludes invalid pixels, but if you recalculate protein mean intensity in third-party tools, we recommend applying the same filtering criteria or limiting analysis to regions with valid pixels to avoid artifacts that can occur at ROI edges.

Here is an example for calculating the scaled mean fluorescence intensity for one protein in the [3 x 3 "cell" example illustrated above](/support/software/xenium-onboard-analysis/latest/algorithms-overview/protein-algorithms#bg-sub):

1. **Total cell area**: 3 x 3 = 9 pixels
1. For simplicity, we assume no further changes to the pixel values during crosstalk correction in this example. **Total fluorescence intensity per cell per protein**: 100 + 120 + 100 + 110 + 70 + 100 + 110 + 100 + 100 = 910 units
1. Mean intensity is calculated as the total fluorescence intensity per cell per protein divided by the total cell area in pixels. **Protein mean intensity for this "cell"**: 910 / 9 = 101.11
1. XOA **subtracts the 100 unit offset** prior to saving the values in the cell-feature matrix: 101.11 - 100 = 1.11
1. The mean fluorescence intensity is then scaled by multiplying the average pixel values by a scaling factor of 10 and rounded to an integer value in XOA v4.0. The scaling factor is applied to retain precision, as the matrix format only supports integer values. The scaling factor is stored in the cell-feature matrix [HDF5 file](/support/software/xenium-onboard-analysis/latest/analysis/protein-data#cf-matrix). **Multiply by scaling factor of 10 and round to integer value**: 1.11 x 10 = 11.1 -> round to 11
1. The final scaled mean protein intensity value for this example "cell" is 11.

These values are _rescaled_ as **mean fluorescence intensity** (divided by 10) for the following outputs to use the same scale as the protein image pixel intensities:

- In the analysis summary (per-cell; Protein Analysis tab)
- Xenium Explorer v4.0 (per-cell or per-selection; [Selected ROIs](/support/software/xenium-explorer/latest/tutorials/xe-selecting-multiple-regions-of-interest))

</LinkableSection>

<LinkableSection id={"offset-xe-xr"} title={"Xenium Explorer and Xenium Ranger"} nested={true}>

**Xenium Explorer** v4.0 automatically subtracts the offset from the intensity slider display range when you visualize protein images. This keeps the image values consistent with per cell and per selection quantification stats, which are read from the cell-feature matrix.

<Flex style={{
 flexDirection: "column",
 gap: "18px",
 padding: "1rem 0 2rem 0"
}}>
<WorkflowCard link={"/support/software/xenium-explorer/latest/tutorials/nav-images#image-protein"} title={"Learn about viewing protein images in Xenium Explorer and in community-developed tools"} />
</Flex>

[**Xenium Ranger** v4.0](/support/software/xenium-ranger/latest) follows the same algorithms as XOA v4.0 and subtracts the offset to recalculate mean fluorescence intensities if cell segmentation is changed.

</LinkableSection>

<LinkableSection id={"offset-commtools"} title={"Account for offset in downstream analysis"} nested={true}>

Community-developed tools are _not_ necessarily set up to automatically account for this offset value. For downstream analysis workflows, here are a few scenarios for deciding whether you need to manually account for the offset:

| Analysis | Workflow example | Manually subtract offset? |
| ---------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------- |
| Gating cells just with the cell-feature matrix data | For example, using Python to plot histograms of the _scaled_ mean protein intensities from the matrix. Since the offset is already subtracted from the matrix values, no further changes are needed. | No changes needed |
| Gating cells starting from the protein image data | For example, using QuPath to do cell segmentation, quantification, and gating. The distribution will be shifted by 100, but no need to worry about the offset while working in QuPath. If you import the QuPath cell segmentation to Xenium Ranger to integrate protein and RNA data, then Xenium Ranger will subtract the offset. | No changes needed |
| Picking thresholds using the protein image data and then filtering cell-feature matrix by protein quantity | For example, importing XOA cell segmentation results into QuPath to see histograms and manually set thresholds, and subsequently using those thresholds to filter the cell-feature matrix by _scaled_ mean protein intensities. | Yes, you need to subtract 100 from the thresholds |

</LinkableSection>

<LinkableSection id={"secondary-analysis"} title={"Secondary analysis"}>

For protein secondary PCA, UMAP, clustering, differential gene express analysis, XOA processes the protein-only features with the same filtering and secondary analysis algorithms as for RNA (described [here](/support/software/xenium-onboard-analysis/latest/algorithms-overview/gex-secondary-analysis-algorithms)).

