Getting Started with Xenium Panel Design

Table of contents

Additional resources

Download and explore 10x Genomics public datasets that use pre-designed and add-on custom panels here. Xenium panel part numbers (PN) are provided here.

Click here to generate a custom panel with the Xenium Panel Designer. Find answers to frequently asked questions here.

Panel overview

Xenium In Situ measures gene expression in tissue sections derived from either formalin fixed & paraffin embedded (FFPE) or fresh frozen (FF) tissue samples placed on Xenium slides. Gene expression is measured with panels of probes that target genes of interest. Three types of panels are available:

10x Genomics pre-designed Xenium panels are designed to cover all major cell types found in a tissue for specific species and tissue types (e.g., human brain).
Researchers may also be interested in selecting markers for specific subpopulations of cells, or genes expressed under specific conditions (e.g., disease). 10x Genomics supports add-on custom panels (with up to 100 custom genes) to supplement the pre-designed panels.
Finally, researchers can also request standalone custom panel designs for mouse or human that are not dependent on pre-designed panels.

If your experimental goals do not align with the above options (e.g., you want to target another species or specific isoforms) please contact your sales representative.

Pre-designed panel use cases

For new Xenium users, 10x Genomics recommends starting a pilot experiment with pre-designed panels (see Pre-designed Xenium Gene Expression Panels). There are several advantages to this approach:

Pre-designed panels have been extensively tested by 10x for sensitivity and specificity on the specific tissue types.
Pre-designed panels are in inventory, so researchers can get started right away. No need to wait to design and manufacture a custom panel.
Pre-designed panels are well-suited for exploratory analyses that identify major cell types in the selected tissue.
Researchers can ensure their lab is comfortable with the entire workflow, and their first datasets can be compared to 10x Genomics public data demonstrations.
It is easier to reproduce experiments across labs or research groups.
If there is any experimental optimization or troubleshooting to be done with the 10x Genomics Support Team, starting with pre-designed panels may help by first addressing any issues unrelated to specific panels.

Add-on custom panel use cases

Add-on custom panel designs offer the flexibility to include more genes to available pre-designed panels (see Xenium Add-on Panel Design Technical Note). The add-on custom configuration is only compatible with pre-designed panels; it is not compatible with a standalone custom base.

Review the curated pre-designed panel gene lists carefully. The lists are designed to comprehensively cell type a sample. Do the genes adequately cover the biology you are exploring? If not, consider adding additional genes to complement it.

To test a biological hypothesis, researchers may want to select markers for certain subpopulations of cells or genes expressed under specific conditions (e.g., diseased vs. healthy). For example, 10x scientists designed an add-on panel to better understand breast cancer biology in the preprint: "High resolution mapping of the breast cancer tumor microenvironment using integrated single cell, spatial and in situ analysis of FFPE tissue". They analyzed data from 313 genes total: 280 genes from common cell types using the pre-designed Xenium Human Breast Panel and 33 additional genes, like invasive myoepithelial marker genes. The add-on genes were selected and curated primarily based on single cell reference data for both healthy and tumorigenic human breast tissues (Pal et al. 2021, Bhat-Nakshatri et al. 2021, Karlsson et al. 2021) (see Supp. Fig. 1).

Standalone custom panel use cases

Standalone custom panels can be designed for human or mouse. Standalone custom panels may be useful when the tissue being analyzed has a very different expression signature compared to the specific tissues covered by pre-designed panels.

We support standalone custom panels up to 480 genes. However, depending on the desired gene list and tissue(s) being analyzed, the final number of genes may be reduced in our design process to ensure high sensitivity across all biology of interest. Thus, we recommend submitting ~10-20 additional genes that could be used in the design process to replace genes in the main set as needed.

For information about advanced customizations, please see the Species Standalone Custom and Advanced Custom Panel Design for Xenium In Situ Technical Note (CG000683) for guidance.

Deciding between panel types

Do you need a standalone custom panel, or can you use an add-on with the pre-designed panel? To help decide, check out the Xenium Panel Selector tool, which allows you to enter a list of genes and compare the overlap with available pre-designed panels:

The first time you use this tool, you will be prompted to enter an email address to set up an account. Check your email for a log in link to complete account setup.
Select either "Human" or "Mouse" from the dropdown menu.
Enter a list of genes. Use HGNC/MGI names or Ensembl IDs, and uppercase for human genes (e.g., "ABCC9") or capital case for mouse genes ("Acta2"). Both gene names and IDs can be entered, and should be separated by a space, comma, or new line.

