Integrated single cell and unsupervised spatial transcriptomic analysis defines molecular anatomy of the
human dorsolateral prefrontal cortex
Louise A. Huuki-Myers1, Abby B. Spangler1, Nicholas J. Eagles1, Kelsey D. Montgomery1, Sang Ho Kwon1,2, Heena R. Divecha1, Madhavi Tippani1, Thomas M.
Hyde1, Stephanie C. Hicks3, Stephanie C. Page1, Keri Martinowich1, Leonardo Collado-Torres1,4, Kristen R. Maynard1,5
1. Lieber Institute for Brain Development, 2. Department of Neuroscience Johns Hopkins School of Medicine, 3. Department of Biostatistics Johns Hopkins Bloomberg School of Public Health, 4.
Department of Psychiatry and Behavioral Sciences JHSOM, 5. Center for Computational Biology Johns Hopkins University
Abstract
Conclusion
Acknowledgements
Identification of Neuronal Subpopulation In Layers
Presenter & Poster requests:
[email protected]
@lahuuki
Download
this Poster:
Methods
Access this Dataset
DE Genes genes in Spatial Domains
Maynard et al., Nat Neurosci, 2021, 10.1038/s41593-020-00787-0
Zhao et al., Nat BioTech, 2021, 10.1038/s41587-021-00935-2
Emani et al, 2022, (syn30106435) 10.7303/syn4921369
Assay n Number of
Samples
Number of
Donors
10x Visium 113,927 spots 30 10
snRNA-seq 77,604 nuclei
(2661,5911)
19 10
snRNA-seq
Identification of Data-Driven Spatial Domains
BayesSpace: spatially-aware unsupervised clustering Spk
Ds
• K = 2: white matter vs. grey matter
• K = 9: classic histological layers
• K = 16: laminar with 2+ domains per histological layer
Spatial Registration of PsychENCODE Datasets
To facilitate ongoing efforts to spatially map and annotate cell
types in the human cortex, we generated a datahub for the
scientific community to explore these integrated spatial and
single cell datasets. This contribution represents a landmark
reference resource for registering cortical cell types identified
from snRNA-seq to spatial domains, and assessing enrichment
of clinical gene sets from large-scale association studies, snRNA-
seq studies and bulk RNA-seq studies within a spatial context.
The molecular organization of the human neocortex has been
extensively studied in the context of its classic histological layers.
However, emerging spatial transcriptomic technologies, such as the
10x Genomics Visium spatial gene expression platform, enables
unbiased identification of transcriptionally-defined spatial domains
that reflect features of existing brain architecture.
Using spatially-aware unsupervised clustering across all samples,
we generated a comprehensive, data-driven molecular
neuroanatomical atlas. We defined unique gene expression
signatures for unsupervised spatial domains at different
resolutions: one broad resolution (k = 9), which most closely
aligned with classic histological layers, and finer unsupervised
resolutions (k = 16, 28), which identified novel spatial patterns. We
identified 29 fine resolution cell type clusters that we reduced to
12 layer-level associated cell types. We added cellular resolution to
our molecular atlas by performing cell type spot deconvolution
using the snRNA-seq data as reference. Using publicly available
neuropsychiatric disease-associated data, we identified gene sets
enriched across the spatial domains.
Single Nucleus RNA-seq
Sp9
D1
is enriched for CLDN5: vascular domain (Endothelia cells)
Higher Resolution for Layer 1
QR
ToDo
• Spatially registered nuclei on to classic layers reduces
dataset to 12 layer associated cell types
• Spatial registration on finer resolution novel domains may
reveal additional information about cell type clusters
• Spatially registered multiple DLPFC snRNA-seq datasets
shows spatial specificity of some cell types
Louise Huuki-Myers Kelsey D. Montgomery Sang Ho Kwon
Stephanie C. Page
Stephanie C. Hicks Kristen R. Maynard
Leonardo Collado-Torres
Abby Spangler Nicholas J. Eagles
Keri Martinowich
Heena R. Divecha Madhavi Tippani
Thomas M. Hyde
k09 libd.shinyapps.io/k9_spatial
DLPFC_Spangler2022
k16 libd.shinyapps.io/k16_spatial
DLPFC_Spangler2022
Spatial Data Shiny App snRNA-seq on iSEE App
Spot Deconvolution Benchmarking
Check out poster 73
by Nick Eagles
Spatial Registration of Clinical Data
Pilot Data
• Improved reproducibility
with k=9 domains &
expanded dataset
k = number of domains
s = domain number
Identified 29 fine resolution clusters across 70k nuclei
Sp16
D14
is enriched SPARC Sp16
D2
is enriched HTRA1
libd.shinyapps.io/DLPFC_snRNA-
seq_2022