Xenium pre-designed panels are for human or mouse. Xenium pre-designed panels have transcript specific padlock probes that were not designed with the intent to bind to other species. Due to the highly specific nature of Xenium probes, the overall efficiency of the probes in other species is generally suboptimal. However, because of sequence conservation, it is possible that some probes will bind specifically in other species.
We strongly recommend designing species-specific probes. A species-specific panel is expected to perform well assuming the species offers a well annotated reference transcriptome. Panel design still requires single-cell counts data with cell type annotations. For common model organisms such as rat, panel genes and data can use rat gene names and Ensembl IDs. See the ‘Advanced Custom’ Technical Note for details.
This article is for researchers who still wish to leverage pre-designed Xenium panels for their non-human-non-mouse samples. To enable users to consider using pre-designed panels on other species, the possibility of probes in the panels to bind was tested for a variety of non-human primates for human panels, and rat for mouse panels.
These calculations were performed by using BLAST to align the RNA binding domain of every probe to the respective transcriptomes. These transcriptomes were obtained from Ensembl. A probe is predicted to bind in a cross-species fashion if the overall melting temperature of the aligned arms exceeds the platform minimum of 50C for Xenium V1, and 65C for Xenium Prime (see the advanced custom tech note), and if there are no mismatches present at the ligation site. Xenium Prime melting temperatures were calculated by concatenating the two arms due to the shorter overall probe lengths.
The analysis reports give the fraction of probe sets that bind the annotated ortholog for each species. The orthology relationships were obtained from Ensembl BioMart. The numerator of these fractions are calculated on a per-probe basis, meaning that if a single probe in a probeset is predicted to bind, then the probeset is predicted to bind.
Additionally, a column is included that reports the potential off-target genes in the other species. If any probe in any of the probesets for that gene exceeds the binding thresholds above, then the Ensembl Gene ID of that gene is reported.
The best outcome is where a gene has a 1.0 in the “Fraction probesets binding ortholog” column, and a null value in the “Potential off-target gene IDs” column. Genes with a score below 1 may have reduced sensitivity.
If you want to perform cross-species analysis with Xenium, it would be advisable to normalize the resulting counts based on the probeset coverage fractions reported.
Figure 1. Distribution of genes that are predicted to bind to Crab-eating Macaque from the human Brain panel. 79.3% of genes are predicted to have all of their probe sets bind.
| Gene name | Fraction probesets binding ortholog | Potential off-target gene IDs |
| ABCC9 | 1.0 | None |
| ADAMTS12 | 0.75 | ENSMFAG00000046344 |
Table 1. Example contents of the offtarget files. The first column contains the human or mouse gene name, the second column contains the fraction of panel probe sets for that gene predicted to bind to the ortholog, and the third column contains Ensembl gene IDs of any genes in the other species for which one or more probes might bind. The best outcome is where a gene has a 1.0 in the “Fraction probesets binding ortholog” column, and a null value in the “Potential off-target gene IDs” column. Genes with a score below 1 may have reduced sensitivity.