The tool will show which genes are included in pre-designed panels and which genes could be considered for add-on custom design. For example:

Xenium Panel Designer

The following sections walk you through the design process using the Xenium Panel Designer on the 10x Genomics Cloud (available globally). First-time users of the 10x Cloud will be asked to create a new account. At any point during the design process, you can save ("Finish later") and resume work on the panel design later if needed. We provide examples of common design workflows here.

When you click "Start a New Panel Request", you'll enter sample information (sample preparation, study organism) and select whether to design the panel using standard genes (e.g., genes in the human GRCh38 or mouse mm10 reference transcriptome).

Depending on the genes you want to design for, there are two workflows for panel design and review with the Xenium Panel Designer tool:

If "Yes": you are designing an add-on custom or standalone custom panel

The design tool enables you to submit, modify, and finalize your own requests for a standalone custom gene panel or an add-on custom gene panel that complements a 10x pre-designed panel.

Go to the Request standalone or add-on custom panels section to learn about the self-serve workflow steps.

If "No": you are designing a species standalone or advanced custom panel

The design tool enables you to submit a request for species standalone or advanced custom panel designs. At this time, our 10x Support teams will work with you to review and finalize the panel design; you will not be able to finalize these panel design types within the Xenium Panel Designer on your own.

Go to the Request species standalone or advanced custom panels section to learn about the workflow steps.

Workflow 1: Generate standalone or add-on custom panels with Xenium Panel Designer

To generate a human or mouse standalone custom gene panel or an add-on custom gene panel that complements a 10x pre-designed panel, use the design tool to select panel type, choose single cell references, and enter a gene list.

Details for these steps are available in the links below:

Once the input steps are completed, review recommendations provided by the design tool for your panel design.

The panel design summary will display suggestions for optimizing the design. When you are satisfied with the results, finalize the design.

Workflow 2: Request species standalone or advanced custom panels

To request a species standalone or advanced custom panel design, first contact your sales representative to generate a unique design ID. It is used to track your panel through the design process. If you do not know your sales representative, please contact [email protected]. Submit a PO for an order containing this design ID prior to working with 10x bioinformaticians to generate your design.

Next, submit your panel request through the Xenium Panel Designer by entering your design ID and selecting a panel type. Then, select or upload single cell references and enter a gene list (inputs described below). If your panel is for human or mouse samples, the design tool can perform a gene clean up step to check formatting and verify whether we can design probes for them.

Details for these steps are available in the links below:

Select panel type
Provide reference
Provide genes
Optional: Provide advanced custom sequence information

After you submit the custom panel request in the design tool, our support team will be notified. 10x bioinformaticians will then share and discuss the panel design summary results with you to fine-tune custom designs before they are finalized. It typically requires a few rounds of review and discussion between 10x and the customer to finalize a high quality panel.

When you are satisfied with the results, 10x support will finalize the design. You will be able to go to the Xenium Panel Designer tool entry associated with your request to view the final design and download output files.

Panel design input information

The following inputs are part of the design process for both designing your own custom panel or submitting a panel request workflows.

1. Select panel type

Select either the pre-designed panel you wish to build upon with add-on custom genes or standalone custom panel, as well as the number of custom genes you would like to include in your custom panel.

2. Provide reference: single cell reference data and cell type annotation format

You have three options to specify up to five reference datasets:

Select from a collection of publicly available reference datasets with similar tissue type and condition to the samples you plan to run from our provided list of pre-built references.
Upload your own annotated single cell gene expression data in HDF5, MEX, or CLOUPE format. See Creating Single Cell References for Xenium Custom Panel Design from Seurat or AnnData for additional guidance on reference file format conversion.
Choose a combination of public reference datasets and your own single cell reference datasets.