</LinkableSection>

<LinkableSection id={"protein-exp"} title={"Protein expression"} nested={true}>

The pipeline also produces outputs quantifying the protein features that exhibit higher expression levels within a specific cluster compared to the rest of the sample (for each clustering method). The per-cell protein mean fluorescence intensities are standardized using a Z-score normalization process, which enables comparison of proteins with different distributions. For each protein feature, the median Z-score is computed across the cells within each cluster. The calculation is shown below:

For each protein feature and each cluster `i`, we compute the median Z-score of all its cells, based on the average and standard deviation of that protein feature’s expression in the rest of the clusters.

$$
Z(p,i) = \mathrm{median} \left( \frac{P(p,i,k) - \mathrm{median}(p,!i)}{\sigma(p,!i)} \quad \text{for } k=1,\ldots,C(i) \right)
$$

Where

- $p$ identifies a protein feature;
- $k = 1,...,C(i)$ is the index of a cell in the $i$-th cluster;
- $Z(p,i)$ is the average Z-score of the $i$-th cluster for the protein feature $p$;
- $C(i)$ is the number of cells in the $i$-th cluster;
- $P(p,i,k)$ is the expression of protein feature $p$ for the $k$-th cell the $i$-th cluster;
- and $\mathrm{median}(p,!i)$ and $\sigma(p,!i)$ are respectively the median and the standard deviation of the expression of protein feature $p$ across all cells that are not in the $i$-th cluster.

</LinkableSection>

<Flex style={{
 flexDirection: "column",
 gap: "18px",
 padding: "1rem 0 2rem 0"
}}>
<WorkflowCard link={"/support/software/xenium-onboard-analysis/latest/algorithms-overview/xoa-algorithms"} title={"Back to algorithms overview"} />
</Flex>

Xenium Protein Algorithms

An introduction to all algorithms for datasets generated with the Xenium In Situ Gene and Protein Expression assay.

The goal of cell segmentation is to approximate boundaries between cells so that transcripts can be assigned to cells. Downstream, these results will be used to produce a cell-feature matrix, similar to those output by existing single cell and spatial technologies.

These segmentation algorithms are supported for the following assays depending on whether Cell Segmentation Staining is used (algorithms apply for both Xenium v1 and Xenium Prime chemistries):

| Xenium In Situ assay | Nucleus segmentation | Cell segmentation |
| -------------------------------------------- | -------------------- | ----------------- |
| Gene Expression + Cell Segmentation Staining | DAPI-based | Multimodal |
| Gene Expression | DAPI-based | Nuclear expansion |

<LinkableSection id={"seg-nucleus"} title={"Nucleus segmentation"}>

The first step is to detect the locations of nuclei using the DAPI images and a custom neural network for nucleus segmentation. The neural network is trained on thousands of manually labeled image patches covering multiple tissue types. Any nucleus that has 95% or more of its pixel intensity lower than an intensity threshold of 15 photoelectrons (pe) will be removed for deconvolved DAPI images and 105 pe for the DAPI Z-stack.

Prior to XOA v4.0, nucleus segmentation was performed on each Z-slice of the DAPI Z-stack. In XOA v4.0, the nucleus segmentation algorithm prioritizes using the DAPI and 18S rRNA marker (when Cell Segmentation Staining workflow used) [2D global focus map images](/support/software/xenium-onboard-analysis/latest/algorithms-overview/image-processing#focus-processing), and is supplemented by the 3D DAPI Z-stack results. This step enables better detection of nuclei in regions with saturated DAPI signal.

