Hands-on Bioinformatics Workshop (HBW), Series 1
Spatial Transcriptomics Data Analysis
by Applied Spatial Omics Centre at University of Calgary
1 Introduction
This report includes stats and segment/ROI Figures of the Human Brain
Nanostring GeoMx Digital Spatial (DSP) data.
Nanostring
offers a set of R packages that enables visualization of the
GeoMx DSP data and, visit the SpatialOmicsOverlay
vignettes for more information. Nanostring generously developed
the package for the community and is actively working on it to
implement and bring more functionalities. If you have any issues or suggestions, please leave
them on
the SpatialOmicsOverlay GitHub repo.
2 Prerequisite
2.1 Environment
- Software: R (v4.3.1)
- Pakcages:
- GeomxTools (v3.4.0)
- SpatialOmicsOverlay (v1.0.0)
2.2 GeoMx DSP Data
- Sample, N=1: Out of the five samples, the following one sample was selected because the TIFF file size was the smallest, 5.74 GB
| Run date | Slide (Individual) | Sample | Autopsy date | Sex | Age |
|---|---|---|---|---|---|
| SEP 01, 2021 | hu_brain_004b | 81084A (Hippocampus) | SEP 04, 2020 | M | 75 |
- Ethnicity: Caucasian
- Diagnosis: Non-disease normal tissue
- Sample type: FFPE
- Segment, N=4: Neuron â—¼, Iba1 â—¼, Neuropil â—¼, Full â—¼
- Files: OME.TIFF and Lab Worksheet are under the Images, and LabWorksheet folders, respectivley
WORKING_DIRECTORY
├── Images
│ └── *.ome.tiff
└── LabWorksheet
└── *_LabWorksheet.txt3 GeoMx DSP Data Visualization
3.1 Preparation
- Set working directory to load files
# Do not run
baseDir <- "WORKING_DIRECTORY" # where the Images/LabWorksheet folders are located
setwd(baseDir)library(SpatialOmicsOverlay)
library(GeomxTools)3.2 Compile a SpatilOverlay object
- Load OME.TIFF and Lab Worksheet files and export to a SpatialOverlay (SO) object, e.g., soSet.RDS
tiffFile <- file.path(baseDir, "Images", "hu_brain_004b.ome.tiff")
labWorksheet <- file.path(baseDir, "LabWorksheet", "01SEPT2021_HWTA_20210929T1939_LabWorksheet.txt")
lwAnnot <- readLabWorksheet(lw = labWorksheet, slideName = "hu_brain_004b")
soObj <- readSpatialOverlay(ometiff = tiffFile, annots = labWorksheet,
slideName = gsub(".ome.tiff", "", basename(tiffFile)),
image = FALSE, # will load the image later
saveFile = FALSE, outline = FALSE)
soObj## SpatialOverlay
## Slide Name: hu_brain_004b
## Overlay Data: 21 samples
## Overlay Names: DSP-1009880000092-G-A02 ... DSP-1009880000092-G-B10 ( 21 total )
## Scan Metadata
## Panels: Human NGS Whole Transcriptome Atlas RNA
## Segmentation: Segmented
## Outline: FALSEslideName(soObj)## [1] "hu_brain_004b"sampNames(soObj)## [1] "DSP-1009880000092-G-A02" "DSP-1009880000092-G-A04"
## [3] "DSP-1009880000092-G-A03" "DSP-1009880000092-G-A05"
## [5] "DSP-1009880000092-G-A08" "DSP-1009880000092-G-A06"
## [7] "DSP-1009880000092-G-A07" "DSP-1009880000092-G-A09"
## [9] "DSP-1009880000092-G-A10" "DSP-1009880000092-G-A11"
## [11] "DSP-1009880000092-G-A12" "DSP-1009880000092-G-B01"
## [13] "DSP-1009880000092-G-B02" "DSP-1009880000092-G-B03"
## [15] "DSP-1009880000092-G-B04" "DSP-1009880000092-G-B05"
## [17] "DSP-1009880000092-G-B06" "DSP-1009880000092-G-B07"
## [19] "DSP-1009880000092-G-B09" "DSP-1009880000092-G-B08"
## [21] "DSP-1009880000092-G-B10"meta(overlay(soObj))## ROILabel Sample_ID Height Width X Y Segmentation
## 1 001 DSP-1009880000092-G-A02 1375 1619 13247 5970 Segmented
## 2 001 DSP-1009880000092-G-A04 1375 1619 13247 5970 Segmented
## 3 001 DSP-1009880000092-G-A03 