The design tool uses single cell data to create a model of the expression profiles of the cell types present in your samples. It then uses this model to evaluate the risk of optical crowding, check for highly expressed genes, and assign codewords to the genes you selected in an optimal fashion. Here are important considerations for panel design and sample preparation that can affect detection budget:

Mismatched panels: It is important that panels are designed and used with tissues that are representative of the reference data included in the panel design. More than one reference can be used in custom panel designs to represent the diversity of samples that may be studied using a single panel. Detection budget may be exceeded if a panel designed for one tissue is used on another with relatively higher expression in certain genes or cell types (i.e., healthy vs. tumor tissue). This affect may be prevented by including a tumor reference in panel design.
High utilization panels: Panels that are designed with very high utilization are more likely to exceed the detection budget described in the Xenium Add-on Panel Design Tech Note. In addition, panels that are designed using reference data that are not representative of the samples being analyzed may be designed with higher than calculated utilization. It is important to review recommendations during custom panel design to avoid panels with utilization over the detection budget.
Abnormally large tissue thicknesses: We recommend a section thickness of 5 µm for FFPE and 10 µm for fresh frozen as described in Xenium In Situ for FFPE - Tissue Preparation Guide and Xenium In Situ for Fresh Frozen Tissues - Tissue Preparation Guide. Sections that are thicker than this will be more likely to exceed the detection budget.

The single cell reference must be accompanied by cell type annotations for the barcodes. In the design process, the expression levels are aggregated across each cell type. This information is used to assign codewords that minimize optical crowding, as well as ensure that cell type clusters match the broad, expected categories. See general guidelines for panel gene selection in the Xenium Add-On Panel Design Technical Note.

Single cell data can come from Chromium Single Cell Gene Expression or Single Cell Gene Expression Flex assays. If the single cell data comes from Flex, it is important to note that this product does not include genes that are highly and ubiquitously expressed such as mitochondrial genes, ribosomal genes, and HLA class 1 genes. We strongly discourage the inclusion of those genes on Xenium custom panels as well. They take up a large portion of the available optical budget and increase the risk of optical crowding. However, including a small number of genes not present in Flex data generally poses a low risk to assay performance. During the custom panel design process, those genes will be assigned an averaged expression level for the purposes of the utilization analysis.

The information the design tool needs is a gene list and a measure of expected gene expression in the sample stratified by cell type.

If providing your own reference data, the Xenium Panel Designer will accept one of the following formats:

One or more unnormalized whole transcriptome filtered feature-barcode matrices with cell type annotations for each matrix. The matrix and annotation files should be bundled as a .zip, .tar, or tar.gz file (one matrix + one annotation file per bundle).

The feature-barcode matrix can be in either Cell Ranger HDF5 or Matrix Exchange (MEX) format. The HDF5 matrix is a single file, while the MEX format is a folder containing three files (matrix.mtx.gz, barcodes.tsv.gz, features.tsv.gz).
The cell type annotations file can be in CSV or TSV format. It is a two-column file and headers are required. The first column must be "barcode". For example:


barcode,annotation
ATGCATTGCGTAAGTG-1,fibroblast
TTGCAAAGCCGAAGTG-1,fibroblast
CATCATTGCGTAATTG-1,T cell
...

Important

It is critical that barcode suffixes and prefixes in the annotations file exactly match those for barcodes in the matrix file.

A single uncompressed CLOUPE file generated by Cell Ranger. The Xenium Panel Designer uses the graph-based clustering results for cell annotations, so no additional annotation file is needed for this input format.

If looking for rare cell types, providing matrix files for multiple samples may yield better results. We recommend providing a matrix file per sample; it does not need to be aggregated. If multiple matrices are provided, the cell type information across all of the matrices will be evaluated.

Important

It is very important that this matrix is not normalized or gene-filtered. Normalizing/filtering limits our ability to assess the impacts of optical crowding. If the matrix contains a subset of the total gene count data, the representation per gene will be skewed.

An error message will be shown for these input file issues:

Gene IDs and/or gene symbols do not match between the matrix, gene list, and the 2020-A reference.
Gene names contain spaces/blanks in the gene name or have typos.
Files have missing column headers or headers with unexpected or misspelled names.
Matrix and annotation CSV files do not have exactly the same barcodes. This is often seen when barcodes in the annotation file have an extra sample suffix after aggregation, but the matrix itself does not.