Here is an example of the nucleus and cell boundaries from a mouse retina dataset† before and after analyzing with the XOA v4.0 nucleus segmentation algorithm. Nucleus boundaries are in red and cell boundaries are colored by segmentation method.

| Description | Example in Xenium Explorer |
| ----------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Nuclear morphology image (2D, DAPI) | <Image src={"https://cdn.10xgenomics.com/image/upload/v1756424324/software-support/Xenium-onboard-analysis/v4.0/xoa4-0_nucleus_seg_DAPI.png"} /> |
| Interior RNA morphology image (18S) | <Image src={"https://cdn.10xgenomics.com/image/upload/v1756424328/software-support/Xenium-onboard-analysis/v4.0/xoa4-0_nucleus_seg_18S.png"} /> |
| **Pre-XOA v4.0** nucleus segmentation algorithm | <Image src={"https://cdn.10xgenomics.com/image/upload/v1756424330/software-support/Xenium-onboard-analysis/v4.0/xoa4-0_nucleus_seg_oldalgorithm.png"} /> |
| **XOA v4.0** nucleus segmentation algorithm | <Image src={"https://cdn.10xgenomics.com/image/upload/v1756424329/software-support/Xenium-onboard-analysis/v4.0/xoa4-0_nucleus_seg_newalgorithm.png"} /> |

† Mouse retina tissue dataset provided by Mi-Hyun Nam, Alon Poleg-Polsky, and Samuel Budoff. Retinal tissue preparation detailed in Budoff S and Poleg-Polsky A. [_En face_ Cryosectioning of Mouse Retina for High-dimensional Spatial Molecular Analysis](https://www.jove.com/t/68171/en-face-cryosectioning-mouse-retina-for-high-dimensional-spatial). JoVE, 2025 and Budoff S and Poleg-Polsky A. [A Complete Spatial Map of Mouse Retinal Ganglion Cells Reveals Density and Gene Expression Specializations](https://pmc.ncbi.nlm.nih.gov/articles/PMC11844403). Biorxiv, 2025.

</LinkableSection>

<LinkableSection id={"seg"} title={"Cell segmentation"}>
</LinkableSection>

<LinkableSection id={"seg-multimodal"} title={"Multimodal"} nested={true}>

_This algorithm step is only compatible for tissues prepared with the Multi-Tissue Stain Mix following the Cell Segmentation Staining workflow (see [CG000749](/support/in-situ-gene-expression/documentation/steps/assay/xenium-in-situ-gene-expression-with-morphology-based-cell-segmentation-staining-user-guide), [CG000760](/support/in-situ-gene-expression/documentation/steps/assay/xenium-prime-in-situ-gene-expression-with-optional-cell-segmentation-staining-user-guide), or [CG000819](/support/in-situ-gene-expression/documentation/steps/assay/xenium-in-situ-gene-and-protein-expression-with-cell-segmentation-staining-user-guide))._ Read the [Xenium In Situ Multimodal Cell Segmentation: Workflow and Data Highlights Technical Note](/support/in-situ-gene-expression/documentation/steps/assay/cell-segmentation-tech-note) to learn about assay performance.

The multimodal cell segmentation algorithm uses custom deep learning models to analyze the multi-channel stain images. The models are trained on 10x Genomics Xenium In Situ data, which include a range of [tissue types](/support/in-situ-gene-expression/documentation/steps/experimental-design-and-planning/multimodal-cell-segmentation-validated-tissue-list) and preparations (fresh frozen, FFPE). Starting in XOA v4.0, cell segmentation is performed on the deconvolved, global focus map images (described in [Image Processing Algorithms](/support/software/xenium-onboard-analysis/latest/algorithms-overview/image-processing)).

After nucleus segmentation with DAPI, the algorithm uses three methods to segment cells.

