Question: Did my alignment fail or did I input the wrong image?
Answer: Visium images and gene expression datasets do not have embedded identifiers that can be used to determine which files go together. Users will need to rely on record-keeping to determine which files to use. We recommend including the slide serial number, capture area, and tissue block ID in the image file name to help keep track of correct image-gene expression library pairs. Please see this KB article for more information.
What if I did not name my image using your recommendations?
Unfortunately, there is no other definitive way to identify if the correct image was used; however, there are some visual clues that may help with your investigation.
- Check your alignment: First, please examine your image alignment in your web summary to determine if the alignment was successful. Please see this article for more information.
- Verify that cluster group patterns make sense: Cluster group patterns that do not match tissue structure may be indicative of an incorrect image or image alignment.
- Verify that the capture in your image matches your records. The capture area can be discerned from the pattern on the slide. Please confirm that the capture area for your slide matches your records.
- Examine metrics (e.g. Fraction of Reads in Spots Under Tissue): The web summary produced by Space Ranger includes a metric called “Fraction of Reads in Spots Under Tissue”. Lower than expected results for this metric may indicate that the incorrect image was used, an incorrect alignment occurred, or there was a sample prep issue. Successful samples should have a Fraction of Reads in Spots Under Tissue >50%; however, as this metric approaches the 50% threshold, users should verify that the correct image and sample preparation steps were carried out. Here is an example where automatic alignment was carried out and the Fraction of Reads in Spots Under Tissue is 66.5%:
Please note that using an incorrect image may not trigger a “Low Fraction of Reads in Spots Under Tissue” warning; however, low values may be a clue that more investigation is required. This metric alone is not sufficient to determine if the correct image was used.
What can you do if you suspect the wrong image was used and the Fraction of Reads in Spots Under tissue is lower than expected?
- If the size or orientation of your sections is distinct, selecting the entire capture area during manual alignment in Loupe Browser may reveal areas of high UMI counts outside the tissue border. Customers may be able to use this pattern to select the correct image.
- Customers may explore UMI counts outside the tissue-covered spots using programmatic methods. Here are some examples in R and Python.
Below is an example tissue plot with UMI counts where we selected the entire capture area in the Loupe Browser and reran Space Ranger with the same image and gene expression library.
This example shows spots with high UMI counts outside the tissue-covered spots and areas under the tissue that have few or zero UMIs. This is a good indicator that the incorrect image was used. Here is a second Space Ranger run with the correct image:
* All data used in this article were retrieved from the 10x Genomics Spatial Datasets gallery.
- Manual Alignment in Loupe Browser: https://support.10xgenomics.com/spatial-gene-expression/software/visualization/latest/alignment
- The fiducials are misaligned in my web summary — how can I fix this? https://kb.10xgenomics.com/hc/en-us/articles/360059388512-The-fiducials-are-misaligned-in-my-web-summary-how-can-I-fix-this-
- What information is necessary to record while performing the Visium Spatial Gene Expression assay? https://kb.10xgenomics.com/hc/en-us/articles/360060179012-What-information-is-necessary-to-record-while-performing-the-Visium-Spatial-Gene-Expression-assay-