Common input file issues that do not halt panel design but give poor results:

The design tool will not error with normalized counts data, but results will be skewed. The design tool should be used with integer counts data.
The design tool will not error with matrix files that filter many genes, but results will be skewed and consequently generate a suboptimal design.
Matrix files that are missing genes in the gene list.
Poorly matched expression data.
Annotation CSV files where the first two columns are not "barcode,annotation". If hierarchical annotations are present in additional columns, they are ignored.

3. Provide genes

The Xenium Panel Designer will check your gene list and provide feedback if any problems or ambiguous genes are encountered for human or mouse samples. You can modify the list based on the feedback before moving to the next step.

Enter a list of gene names and/or Ensembl IDs. Sort the list from most to least important to help us prioritize genes to retain for the final design.

For add-on custom panels, up to 100 genes can be added to a 10x pre-designed panel. The add-on panel list should not include genes that are already in the selected pre-designed panel.
For standalone custom panels, up to 480 genes are supported.
Recommended for any panel type: submit ~10-20 additional genes in case any in the main set need to be replaced during the design process.

10x Genomics can design probes for the human and mouse genes listed here. These genes cover:

96.8% of the human transcriptome and 99.7% of the human coding transcriptome
85.4% of the mouse transcriptome and 99.8% of the mouse coding transcriptome

The genes not covered by our existing probes based on their respective human or mouse 2020-A reference transcriptomes are listed in the following files (gene name, Ensembl ID, and gene type):

Human
Mouse

It is important that the gene symbols and feature IDs match the features in the human or mouse 2020-A reference GTF file exactly. If provided, the Ensembl IDs should not include "version decimals" (e.g., "ENSG00000010404" is ok, while "ENSG00000010404.3" is not). The Ensembl IDs must be the gene ID (ENSG, ENSMUSG) not the transcript ID.

4. Provide advanced custom sequence information

This step is optional and only applies to species standalone custom or advanced custom panel designs.

10x Genomics supports adding these probes to the panel design, but cannot guarantee that they will successfully detect their targets.

There are a few common target design modes (click the links for details): isoform, SNV, exogenous genes, T cell receptors, and barcode detection. The files associated with these targets can be sent to your 10x bioinformatics representative.

To specify RNA binding sequences for advanced custom probe sequences, create a CSV file formatted as specified in the Species Standalone Custom and Advanced Custom Panel Design for Xenium In Situ Technical Note. This CSV file can be uploaded to the Xenium Panel Designer (see Provide Sequences step below) or you can work with your 10x bioinformatics representative to generate it.

Isoform

Please supply a FASTA file with the sequence of each splice junction you want to target, where the splice junction is centered and there are at least 30 bases of transcribed sequence on both sides in transcription orientation.

Here is an example FASTA file for specifying probes for the human EGFRvIII variant, where exon 1 aberrantly splices to exon 8.
```
>EGFR_exon1_exon8
GCGCTCTGCCCGGCGAGTCGGGCTCTGGAGGAAAAGAAAGGTAATTATGTGGTGACAGATCACGGCTCGTGCGTCCGAGC
>EGFR_exon1_exon2
GCGCTCTGCCCGGCGAGTCGGGCTCTGGAGGAAAAGAAAGTTTGCCAAGGCACGAGTAACAAGCTCACGCAGTTGGGCAC
```
These sequences are 80 bp in total length with the splice junction in the center.

The above FASTA file is aligned to the human GRCh38 reference transcriptome. Each sequence is evenly split over the target splice junctions. Note that the sequences are in transcription orientation and only contain exonic sequence.
Single Nucleotide Variant (SNV)