<Image src={"https://cdn.10xgenomics.com/image/upload/v1709342519/software-support/Xenium-onboard-analysis/multimodal_seg_stains.jpg"} />
 

The segmentation results are prioritized in this order for each cell:

- **Segment cells based on their cell boundary stain:** The inferred segmentation from this method should be closest to the true cell membrane boundary. It uses cell-surface marker antibodies to target epithelial markers (ATP1A1, E-Cadherin) and immune markers (pan-lymphocyte: CD45). This method can split nuclei, define cells missing a nucleus, and identify multinucleate cells. Nuclei that overlap with anucleate cells are assigned to the cell.
- **Segment cells based on expansion from the nucleus to the cell interior stain edge:** This method includes both a deep learning model and a nuclear expansion method using the interior stain to infer cell boundaries. It uses the interior stain (18S rRNA marker) and the DAPI stain for nuclei. The XOA cell segmentation algorithm does not currently use the interior protein stain (alphaSMA/Vimentin), as the 18S marker has been sufficient for most tissue types. The inferred cell outline may look irregular in cases where boundaries between cell interior stains are challenging to identify (i.e., for dense tissue types).
- **Nuclear expansion:** For cases where cells that do not have boundary or interior stains, segment cells with a nuclear (DAPI) expansion distance of 5 &micro;m or until another cell boundary is encountered (described more [below](/support/software/xenium-onboard-analysis/latest/algorithms-overview/segmentation#seg-nucleus-expansion)).

The `analysis_summary.html` and `metrics_summary.csv` outputs will provide metrics for the proportion of cells segmented with each method (described [here](/support/software/xenium-onboard-analysis/latest/analysis/xoa-output-analysis-summary#seg-metrics)).

</LinkableSection>

<LinkableSection id={"seg-nucleus-expansion"} title={"Nuclear expansion"} nested={true}>

_This algorithm step is used in DAPI-only workflows or for cells that do not have boundary or interior stains (see algorithm order in section above)._

Once the locations of nuclei in the sample have been identified by the model, a heuristic cell boundary expansion step is performed. The nucleus boundaries are expanded by 5 &micro;m or until they encounter another cell boundary in X-Y. If cell boundaries overlap during expansion, they are resolved using an algorithm that is conceptually similar to Voronoi tessellation.

<SupportCallout level="info">In XOA v1.0 - 1.9, the default expansion distance was 15 µm. It is reduced to 5 µm in v2.0 and later.</SupportCallout>
 

</LinkableSection>

<LinkableSection id={"seg-mask-polygon"} title={"Segmentation mask and polygon outputs"}>

Xenium cell segmentation takes into account the 3D output from the [DAPI image processing step](/support/software/xenium-onboard-analysis/latest/algorithms-overview/image-processing#dapi) for all Z-slices for better accuracy, as well as the 2D focus image outputs if the multimodal cell segmentation workflow is used. The pipeline ultimately produces a flattened 2D segmentation mask for ease of use.

The nuclear boundaries are consolidated to form non-overlapping 2D objects when projected in X-Y. Since the segmentation mask is 2D, transcripts are assigned to 2D shapes based on their X and Y coordinates.

Nucleus and cell boundary polygons are also provided in the outputs. They are approximations of the segmentation masks and are used for efficient visualization of nucleus and cell segmentation in Xenium Explorer and other analysis software. Cell and nucleus mask and polygon data are stored in the `cells.zarr.zip` output file.

You can also use XOA outputs to explore other segmentation methods developed in the community. Analysis Guides on this topic can be found on the [10x website](/analysis-guides). These third-party segmentation results can be viewed in Xenium Explorer with updated XOA metrics and outputs using [Xenium Ranger's import-segmentation pipeline](/support/software/xenium-ranger/latest).

</LinkableSection>

<Flex style={{
 flexDirection: "column",
 gap: "18px",
 padding: "1rem 0 2rem 0"
}}>
<WorkflowCard link={"/support/software/xenium-onboard-analysis/latest/algorithms-overview/xoa-algorithms"} title={"Back to algorithms overview"} />
</Flex>

Segmentation

Segmentation algorithms and mask or polygon outputs for nuclei and cells.

Nucleus and Cell Segmentation Algorithms

This page describes the algorithms underlying the processing and analysis steps that occur on the Xenium Analyzer.