1375 1619 13247 5970 Segmented
## 4 002 DSP-1009880000092-G-A05 1271 1527 12420 7442 Geometric
## 5 003 DSP-1009880000092-G-A08 1271 1527 12958 8746 Segmented
## 6 003 DSP-1009880000092-G-A06 1271 1527 12958 8746 Segmented
## 7 003 DSP-1009880000092-G-A07 1271 1527 12958 8746 Segmented
## 8 004 DSP-1009880000092-G-A09 1968 1642 14776 3874 Geometric
## 9 005 DSP-1009880000092-G-A10 1232 1650 16539 3519 Segmented
## 10 005 DSP-1009880000092-G-A11 1232 1650 16539 3519 Segmented
## 11 005 DSP-1009880000092-G-A12 1232 1650 16539 3519 Segmented
## 12 006 DSP-1009880000092-G-B01 1113 332 18510 10008 Geometric
## 13 007 DSP-1009880000092-G-B02 1240 384 18638 8362 Geometric
## 14 008 DSP-1009880000092-G-B03 808 1069 20887 8330 Geometric
## 15 009 DSP-1009880000092-G-B04 1969 1527 19982 9052 Segmented
## 16 009 DSP-1009880000092-G-B05 1969 1527 19982 9052 Segmented
## 17 010 DSP-1009880000092-G-B06 1969 1527 19263 11231 Segmented
## 18 010 DSP-1009880000092-G-B07 1969 1527 19263 11231 Segmented
## 19 011 DSP-1009880000092-G-B09 1969 1527 20214 13445 Segmented
## 20 011 DSP-1009880000092-G-B08 1969 1527 20214 13445 Segmented
## 21 012 DSP-1009880000092-G-B10 1271 1527 13507 10070 Geometrichead(coords(soObj))## sampleID ycoor xcoor
## 1 DSP-1009880000092-G-A02 7144 13256
## 2 DSP-1009880000092-G-A02 7145 13256
## 3 DSP-1009880000092-G-A02 7146 13256
## 4 DSP-1009880000092-G-A02 7147 13256
## 5 DSP-1009880000092-G-A02 7148 13256
## 6 DSP-1009880000092-G-A02 7149 132563.3 Generating a Figure without image
- ROIs will be plotted without the tissue image
- Colors are by default Sample_ID (AOIs)
- Check the original tissue image from the Nanostring GeoMx DSP instrument
plotSpatialOverlay(overlay = soObj, hiRes = FALSE, legend = FALSE, scaleBar = FALSE)
3.4 Check the level of gene expression
- Visualize one of the DEGs (Hippocampus up-regulated) and color the segments by the expression level
- Assign plotting factor(s) to the SO object
- Segment from Lab Worksheet
- HPCA expression from the 08_GeoMxSet_AnalysisReady.RDS file
geomxObj <- readRDS(file.path(outDir, "08_GeoMxSet_AnalysisReady.RDS"))
soObj <- addPlottingFactor(overlay = soObj, annots = lwAnnot, plottingFactor = "segment")
soObj <- addPlottingFactor(overlay = soObj, annots = geomxObj, plottingFactor = "HPCA")
soObj <- addPlottingFactor(overlay = soObj, annots = 1:length(sampNames(soObj)), plottingFactor = "ROILabel")
soObj## SpatialOverlay
## Slide Name: hu_brain_004b
## Overlay Data: 21 samples
## Overlay Names: DSP-1009880000092-G-A02 ... DSP-1009880000092-G-B10 ( 21 total )
## Scan Metadata
## Panels: Human NGS Whole Transcriptome Atlas RNA
## Segmentation: Segmented
## Plotting Factors:
## varLabels: segment HPCA ROILabel
## Outline: FALSEhead(plotFactors(soObj))## segment HPCA ROILabel
## DSP-1009880000092-G-A02 Neuron 1078 1
## DSP-1009880000092-G-A04 Neuropil 475 2
## DSP-1009880000092-G-A03 Iba1 84 3
## DSP-1009880000092-G-A05 Full ROI 3144 4
## DSP-1009880000092-G-A08 Neuropil 631 5
## DSP-1009880000092-G-A06 Neuron 1414 6plotSpatialOverlay(overlay = soObj, hiRes = FALSE, colorBy = "HPCA", scaleBar = FALSE) +
viridis::scale_color_viridis() +
ggplot2::labs(title = "HPCA Expression in Hippocampus")
3.5 Customizing the Figure
3.5.1 Add a scale bar
- If scaleBar = TRUE, a scale bar is automatically calculated when plotting
- Scale bars can be customized with the following parameters: scaleBarWidth, scaleBarColor, corner, etc.