SNV performance will vary due to the ligation junction preferences of the ligase used in the Xenium In Situ assay workflow. Preferred ligation junctions are described in the Species Standalone Custom and Advanced Custom Panel Design for Xenium In Situ Technical Note. Similar to isoform targets, provide a FASTA file with exonic sequences. Here is an example FASTA file for specifying the KRAS G12A mutation:
```
>NM_004985 KRAS-CDS-sequence
ATGACTGAATATAAACTTGTGGTAGTTGGAGCTGGTGGCGTAGGCAAGAGTGCCTTGACGATACAGCTAA
TTCAGAATCATTTTGTGGACGAATATGATCCAACAATAGAGGATTCCTACAGGAAGCAAGTAGTAATTGA
TGGAGAAACCTGTCTCTTGGATATTCTCGACACAGCAGGTCAAGAGGAGTACAGTGCAATGAGGGACCAG
TACATGAGGACTGGGGAGGGCTTTCTTTGTGTATTTGCCATAAATAATACTAAATCATTTGAAGATATTC
ACCATTATAGAGAACAAATTAAAAGAGTTAAGGACTCTGAAGATGTACCTATGGTCCTAGTAGGAAATAA
ATGTGATTTGCCTTCTAGAACAGTAGACACAAAACAGGCTCAGGACTTAGCAAGAAGTTATGGAATTCCT
TTTATTGAAACATCAGCAAAGACAAGACAGGGTGTTGATGATGCCTTCTATACATTAGTTCGAGAAATTC
GAAAACATAAAGAAAAGATGAGCAAAGATGGTAAAAAGAAGAAAAAGAAGTCAAAGACAAAGTGTGTAAT
TATGTAA
```
Additionally, provide a separate CSV file describing the single nucleotide variants you would like to target. This file should use 0-based coordinates in the position (pos) column, with each alternative base in the alt column (one row per alternative base). In the example below, pos = 34 indicates the 35th base of the sequence above in a 0-based coordinates system, where the SNV is an A instead of a G:
```
sequence,pos,alt
NM_004985,34,A
```
An example of the designed probes aligned to the human GRCh38 reference transcriptome is shown below:

Exogenous genes

Exogenous genes include reporter genes, fluorophores, transgenes, or expressed sequences such as CRISPR guides. All of these genes can be specified by providing a FASTA file of the sequence you would like to target. Here is an example FASTA file for GFP:


>L29345.1 Aequorea victoria green-fluorescent protein (GFP) mRNA, complete cds
TACACACGAATAAAAGATAACAAAGATGAGTAAAGGAGAAGAACTTTTCACTGGAGTTGTCCCAATTCTT
GTTGAATTAGATGGCGATGTTAATGGGCAAAAATTCTCTGTCAGTGGAGAGGGTGAAGGTGATGCAACAT
ACGGAAAACTTACCCTTAAATTTATTTGCACTACTGGGAAGCTACCTGTTCCATGGCCAACACTTGTCAC
TACTTTCTCTTATGGTGTTCAATGCTTTTCAAGATACCCAGATCATATGAAACAGCATGACTTTTTCAAG
AGTGCCATGCCCGAAGGTTATGTACAGGAAAGAACTATATTTTACAAAGATGACGGGAACTACAAGACAC
GTGCTGAAGTCAAGTTTGAAGGTGATACCCTTGTTAATAGAATCGAGTTAAAAGGTATTGATTTTAAAGA
AGATGGAAACATTCTTGGACACAAAATGGAATACAACTATAACTCACATAATGTATACATCATGGCAGAC
AAACCAAAGAATGGAATCAAAGTTAACTTCAAAATTAGACACAACATTAAAGATGGAAGCGTTCAATTAG
CAGACCATTATCAACAAAATACTCCAATTGGCGATGGCCCTGTCCTTTTACCAGACAACCATTACCTGTC
CACACAATCTGCCCTTTCCAAAGATCCCAACGAAAAGAGAGATCACATGATCCTTCTTGAGTTTGTAACA
GCTGCTGGGATTACACATGGCATGGATGAACTATACAAATAAATGTCCAGACTTCCAATTGACACTAAAG
TGTCCGAACAATTACTAAATTCTCAGGGTTCCTGGTTAAATTCAGGCTGAGACTTTATTTATATATTTAT
AGATTCATTAAAATTTTATGAATAATTTATTGATGTTATTAATAGGGGCTATTTTCTTATTAAATAGGCT
ACTGGAGTGTAT

T cell receptors

T cell receptor (TCR) detection in Xenium In Situ can come in one of two general flavors: 1) detection of Complementarity-Determining Region 3 (CDR3) sequences or 2) unbiased profiling of V/J segments. Unbiased V/J profiling is actively being developed by 10x Genomics, but is not currently available.

CDR3 sequence detection relies on using an assembled CDR3 sequence to design probes specific for the target CDR3 (such as one produced by the 10x Immune Profiling solution). By doing this, we make the assumption that there are no other CDR3 present at appreciable levels in the sample that have close sequence similarity.