<LinkableSection id={"overview-workflow"} title={"Workflow overview"}>

The Xenium In Situ Gene Expression ([CG000582](/support/in-situ-gene-expression/documentation/steps/assay/xenium-in-situ-gene-expression-probe-hybridization-ligation-amplification), [CG000749](/support/in-situ-gene-expression/documentation/steps/assay/xenium-in-situ-gene-expression-with-morphology-based-cell-segmentation-staining-user-guide)) and Xenium Prime In Situ Gene Expression ([CG000760](/support/in-situ-gene-expression/documentation/steps/assay/xenium-prime-in-situ-gene-expression-with-optional-cell-segmentation-staining-user-guide)) workflows begin with sample preparation. Fresh frozen (FF) or formalin-fixed paraffin-embedded (FFPE) samples are mounted on Xenium slides. The samples are fixed and permeabilized (FF samples) or deparaffinized and decrosslinked (FFPE samples). Then, probe hybridization, ligation, and rolling circle amplification (RCA) are performed following the steps in the respective assay user guides.

Once the sample has been prepared, imaging is performed in cycles on the Xenium Analyzer. During each cycle, fluorescently labeled probes for detecting RNA target sequences and other reagents are automatically cycled in, imaged, and removed. The internal image sensor captures data across multiple Z-planes with a 0.75 &micro;m step size across the entire tissue thickness for every field of view (FOV) in the user-selected region (see Region Selection Guidelines in the [Xenium Analyzer instrument user guide](/support/in-situ-gene-expression/documentation/instruments/xenium-analyzer/xenium-analyzer-user-guide)). Image data are captured for multiple fluorescence channels in every cycle, which are processed and stitched to build a spatial map of the transcripts across the selected tissue section.

<Image src={"https://cdn.10xgenomics.com/image/upload/v1717125490/software-support/Xenium-onboard-analysis/InSitu_workflow_combined.png"} />
 

Thus, over the course of a run, the Xenium Analyzer’s internal image sensor collects 3D volumes across: **1) multiple FOVs, 2) multiple fluorescence channels, and 3) multiple cycles of chemistry and imaging.** This produces terabytes of internal sensor data that are processed efficiently and analyzed across all cycle-channels to decode transcripts. Once transcripts are decoded, downstream analysis and visualization of Xenium raw data output &mdash; the spatial map of transcripts &mdash; can proceed.

</LinkableSection>

<LinkableSection id={"overview-pipe"} title={"Onboard analysis pipeline"}>

The Xenium Onboard Analysis (XOA) pipeline supports the analysis of gene and protein expression data.

<SupportCallout level="info">XOA v4.0 introduces support for analysis of data from the Xenium In Situ Gene and Protein Expression with Cell Segmentation Staining assay (CG000819) (referred to as Xenium Protein assay). The analysis includes the same steps shown for the Multimodal cell segmentation illustration, along with additional steps for protein image processing and analysis. XOA protein data algorithms are described [here](/support/software/xenium-onboard-analysis/latest/algorithms-overview/protein-algorithms).</SupportCallout> 

The **image processing** and **segmentation** steps of the XOA pipeline depend on the assay workflow used to prepare samples. The final steps of both pipelines include **RCA product image registration, decoding, deduplication, secondary analysis**, and generation of **Xenium Onboard Analysis output data**. Quality scores (Q-Scores) are estimated using controls for calibration (see [Controls](/support/software/xenium-onboard-analysis/latest/algorithms-overview/xoa-algorithms#qvs)).

