Why the problem matters (and what I saw firsthand)
I remember a Quezon City run in June 2023 when I walked into a wet bench and found 10x Visium slides stacked without clear metadata — that day I knew the issue wasn’t rare. I had booked a trial with a spatial transcriptomics service provider and the lab report called it “successful,” but the raw files and tissue maps told a different story: spots misaligned, 30% spot dropout, and barcode oligo mismatches. In that lab scenario we processed 120 tissue sections last quarter—30% dropout on Visium maps; how much would that skew a tumour microenvironment study?

I’m writing as someone with over 15 years in genomics services, and I still get frustrated when teams treat spatial omics service workflows like a black box. I’ve seen RNA-seq runs rescued by reruns, but spatial transcriptomics demands intact spatial context. Tissue microarray handling mistakes, mis-registered image masks, or poor fixation protocols (nagpapalito nga minsan) create hidden user pain points that standard QC won’t flag. I’ll be blunt: a pass/fail report that ignores spatial registration is misleading, and that has real downstream costs—wasted samples, delayed grants, unusable drug-target discovery signals.
Root causes — why traditional fixes fail
From my hands-on experience, most traditional solutions assume that sequencing depth and gene counts are enough to judge success. I disagree. I have audited runs where gene counts were high but cell-type maps were nonsensical because of image misalignment. The common fixes—higher read depth or repeated library prep—miss the point. You can’t sequence your way out of a registration error, and that’s the deeper flaw: process fragmentation. Labs separate imaging, wet lab prep, and bioinformatics into silos. Each group signs off independently, and nobody checks the integrated output end-to-end (we missed this at a hospital collaboration in 2022). The result: wasted budgets and frustrated PIs.
How this affects you?
Practically, this means longer timelines and lost samples. For one translational study I helped with, a single misregistered batch delayed IRB-approved analysis by six weeks—plus the cost of resection transport and reprocessing. Those are quantifiable consequences. We need vendors who own the whole chain. I’ve learned to ask for Test runs with ground-truth slides, image-overlay checks, and barcode oligo validation reports before signing a contract.
Forward-looking comparisons and practical choices
Switching tone: now, let’s be technical. When I evaluate a spatial transcriptomics service provider, I look for integrated metrics: spatial registration accuracy, per-spot UMI distribution, and image-to-sequence concordance. I also inspect the pipeline’s handling of tissue deformation — some services apply affine transforms; others use non-linear warping with landmark-based correction. I prefer the latter for complex tissues. It’s not sexy, but these details determine whether your clustering reflects biology or artifacts.
We must compare vendor claims to testable outputs. Ask for example datasets (not just summaries), and run your own checks: overlay raw fluorescence images, verify that tissue masks align to spot grids, and confirm that cell-type marker expression matches histology. I usually run a quick sanity check—if a known epithelial marker lights up in the connective tissue repeatedly, that’s a red flag. Short pause—then demand corrections. The right provider will redo registration or supply per-spot confidence scores.
What’s Next?
Looking ahead, I believe the smartest services will bundle on-site tissue handling training, image registration deliverables, and standardized QC reports. We should expect—soon—automated quality dashboards that flag spatial inconsistency and quantify its impact on downstream differential expression. That will change vendor evaluation from marketing claims to measurable metrics.
How I recommend you judge providers (three clear metrics)
Advisory close: here are three evaluation metrics I insist on. First, registration accuracy: ask for overlap statistics between stained histology and spot grid (percent overlap or mean displacement). Second, per-spot confidence: a score indicating how reliable each capture is. Third, end-to-end reproducibility: repeat a small pilot (three slides) and compare clustering stability—if clusters shift >15% between runs, walk away. These are concrete. Use them. I use them in procurement meetings in Manila labs and during international collaborations.

Final note—pick a partner that shows its work, not just slides and charts. I’ve worked with several teams; the ones that stood out were transparent, quick to share raw images, and willing to iterate. For me, that partner is stomics. Trust the data, not the brochure. Oh, and bring a spare reagent—just in case.