If you want to target known CDR3 sequences, please provide a FASTA file containing the CDR3. You may wish to also include some flanking sequence from framework regions (FWR3/FWR4) to ensure an optimal probe is picked.

For example, this representative PBMC dataset contains assembled CDR3 sequences that range from 42 bp to 48 bp. Here is an example assembled clonotype:

Type	Sequence
Barcode	TATCAGGCACGCGAAA-1
FWR3	ACTGACCAAGGAGAAGTCCCCAATGGCTACAATGTCTCCAGATCAACCACAGAGGATTTCCCGCTCAGGCTGCTGTCGGCTGCTCCCTCCCAGACATCTGTGTACTTC
CDR3	TGTGCCAGCAGCCGGGACAGGGTAAATCAGCCCCAGCATTTT
FWR4	GGTGATGGGACTCGACTCTCCATCCTAG

Converting this to a FASTA file, the inputs would look like this using 20 bp of each framework region:


>Clonotype1
CCCAGACATCTGTGTACTTCTGTGCCAGCAGCCGGGACAGGGTAAATCAGCCCCAGCATTTTGGTGATGGGACTCGACTCTC

Barcode detection

Xenium In Situ chemistry can detect ssDNA or ssRNA at sufficient copy number. We recommend at least 2-4 copies per cell, but ideally with ≥ 10 copies. The ideal number of copies per cell will depend on the proportion of cells expressing your barcode, the efficiency of the probe, and the importance that every cell contains the barcode you would like to detect.

A common use case for barcode detection is lineage tracing. Some use cases involve multiple barcodes in a combinatorial fashion. This is possible to read out with Xenium In Situ, however the probability that all of the events happen in a given cell decreases as the number of individual events that must be detected in a single cell increases.

Placing multiple copies of the same barcode on your construct will increase sensitivity by boosting the effective copy number. If you want to design probes in this fashion, it is ideal to put 50 bp of space between each barcode copy.

Review recommendations

After the input information is submitted, the Xenium Panel Designer will generate a custom panel for you. The page will update as soon as the design is complete (typically ~5 minutes).

During the process, the design tool:

Screens the customer-provided gene list for genes that cannot have probes designed for them (given the reference data and maximum number of probe sets). It may also screen for low-probe targets before running the design tool. The designer will explicitly check for the following:
Genes whose expression is predicted to be too high to assay using Xenium technology
Genes that we cannot design probes for
Combinations of probes that are predicted to cause issues
Genes that are moderately expressed in many cell types and are unlikely to be beneficial in your panel
Genes that we have previously assayed for and pose issues (in sensitivity, specificity, etc.)
Creates a cell type model based on the reference data (customer-submitted or in-house).
Generates the panel design summary to assess the design. The file includes a panel summary with panel metadata and plots to assess the design. If the tool detects any potential issues, alerts or warnings are also shown.

Interpreting your panel design summary

Review the panel design summary (HTML file) to determine whether to further modify or move forward with the current panel design.

Genes of concern (e.g., high expressors; see this Technical Note for examples) are identified in the panel design summary (both as alerts and in the plots). It is common that there will be such warnings early in the design process and it typically requires a few rounds of review to finalize a high quality panel.

While the panel is being designed, the summary HTML will show results for the current design on the left side, as well as what the results would look like with suggested optimizations on the right. Click on this link to download an example HTML panel summary showing pre- and post-optimization design results.

All plots are interactive, and you can zoom in to see all gene labels on the x-axis. Double-click to the right of the plots to reset to the original plot view. Plot guidance is provided in the "How to interpret this?" dropdown menus.

Warnings

If applicable, the summary will display recommendations (red), warnings (yellow), and/or information about genes or cell types on your panel that exceed recommendations for specific sections of the summary.

One example is shown below:

A similar notice is generated when we are unable to design the requested number of probe sets per gene (shown below). This will not halt your panel design, but it does mean the sensitivity of the corresponding gene will be somewhat reduced.

Xenium panels default to eight probe sets per gene, but we have seen detection with as few as two probe sets. Some genes are not long enough or not unique enough to have eight probe sets. For more information, see the probe sets section of the glossary.

