Controls, QC, and Troubleshooting in Spatial Transcriptomics Assays

Spatial transcriptomics quality control (QC) determines whether a costly assay produces trustworthy biology or an expensive pile of noise. Because both sequencing- and imaging-based platforms are sensitive to probe specificity, tissue handling, and optical background, a deliberate control and troubleshooting strategy has to be built into the experiment from the start, not added after a run fails.

Why controls matter in spatial transcriptomics QC

A rigorous QC framework exists to detect technical failures in a spatial transcriptomics assay before they are mistaken for real biological findings (Figure 1). Controls matter more here than in standard RNA sequencing (RNA-seq) because spatial experiments carry a per-sample cost that most sequencing workflows do not, making a failed run more expensive than a failed bench experiment.

Figure 1: A linear schematic of the four sequential QC checkpoints in a spatial transcriptomics workflow, from probe design through batch-aware data analysis. Credit: AI-generated image created using Google Gemini (2026).

Every spatial platform generates a background signal from sources unrelated to true transcript detection, including nonspecific probe binding, optical noise, and stray fluorescence. Without a dedicated control strategy, background noise can be indistinguishable from genuine biological signals, and downstream cell typing or spatial statistics inherit that ambiguity.

Spatially aware QC methods have shown that outlier detection approaches borrowed directly from single-cell RNA sequencing (scRNA-seq) can be confounded by spatial biology itself, thereby missing artifacts unique to tissue sections. One spatially aware QC framework identified consistent, technology-specific patterns of low-quality spots and regional artifacts that standard scRNA-seq QC metrics failed to catch, underscoring that spatial QC needs its own toolkit rather than a repurposed one.

Designing probes for spatial assay controls

Negative control probes and blank barcodes give a direct, quantitative estimate of background noise on any given spatial transcriptomics run. Negative controls are real probes designed against sequences absent from the target genome, while blank barcodes are algorithmically valid barcodes deliberately left unassigned to any gene in the panel.

Platform designs differ meaningfully in how they implement this control layer. A systematic comparison of imaging platforms in formalin-fixed, paraffin-embedded tumor samples found that the absence of sufficient negative controls limited the ability to properly evaluate transcript count quality. Additionally, false discovery rates calculated from available controls varied across platforms and gene panel sizes.

That variability means a false discovery rate calculated on one platform is not automatically comparable to that calculated on another, since the number and type of control probes differ across commercial panels. Reporting which control type generated a given quality metric, negative probe, or blank barcode, is essential for anyone comparing results across platforms or publications.

Positive controls play a complementary role by confirming that the assay chemistry works, typically through housekeeping genes or spike-in sequences with predictable expression. A run that passes positive control checks but shows an unusually high negative control signal points toward a specificity problem rather than a chemistry failure, which changes where troubleshooting effort should be directed (Table 1).

Table 1: A comparison of control types used to assess spatial transcriptomics assay quality.

Control type	Purpose	What a failure indicates
Negative control probe	Estimates nonspecific binding background	High off-target probe binding
Blank barcode	Estimates background from unassigned barcodes	Barcode misassignment or optical noise
Positive control or housekeeping gene	Confirms assay chemistry functioned	Failed hybridization, permeabilization, or amplification

Managing autofluorescence in spatial transcriptomics assays

Autofluorescence from endogenous tissue components, most notably lipofuscin in aged or neural tissue, can overwhelm specific fluorescent signals that a spatial assay is designed to detect. Because lipofuscin fluoresces across the visible spectrum, it interferes with multiple imaging channels rather than a single wavelength.

A systematic evaluation of quenching treatments in fixed tissue found that the two most effective chemical treatments reduced autofluorescence intensity by 89–93% and 90–95%, respectively. Less effective treatments achieved autofluorescence reductions as low as 12%. The range between most and least effective treatments illustrates why a quenching protocol validated on one tissue type should not be assumed to transfer directly to another.

Photobleaching offers a complementary, reagent-free approach to the same problem. In one study, a high-power LED photobleaching device applied before staining eliminated lipofuscin-associated autofluorescence in neural tissue while minimizing tissue damage. This approach avoids introducing additional chemical treatments into an already multistep protocol.

A short evaluation sequence helps identify the right autofluorescence strategy for a given tissue type before committing to a full experiment.

