Table 1 Summary of key challenges, tips, and examples from current data

Establishing permeabilization time

As a bright signal is desired, scientists may choose exposure times that are too long, too much gain, and/or laser strengths that are too high in an effort to obtain "better" images

Verify that there is no signal in the negative control and adjust parameters accordingly

Choosing tissue section and size

Section has to fit capture area

Trim sample size in the cryostat using a pre-cooled razor blade

Mounting sections onto Visium slide is a final step

Before mounting section onto Visium slide, mount test sections onto standard microscopic slide to verify there is no freezing damage and the muscle fibers are oriented correctly

The same Cq value (as determined by qPCR) are used per slide for cDNA amplification, but optimal Cq values may differ between samples

Mount approximately similar section sizes and sample types per run

If the libraries are going to be pooled for sequencing and these libraries contain information from unequal numbers of capture areas, then sequencing reads are asymmetrically shared, leading to different sequencing depths

Mount approximately similar section sizes and sample types per run

The 2-week sample was too small to be pooled with the other samples (Fig. 1A), leading to unequal sequencing depths (mean reads per spot, Fig. 1B)

Choosing sequencing depth

How deep do these samples have to be sequenced? What is the relationship between false negative rate and sequencing depth?

As a general rule, increasing sequencing depth will increase sequencing saturation which decreases false negative rate. Based on the current data, this has to be addressed separately for each sample depending on tissue size and heterogeneity

Higher mean reads per spot leads to higher absolute sequencing saturation (Fig. 1A + B). Despite this, most genes were detected in the largest sample (Fig. 1E), even though sequencing depth and saturation was lowest in this sample (Fig. 1B + C)

Interpreting false negative rate per cluster or tissue type, respectively

The transcripts per spot are derived from more than one cell (Fig. 2D + E). The more heterogeneous these cells are, the more different transcripts are expected to be; thus, the deeper it must be sequenced

Increase sequencing depth depending on spots and tissue type of interest

More nuclei contribute to capture areas of connective tissue (Fig. 2E). Even though these spots tended to contain fewer RNA molecules compared with the cytosol of myofibers (Fig. 1B), the number of different genes is less affected, suggesting more heterogeneity (Fig. 1B)

Identification of areas of interest and investigating heterogeneity

The heterogeneity of cells contributing to a spot may exceed more subtle tissue-specific transcriptional states in unbiased clustering

Manually perform hypothesis-driven transcriptional marker search

Myofibers undergoing degeneration/regeneration cycles did not cluster separately (Compare Fig. 2A with Fig. 3C). Manual selection and differential gene expression analysis was performed using expression of key transcriptional markers (Fig. 3D + E)