Plots

The Panel Utilization per Cell Type plot shows per cell type utilization (TP10K) and the Panel Utilization per Gene plot shows per gene utilization (TP10K). These are used to guide whether genes should be removed from the design or, similarly, whether the number of probe sets should be modified.

The per gene utilization cutoff is 120 TP10K across cell types and the per cell type utilization is ideally under 400 TP10K, with 600 TP10K as the upper limit. Refining your panel to fit within these recommendations can be somewhat of an art, based on the cell type composition of the sample you intend to study. If individual cell types exceed the absolute threshold of 600 TP10K, it is ultimately up to the researcher to decide if they are willing to accept the risk of significantly degraded sensitivity in those cell types due to optical crowding, or whether they prefer to modify their design instead to reduce the utilization in those cell types (ideally to 400 TP10K or lower).

For additional guidance on adjusting probe set coverage and/or the gene list to reduce optical crowding, see this article.

Early in the design process, you may find your utilization plots show that your panel exceeds these recommendations.

An example is shown below (left-side plots), where several cell types exceed the 400 TP10K recommendation and multiple genes also fall into the red area of the plot (above 120 TP10K). As design iteration occurs, you may choose to reduce the number of probe sets for the highly expressed genes (or remove them) in order to bring these metrics into the recommended ranges (right-side plots):

We make automated recommendations on how to modify your custom panel to fit within the recommended expression limits, either by excluding genes or by reducing the number of probe sets used for them. These are shown below the plots.

The Potential Issues section provides guidance on genes in the current panel that you may wish to reconsider including in the final design.

The Probeset Summary plot provides a quick visual check for the number of probe sets and panel sensitivity for each gene in the panel. The plot helps to confirm the number of probe sets used for designs before any reductions in the design process. Yellow outlines on genes (squares) indicate that we recommend reducing the number of probe sets in order to combat crowding. Purple outlines indicate that the gene is using as many probe sets as we can design for that gene. You can hover over the square to view the current and recommended number. Within each group, genes are sorted alphabetically. The alerts below the plot highlight genes and their current probe set number in parentheses that may be problematic in the design.

The Expression Heatmap plot provides a quick visual check that barcodes correspond to the expected gene expression cluster patterns by cell type. Click and drag the bars above and left of the plots to zoom in.

Z-scores are calculated across all cell types for a given gene. This plot is useful for checking that the clustering pattern for each cell type is distinct between types (row by row). White regions of the plot may indicate genes that are missing data in the matrix file (i.e., filtered data) or removed from the design (in the right-side plot).

Finalize panel design

You may edit the panel based on results in the panel design summary as many times as needed. When you are satisfied with the design, select the relevant option to finalize the design on the Review Design Recommendations page (10x support will do this step for workflow 2).

For workflow 1, a 6-digit design ID will be generated. The design ID is used to order your first panel, as well as reorders of the same panel in the future. The output files associated with your panel will be generated for download.

After the panel has been finalized, it can no longer be modified.

On the finalized panel design screen, you can:

Contact your sales representative with the design ID to request a quote to start the panel ordering process.
Learn what is included in your panel(s) order
Upload the panel JSON file to the Xenium Analyzer instrument
Download panel design output files (see table below)

If you would like to know the status of your order, email 10x customer service at [email protected] with the panel design ID, as well as the sales order and/or PO number.

Panel design output files	Description
Panel design summary HTML	See above for guidance on interpreting the panel design summary.
BED file	A file with custom panel gene targets, which can be used to gauge whether a limited number of probes target CDSs of interest.
JSON file	A file describing the custom panel that is used by the Xenium Analyzer instrument (same format as `gene_panel.json`).
Gene list CSV	The final gene list and the number of probe sets for each gene.

Using the custom panel

For add-on custom and standalone custom panel designs, after shipment has been initiated, your order will include a gene panel file (JSON format). You can download the finalized panel design JSON file from the Xenium Panel Designer.

The JSON file should be uploaded to the Xenium Analyzer instrument by inserting a USB into the Xenium Analysis Computer USB port (exFAT format). The Xenium Analyzer User Guide details the steps to start an instrument run.