1. Image an unstained, unquenched serial section first to characterize the baseline autofluorescence spectrum and intensity.
2. Test one chemical quenching treatment and one photobleaching approach in parallel on adjacent sections.
3. Compare signal-to-noise ratio for the specific fluorophores planned for the main assay, not just overall background reduction.
4. Confirm that the chosen quenching method does not degrade RNA integrity or interfere with downstream probe hybridization.
5. Lock the validated protocol before processing precious or irreplaceable tissue.

Preventing tissue detachment in spatial assays

Tissue detachment during the repeated wash and incubation steps of a spatial transcriptomics protocol is one of the most common causes of a failed or partial run, and it is largely preventable with careful handling. Detachment risk rises sharply whenever a protocol involves multiple liquid exchanges, since each exchange introduces a risk of shear stress at the tissue-slide interface.

Research introducing an RNA-recovery workflow for degraded and challenging samples noted that deparaffinization followed by staining increases the risk of detachment due to repeated washing steps. The same study found that a short baking step after fixation meaningfully improved tissue adhesion to the slide surface, offering a low-cost intervention that fits into existing protocols without major redesign.

Section thickness, tissue hydration, and gentle reagent handling together determine whether a section survives the full assay. Sections that are too thick or derived from a dehydrated block are more prone to cracking, and cracked tissue detaches more readily during subsequent processing than an intact section.

Practical steps can reduce the risk of detachment without requiring specialized equipment:

• Allow adequate hydration time for the tissue block before sectioning.
• Avoid abrupt temperature changes between wash and incubation steps.
• Add reagents gently to the well edge rather than directly onto the tissue.
• Confirm section thickness falls within the platform's validated range before mounting.

Addressing batch drift in spatial transcriptomics data

Batch effects, meaning systematic technical variation between slides, processing runs, or reagent lots, can distort spatial gene expression patterns in ways that are easy to mistake for real biological differences between samples. Because spatial data carries positional information that standard batch correction tools were not designed to preserve, correcting batch drift in spatial transcriptomics requires specialized methods.

Recently, an open-source batch correction algorithm was developed to operate directly on spatial transcriptomics count data, correcting technology-driven batch effects between paired single-cell and spatial datasets from the same tissue type. The approach improved researchers' ability to visualize and detect consistent gene expression patterns across batches that would otherwise appear artificially divergent.

A separate statistical framework for multi-slide data, designed specifically for spatial transcriptomics, corrects batch effects while identifying which spatial domains are shared across slices and which are slice-specific. This is an important distinction when a study spans tissue from multiple donors or processing days. Treating every slide-to-slide difference as pure batch noise risks erasing genuine biological variation.

Practical steps for managing batch drift include:

• Processing case and control samples together within the same batch whenever the study design allows it.
• Recording batch identity as a formal variable at the point of data collection rather than reconstructing it later.
• Applying a spatially aware correction method rather than a standard single-cell tool whenever multiple slides or processing runs are combined for analysis.

Building spatial transcriptomics QC into every run

Controls, autofluorescence management, detachment prevention, and batch correction are not independent checkboxes. Together, they form a single continuous quality framework that starts before the tissue reaches the slide and continues through final data analysis. Treating spatial transcriptomics QC as a one-time check at the end of a run, rather than a series of checkpoints throughout, is the most common reason preventable failures go undetected until it is too late to recover the sample.

Labs that build negative and positive controls, a validated autofluorescence protocol, careful handling practices, and batch-aware analysis into a standard operating procedure can benefit from interpretable data and actionable insights. That upfront discipline pays for itself, given how expensive a repeated spatial transcriptomics experiment can be.

Effective QC also depends on understanding the platforms generating the data in the first place, since platform choice and workflow design shape which control types are available and which failure modes are most likely. Tissue quality entering the assay matters just as much, and careful tissue preparation prevents many of the QC problems this guide addresses from ever appearing on the slide.

Both considerations sit within the broader context of spatial biology methods and technologies, where platform selection, tissue preparation, and QC together determine whether a spatial experiment delivers interpretable data.

This content includes text that has been created with the assistance of generative AI and has undergone editorial review before publishing. Technology Networks' AI policy can be found here.