The high-level pipeline steps for each workflow are shown in the diagrams below:

- (**Left**) Following the _Xenium In Situ Gene Expression - Probe Hybridization, Ligation & Amplification User Guide_
 ([CG000582](/support/in-situ-gene-expression/documentation/steps/assay/xenium-in-situ-gene-expression-probe-hybridization-ligation-amplification))
- (**Right**) Following the _Xenium In Situ Gene Expression with Cell Segmentation Staining User Guide_ ([CG000749](/support/in-situ-gene-expression/documentation/steps/assay/xenium-in-situ-gene-expression-with-morphology-based-cell-segmentation-staining-user-guide)) or _Xenium Prime In Situ Gene Expression with optional Cell Segmentation Staining User Guide_ ([CG000760](/support/in-situ-gene-expression/documentation/steps/assay/xenium-prime-in-situ-gene-expression-with-optional-cell-segmentation-staining-user-guide))

<Image src={"https://cdn.10xgenomics.com/image/upload/v1752255030/software-support/Xenium-onboard-analysis/v4.0/RNAonly_SoftwareWorkflow.png"} />
 

</LinkableSection>

<LinkableSection id={"image-processing"} title={"Image processing"}>

The image processing steps performed by the pipeline across cycles include acquisition of:

- **DAPI images:** DAPI (Xenium Nuclei Staining Buffer) is a blue fluorescent DNA stain for visualizing nuclear DNA in fresh and fixed cells. In the Xenium workflow, DAPI staining is used to locate nuclei, inform cell segmentation, and produce a 3D tissue morphology image. DAPI images are captured across all FOVs in the first cycle.
- **Multi-tissue stain images:** The Xenium Multi-Tissue Stain Mix contains stains for visualizing the cell membrane and cytoplasmic regions of cells. The boundary and interior RNA stains, in addition to the DAPI nuclei stain, are used to inform cell segmentation for the multimodal cell segmentation assay workflow.
- **Protein images:** Antibody staining is performed for several protein markers in the Xenium Protein assay performs.
- **Background images:** Background "blank" image, where a picture of the tissue is taken with no fluorophores. It may also be referred to as an autofluorescence image.
- **RCA product images:** For each cycle, images are captured in multiple color channels across all FOVs. Punctate fluorescent signals (puncta) are detected and filtered, and image distortion is corrected.

Click the link below to read about the algorithms used to process these images.

<Flex style={{
 flexDirection: "column",
 gap: "18px",
 padding: "1rem 0 2rem 0"
}}>
<WorkflowCard link={"/support/software/xenium-onboard-analysis/latest/algorithms-overview/image-processing"} title={"10x general image processing algorithms"} />
</Flex>

<Flex style={{
 flexDirection: "column",
 gap: "18px",
 padding: "1rem 0 2rem 0"
}}>
<WorkflowCard link={"/support/software/xenium-onboard-analysis/latest/algorithms-overview/protein-algorithms"} title={"10x protein image processing algorithms"} />
</Flex>

</LinkableSection>

<LinkableSection id={"seg"} title={"Nucleus and cell segmentation"}>

Nucleus segmentation occurs between cycles, while cell segmentation occurs after the imaging cycles have completed.

- In the _Xenium In Situ Gene Expression_ assay, DAPI images are used to infer nucleus boundaries using a machine learning algorithm; cell boundaries are then inferred by isotropic nuclear expansion.
- In any Xenium In Situ assay following the _Cell Segmentation Staining_ workflow, multi-tissue stains and multimodal cell segmentation methods are used to determine cell boundaries.

Click the link below to read about the algorithms used to segment nuclei and cells.

<Flex style={{
 flexDirection: "column",
 gap: "18px",
 padding: "1rem 0 2rem 0"
}}>
<WorkflowCard link={"/support/software/xenium-onboard-analysis/latest/algorithms-overview/segmentation"} title={"10x nucleus and cell segmentation algorithms"} />
</Flex>

</LinkableSection>

<LinkableSection id={"registration"} title={"RCA product image registration"}>

For each FOV, puncta across channels and cycles must be aligned to reduce differences in image offset, rotation, and magnification. This is important for accurate transcript decoding. The localized puncta from each channel and cycle are registered so that puncta corresponding to the same original RNA molecule are aligned tightly in 3D. Nonlinear transformations are fitted such that all the puncta are aligned to the reference morphology image.