Glossary

The factors below may contribute to the overall success of add-on custom and standalone custom panel designs and impact downstream data analysis:

Probe sets: The Xenium chemistry is probe-based. Multiple probes are used per gene in order to successfully capture biological variation (i.e., multiple isoforms). A probe set is a collection of probes designed to detect as many isoforms of a gene as possible based on GENCODE BASIC annotations.

In the example illustrated above, the four isoforms of the gene are targeted by three probe sets. "Probe set 1" and "Probe set 2" each consist of one probe and target four isoforms. "Probe set 3" has two probes because it is not possible to target all four isoforms with a single probe.

The probe designer targets eight probe sets per isoform. It is not always possible to design a single probe that covers every isoform for a gene because some genes have isoforms that do not share sequence, or more likely the sequence they do share is not a good target for probe design. This means that some genes in the panel will have probe sets consisting of multiple probes (a probe set is not necessarily a single oligonucleotide).

Probes should unambiguously bind to the intended target transcript and not other transcripts. The transcript-complementary regions of the probe must bind specifically to the target sequence within the limits of sample preparation conditions and must enable ligation at the padlock junction and efficient rolling-circle amplification. We have pre-designed probe sets for all genes, except for a small fraction of genes for which we were unable to design effective probes:

Genes with high sequence homology to paralogs (e.g., genes with zinc-finger motifs) are not amenable to specific probe design because they lack a unique target sequence.
Factors like extreme GC-content can also limit our ability to design effective probes for a given gene.

Based on the above limitations, some genes may have as few as one probe in a probe set.

Generally, genes with at least three probe sets have robust detection, while those with one or two probe sets run the risk of not being detected. In the panel design process, we may include fewer than eight probe sets for a gene if it is very highly expressed to maximize the optical budget efficiency.

A reduction in probe sets for a given gene results in an approximately linear decrease in sensitivity. For example, the detection efficiency for a given gene decreases by ~50% if probe sets are reduced from eight to four.

Detection budget: Xenium In Situ uses fluorescent microscopy to detect individual transcripts in cells. Like all imaging-based technologies, only a finite number of fluorescent signals can be distinguished within a given area or volume, such as a cell. This means that there is a finite number of transcripts that can be detected in a given tissue volume, which is referred to as a detection budget. As a consequence, panel design involves allocating the available budget in each cell type to a set of carefully selected panel genes.

Optical crowding: Xenium In Situ was carefully designed to be able to compensate for optical crowding and enable robust transcript detection in most cases. However, in severe cases, the detection sensitivity for a subset of genes on a panel can be impacted significantly. Our custom panel design process is designed to help researchers minimize the risk of severe optical crowding. You can learn more about optical crowding and the ways in which Xenium compensates for this effect in the Xenium Add-on Panel Design Technical Note.

Sensitivity: Sensitivity refers to the fraction of transcripts detected per cell, i.e. the ratio between the number of transcripts observed and the number of transcripts actually present in the cell. Since it is usually not possible to know the true number of transcripts present, it is typically difficult to quantify sensitivity in absolute terms.

Technical noise: Like other single cell technologies (e.g., Chromium Single Cell 3’ Gene Expression and Chromium Single Cell Gene Expression Flex), Xenium in situ data analysis does not detect every transcript of a given gene present in a cell. Cell-by-cell variation in the fraction of transcripts detected manifests as technical variation (noise) in the number of observed transcript counts per cell.

Codewords: See the Overview of Xenium Algorithms page for decoding explanation.

Utilization: For a given cell type, utilization describes the fraction of a cell's transcriptome that a Xenium Gene Expression panel is targeting. It is measured in units of "TP10K" (transcripts per 10,000).

Higher utilization values can indicate higher predictions of transcript density for cells belonging to a given cell type.

To calculate utilization per gene (measured in units of TP10K) from single cell data, we first calculate the mean expression of each gene in all cells labeled with a given cell type and then scale the resulting profile so it sums to 10,000. This is done by dividing all values of the profile by the total sum and multiplying by 10,000. To calculate utilization of the panel, we sum the TP10K values for all genes on the panel.

This method is similar to how bulk RNA-seq expression profiles are calculated, sometimes in units of "TPM" (transcripts per million). Here, we scale to 10,000 in order to obtain relative values that are closer to the absolute number of transcripts observed.