plotSpatialOverlay(overlay = soObj, hiRes = FALSE, colorBy = "HPCA", scaleBarWidth = 0.3, scaleBarColor = "black", corner = "bottomleft") +
viridis::scale_color_viridis()+
ggplot2::labs(title = "HPCA Expression in Hippocampus")
3.5.2 Add the tissue image
- Images can be added using readSpatialOverlay(image = TRUE) or added afterwards using the addImageOmeTiff function
- Users can specify which resolution of the image to
extrat with the res parameter
- 1 = largest image and the higher values, i.e., 2, 3, …, 6, get smaller
- Using a high resolution image, e.g., res=1, will cause a java memory error
soObj <- addImageOmeTiff(overlay = soObj, ometiff = tiffFile, res = 6)
plotSpatialOverlay(overlay = soObj, hiRes = FALSE, colorBy = "segment", scaleBar = FALSE)
3.6 Image manipulation
3.6.1 Flipping X-axis
- Reflect the image vertically
plotSpatialOverlay(overlay = flipX(soObj), hiRes = FALSE, colorBy = "segment", scaleBar = FALSE)
3.6.2 Flipping Y-axis
- Reflect the image horizontally
plotSpatialOverlay(overlay = flipY(soObj), hiRes = FALSE, colorBy = "segment", scaleBar = FALSE)
3.6.3 Flipping X- and Y-aixs
- Rotate the image 180 degrees
plotSpatialOverlay(overlay = flipY(flipX(soObj)), hiRes = FALSE, colorBy = "segment", scaleBar = FALSE)
3.6.4 Crop tissue
- The cropTissue function automatically detects where the tissue is and removes non-tissue area
- The amount of area added to the cropped area can be defined by the buffer parameter
soObj <- cropTissue(overlay = soObj, buffer = 0.01)
plotSpatialOverlay(overlay = soObj, colorBy = "ROILabel", legend = FALSE, scaleBar = FALSE)
3.6.5 Crop sample
- The cropSamples function automatically crops the image around the ROIs given
- Users can feed in a subset of the samples with the sampleIDs parameter to zoom in
soObjSubset <- cropSamples(overlay = soObj, sampleIDs = lwAnnot$Sample_ID) # all samples in this tutorial
plotSpatialOverlay(overlay = soObjSubset, hiRes = FALSE, colorBy = "HPCA", scaleBar = FALSE) +
viridis::scale_color_viridis()
3.7 Image (re)coloring
- Fluorecence dye colors are determined from the instrument but can be recolored using the changeImageColoring function
- Note that Target = Custom was set from Nanostring and the following information can be found in README_hu_brain.docx
| Channel name | Fluorescent dye | Marker (Target) | Cell/Structure marked |
|---|---|---|---|
| FITC/525 nm | Alexa 488 | GFAP | Astrocytes |
| Cy3/568 nm | SYTO 83 | DNA | Nuclei |
| Texas Red/615 nm | Alexa 594 | Iba1 | Microglia |
| Cy5/666 nm | Alexa 647 | NeuN | Neurons |
fourCh <- add4ChannelImage(overlay = soObj)
fluor(fourCh)## Dye DisplayName Color WaveLength Target ExposureTime MinIntensity
## 1 Alexa 488 FITC Blue 525nm Custom 200.0 µs 42
## 2 Alexa 532 Cy3 Green 568nm DNA 60.0 µs 9
## 3 Alexa 594 Texas Red Yellow 615nm Custom 250.0 µs 33
## 4 Alexa 647 Cy5 Red 666nm Custom 300.0 µs 225
## MaxIntensity ColorCode
## 1 19257 #0000feff
## 2 8719 #00fe00ff
## 3 10392 #fefe00ff
## 4 30000 #fe0000ffgp <- plotSpatialOverlay(overlay = soObj, hiRes = FALSE, colorBy = "segment", scaleBar = FALSE)
legend <- fluorLegend(soObj, nrow = 2, textSize = 4, boxColor = "black", alpha = 0.8)
cowplot::ggdraw() +
cowplot::draw_plot(gp) +
cowplot::draw_plot(legend, scale = 0.2, x = -0.3, y = -0.3)
fourCh <- changeImageColoring(overlay = fourCh, color = "cyan", dye = "Alexa 488")
fourCh <- changeImageColoring(overlay = fourCh, color = "gold", dye = "Alexa 594")
fourCh <- changeImageColoring(overlay = fourCh, color = "magenta", dye = "Alexa 647")
fluor(fourCh)## Dye DisplayName Color WaveLength Target ExposureTime MinIntensity
## 1 Alexa 488 FITC Cyan 525nm Custom 200.0 µs 42
## 2 Alexa 532 Cy3 Green 568nm DNA 60.0 µs 9
## 3 Alexa 594 Texas Red Gold 615nm Custom 250.0 µs 33
## 4 Alexa 647 Cy5 Magenta 666nm Custom 300.0 µs 225
## MaxIntensity ColorCode
## 1 19257 cyan
## 2 8719 #00fe00ff
## 3 10392 gold
## 4 30000 magentagp <- plotSpatialOverlay(overlay = fourCh, hiRes = FALSE, colorBy = "segment", scaleBar = FALSE)
legend <- fluorLegend(fourCh, nrow = 2, textSize = 4, boxColor = "black", alpha = 0.8)
cowplot::ggdraw() +
cowplot::draw_plot(gp) +
cowplot::draw_plot(legend, scale = 0.2, x = -0.3, y = -0.3)