Team lcolladotor Journal Club: Spot Sweeper

Team lcolladotor Journal Club: SpotSweeper spatially-aware quality control for spatial transcriptomics
Michael Totty, Stephanie C. Hicks, and Boyi Guo, bioRxiv, 2024
doi: https://doi.org/10.1101/2024.06.06.597765

Presented by: Louise Huuki-Myers, Sep 4, 2024
Recording of our journal club meeting: https://youtu.be/f72XcXmtkfw

SpotSweeper is a software package for quality control in spatial transcriptomics data. Spot quality is evaluated against a local neighborhood of spots, outliers are identified based on a local z-score. This is more effective in spatial data than finding outliers in the global distribution.

Transcript

1. Team lcolladotor journal club: SpotSweeper spatially-aware quality control for spatial transcriptomics Michael Totty, Stephanie C. Hicks, and Boyi Guo doi: https://doi.org/10.1101/2024.06.06.597765 Presented By Louise Huuki Myers (9/4/2024)
2. Quality Control for Spatial Transcriptomics Data Motivation • Poor quality spots should be removed in spatial transcriptomics (SRT) data to improve accuracy of downstream analysis • Methods for quality control in snRNA-seq don’t account for spatial information in SRT Solution: SpotSweeper • Identify poor quality spots considering the spatial neighborhood
3. Quality Control Metrics • library size ◦ Aka: Counts, Sum Unique Molecular Indices (UMIs) • Number of detected genes ◦ Aka: sum genes, total features • percent of reads mapping to mitochondrial genes ◦ Aka: mito rate
4. Ex. Expected low transcriptional activity in White Matter • With isOutlier(sum_umi) spots in the white matter are often flagged as outliers
5. • Spot-level QC ◦ Define a local neighborhood using k-nearest neighbors based on spatial coordinates ◦ Use neighborhood to calculate z-score for the QC metric (library size, detected genes, mito rate) ◦ Identity outlier spots based on z-score • Artifact QC (more on this later) Fig 1: An overview of SpotSweeper
6. “key advantages of SpotSweeper are it (B) is less biased by differences across spatial domains, (C) retains more high-quality spots, (D) accurately detects local outliers, and (E) accurately detects compromised spots due to region-level artifacts.” Fig 1: An overview of SpotSweeper
7. Fig 2: SpotSweeper improves quality control using local versus global approaches to identify outliers or low quality spots. • Example in Human Pilot DLPFC data, leverage manual annotations • QC metrics are layer dependent • Global outliers would drop large % of spots from Layer 1 & WM • Local z-score has similar distributions between layers
8. Fig 3: SpotSweeper detects consistent set of spots with systematically low library size driven by barcode biases. • 6 barcodes in Visium slides have low library sizes across datasets • Synthetic barcodes at these spots may be responsible for downstream quality issues ◦ Certain k-mers show bias towards larger or smaller library sizes
9. Fig 4: Characterizing region-level technical artifacts unique to spatial transcriptomics. • Dry Spots ◦ incomplete coverage of Visium capture areas by permeabilization liquid ◦ Not visible in H&E image ◦ Low library size but same mito ratio vs. other spots • Hangnail ◦ Visible in H&E ◦ Similar library size and mito rate ◦ Clusters with wrong layer
10. Fig 5 SpotSweeper accurately detects dryspot and hangnail tissue artifacts. • Find regions with low variation in mitochondrial rate ◦ Identity neighborhoods ◦ Calc local variance for central spots, adjust using reweighted least squares ◦ PCA on corrected local variance ◦ Cluster with k=2
11. Questions?
12. Use Case: ERC Visium
13. QC example: Br5460 Local Outliers in ERC data ✅ • Dropping 544 spots (0.44% in tissue spots)
14. SpotSweeper Example “DBSCAN clustering on log2-normalized QC metrics of library size and number of unique genes detected to identify dryspot artifacts” “ this procedure in the findDryspot function within the SpotSweeper package” - Missing function in package?
15. Suspected low quality areas Low UMI but not clear WM