</LinkableSection>

<LinkableSection id={"decode"} title={"Transcript decoding"}>

In order to proceed from puncta to transcripts, decoding must be performed. The Xenium codebook contains a collection of codewords that are assigned to genes in a given [gene panel](/support/in-situ-gene-expression/documentation/steps/panel-design). The pipeline uses the `gene_panel.json` to specify a given gene name to an indexed codeword. Each codeword is defined based on a pattern of fluorescent signal intensities recorded across channels and cycles (see diagram on [panel concepts and terms page](/support/software/xenium-panel-designer/latest/tutorials/xenium-panel-design-concepts-terms#codewords)). Some codewords are reserved for negative controls.

When selecting the codeword assigned to each gene, the Xenium panel design process attempts to keep the predicted transcript density in each cycle-channel in each known cell type as uniform as possible, using single cell reference cell type data. In particular, high expression genes that are coexpressed in the same cell type will be assigned codewords in different cycle-channels as much as possible (more details on the [Getting Started with Xenium Panels](/support/software/xenium-panel-designer/latest/tutorials/create-single-cell-reference) page).

The fluorescent signals from all channels and cycles are compared to the codebook using a global (across all FOVs) maximum likelihood approach based on probabilistic modeling. The decoding process itself does not make per cycle-channel calls. Instead, it performs a filtering step to find codewords that are plausible matches for the puncta signal trace. It assigns probabilities to each candidate codeword based on all the intensity information at the spatial location and chooses the most probable. This approach considers attributes such as puncta location, their color and cycle of detections, and signal intensities.

The XOA v3.0 algorithm increased the sensitivity of transcript detection in transcript dense regions (see [here](/support/software/xenium-onboard-analysis/latest/algorithms-overview/image-processing#rca)). The v3.0 pipeline also improved the filtering of puncta signal traces to remove lower probability codewords that are detected at the same location as higher probability codewords. This filtering was introduced to maintain the false positive transcript detection rate at an acceptable level. Starting in XOA v3.2, the algorithm is better at retaining signal from two real transcripts in close proximity. False positive puncta signal traces that were removed in XOA v3.0 - 3.1 are now included in outputs as transcripts with a calibrated Q-Score of 0. This does _not_ affect the counts in the `cell_feature_matrix` outputs, since those only include high-quality transcripts. However, it will decrease the "Percent of all gene transcripts that are high quality" metric and the "mean calibrated quality of gene transcripts" values for a handful of genes in the Gene-Specific Transcript Quality plot of the `analysis_summary.html`.

</LinkableSection>

<LinkableSection id={"qvs"} title={"Quality scores and controls"} nested={true}>

**The cell-feature matrix and Xenium Analyzer's secondary analyses only include transcripts with a Q-Score ≥ 20.** Final Q-Scores are reported in the [transcript data output files](/support/software/xenium-onboard-analysis/latest/analysis/xoa-output-understanding-outputs#transcript-file).

A [Phred-style calibrated quality score](https://en.wikipedia.org/wiki/Phred_quality_score) is assigned to each decoded transcript to signify the confidence in the decoded transcript identity. The quality score is derived from the likelihood of the maximum likelihood codeword (i.e., the codeword that best explains the observed data), compared to the likelihood of other sub-optimal codewords. This yields a raw Q-Score. Codewords are then mapped to targets using the gene panel information.

Final Q-Scores are obtained first by putting the full range of raw Q-Scores into bins. Then, the raw Q-Scores in each bin are calibrated by the proportion of "Negative Control Codewords" in the bin. A final Q-Score is assigned to each raw Q-Score bin, to ensure that Q-Scores in each bin are correctly calibrated. Control codewords are built into the process to ensure that the final Q-Scores are accurately calibrated.

- **Negative control codewords** are codewords in the codebook that do not have any probes matching that code. They are chosen to meet the same requirements as regular codewords and can be used to assess the specificity of the decoding algorithm.
- **Negative control probes** are probes that exist in the panels but target non-biological sequences. They can be used to assess the specificity of the assay.
- **Genomic control probes** are probes that are designed to bind to intergenic genomic DNA but not to any transcript sequence present in the tissue. They are present in the Xenium Prime assay.
- **Unassigned codewords** are unused codewords. There is no probe in a particular gene panel that will generate the codeword.
- The **Deprecated codewords** category is assigned to codewords that are not used in the Xenium Onboard Analysis pipeline.

</LinkableSection>

<LinkableSection id={"dedup"} title={"Deduplication"}>

After decoding, the results are combined from all FOVs. Duplicate decoded transcripts in overlapping regions between adjoining FOVs are reconciled into one transcript, based on a nearest neighbor analysis that considers transcript identity and Q-Score. The deduplicated transcripts are assigned global coordinates based on the reference morphology image obtained from the [DAPI image processing step](/support/software/xenium-onboard-analysis/latest/algorithms-overview/image-processing#dapi) (`morphology.ome.tif`).

Decoded and deduplicated transcripts in 3D with Q-Scores ≥ 20 are assigned to directly overlapping segmented cells to produce a cell-feature matrix. This matrix can be analyzed using conventional and novel single cell and spatial analysis approaches, facilitating [integration](/analysis-guides/continuing-your-journey-after-xenium-analyzer) with existing single cell and spatial datasets.

</LinkableSection>

<LinkableSection id={"secondary-analysis"} title={"Secondary analysis"}>

Xenium Onboard Analysis performs secondary analysis on the cell-feature matrix to generate PCA, UMAP, graph-based and K-means clustering, and differential expression analysis results. t-SNE projection is not supported in Xenium onboard analysis. Click the links below to read about the algorithms used to generate secondary analysis results:

<Flex style={{
 flexDirection: "column",
 gap: "18px",
 padding: "1rem 0 2rem 0"
}}>
<WorkflowCard link={"/support/software/xenium-onboard-analysis/latest/algorithms-overview/gex-secondary-analysis-algorithms"} title={"10x gene expression secondary analysis algorithms"} />
</Flex>

<Flex style={{
 flexDirection: "column",
 gap: "18px",
 padding: "1rem 0 2rem 0"
}}>
<WorkflowCard link={"/support/software/xenium-onboard-analysis/latest/algorithms-overview/protein-algorithms"} title={"10x protein expression secondary analysis algorithms"} />
</Flex>

</LinkableSection>

Xenium Algorithms

An introduction to all algorithms for the Xenium In Situ Gene Expression and Xenium Prime In Situ Gene Expression products.

Overview of Xenium Algorithms

Overview

Getting Started

Analysis

Algorithms

Example Datasets

Release Notes

Advanced

Xenium Analyzer

Xenium Explorer

Xenium Ranger

Xenium Panel Designer

Analysis Guides

Xenium Onboard Analysis

Estimate and Monitor Xenium Analyzer Instrument Run Time

Estimate and Monitor Run Time

A table of Xenium onboard analysis pipeline output files.

Outputs at a Glance

Overview of all Xenium onboard analysis pipeline output files.

Understanding Xenium Outputs

Overview of the interactive Xenium analysis summary output file metrics and plots.

Analysis Summary Overview

Troubleshooting guide for the Xenium analysis summary alerts.

Analysis Summary Alerts

Descriptions for protein data-specific outputs.

Protein Data Outputs

Storing Xenium raw data for reanalysis or grant requirements.

Archiving Xenium Data

The Xenium In Situ software suite is a set of software applications for analyzing and visualizing in situ gene expression data produced by the Xenium Analyzer. The Xenium Onboard Analysis pipeline enables simultaneous on-instrument data collection and analysis.

What is Xenium Onboard Analysis?

Learn about Xenium Onboard Analysis outputs, algorithms, and software compatibility.

Getting Started with Xenium Onboard Analysis

New to Xenium Onboard Analysis?

View the latest updates to Xenium Onboard Analysis algorithms.

View the latest updates to Xenium Onboard Analysis.

Powerful in situ profiling platform using curated panels, powerful visualization software, and easy-to-follow workflow.

Visualize data from Xenium In Situ solutions.

Reanalyze data from Xenium In Situ solutions.

Browse Q&As from the 10x Genomics support team.

A collection of introductions, tutorials, and blogs for data analysis beyond 10x Genomics software.