Computational Approaches to Transcriptome Signatures in the Human Brain Agilent Technologies eseminar February 18, 2016 Mike Hawrylycz, Ph.D.
ALLEN Adult Human Atlas Online tools
An Anatomic Transcriptional Atlas of Glioblastoma Cellular tumor Leading edge Microvascular proliferation Pseudopalisading cells around necrosis Our GBM atlas allows for computational assessment of anatomical composition of any bulk GBM samples. CD44 EZH2 Only large data set to profile distinct anatomical structures of GBM using RNA-Seq and ISH, and machine learning annotation providing an invaluable resource. Proliferation and migratory classes of stem cell markers correspond to two GBM subtypes: classical and proneural. This novel finding is a potential simplification from current understanding. Puchalski, Shah, Miller, et al, submitted Science Ivy-Allen Glioblastoma atlas project http://glioblastoma.alleninstitute.org/
Allen Human Brain Atlas Platform Hawrylycz, Lein, et al, Nature, 2012 Lein, Hawrylycz, Nature, 2014
Allen Human Brain Atlas Platform
Microarray Data Generation Agilent 8x60K array, custom-designed by Beckman Coulter Genomics in conjunction with the Allen Institute, was used to generate microarray data. The array design included the existing 4x44K Agilent Whole Human Genome probe set supplemented with an additional 16,000 probes. At least two different probes were available for 93% of genes with EntrezGeneIDs (21,245 genes). Probes were located on different exons as much as possible when multiple probes were available for a gene. Other probes on the microarray were for transcripts with UCSC IDs (1,852 transcripts) and Agilent IDs (1,268 transcripts). An additional set of probes were included to overlap with the 1,000- and 60-gene sets that were characterized by ISH for the 1,000 Gene Survey in Cortex and the Subcortex Study, respectively, both of which are integrated into the Allen Human Brain Atlas Total RNA in the amount of 50 ng per sample was sent to Beckman Coulter Genomics for processing on Agilent 8x60K gene expression arrays. RNA-Sequencing (RNA-Seq) data were generated for a selected set of 240 samples (120 from each brain) representing 29 cortical and subcortical regions matched across two brains (H0351.2001 and H0351.2002), using aliquots of the same total RNA isolates used to generate microarray data.
Spatial genomic landscape of gene classes
Spatial Topography of the Neocortex Microarray and RNA-seq Miller et al., BMC Genomics, 2014
Reproducible Differential Gene Expression: Six Brains Genes DE between structures in at least 5 brains 96 regions Same direction FC > 3 B&H p-value < 0.01
The Genetic Geography of the Brain http://casestudies.brain-map.org/ggb Tim Dolbeare
The Genetic Geography of the Brain: Vignette Tim Dolbeare Anil Jegga, CCHMC
Protocadherin 8 (PCDH8) Gene Expression in Six Brains Consistent expression pattern of an exemplary gene, PCDH8, across 96 brain regions for the six brains (numbered 1-6). CTX: cortex; HP: hippocampus; AMG: amygdala; STR: striatum; HY: hypothalamus; TH: thalamus; CB: cerebellum; P: pons; MB: midbrain; WM: white matter. Structures shown are a subset of those in (A) with cortex reduced to its major lobes (FL: frontal lobe, OL: occipital lobe, TL: temporal lobe, PL: parietal lobe).
Differential Stability and the Brain C DS as average Kendall Tau DS versus mean expression level and variability DS versus average Pearson based metric
Differential Stability and the Brain Parvalbumin (PVALB), DS = 0.806 The most stable gene in human brain in this dataset. CKS2 (CDC28 Protein Kinase Regulatory Subunit) DS = 0.245 Consistent expression pattern of an exemplary gene, across 96 brain regions for the six brains (numbered 1-6). CTX: cortex; HP: hippocampus; AMG: amygdala; STR: striatum; HY: hypothalamus; TH: thalamus; CB: cerebellum; P: pons; MB: midbrain; WM: white matter. Structures shown are a subset of those in (A) with cortex reduced to its major lobes (FL: frontal lobe, OL: occipital lobe, TL: temporal lobe, PL: parietal lobe).
High DS Genes and the Brain Top 5 th Percentile, n=864 genes
Consistent patterning of Potassium channels Most stable gene class, p<3.19e-12 Span a broad distribution of anatomic patterns and structural markers. Can we classify or describe these patterns?
Consensus Co-expression Patterns in Adult Brain
Anatomic Architecture of 29 Modules
Potassium Channels span Module types
Consensus Co-expression Patterns in Adult Brain
Modules span expression patterns of the brain Maximum module correlation and DS 5 brains predicts module in 6 th brain ρ>0.4, 85.6% (14,856/17,349)
Consensus Co-expression Patterns in Adult Brain Genetic markers for eight cortical cell types in postnatal mouse were identified based on differential expression of RNA-seq derived transcriptomes Zhang Y, Chen K, Sloan S a, et al. An RNA-Sequencing Transcriptome and Splicing Database of Glia, Neurons, and Vascular Cells of the Cerebral Cortex. J Neurosci. 2014;34(36):11929 47.
Canonical Genetic Signatures of the Adult Human Brain: (M1-M16)
Module Annotation: Anatomy, ontology, drug, disease
Canonical Genetic Signatures of the Adult Human Brain: (M1-M16)
Module Annotation: Anatomy, ontology, drug, disease
Module interaction: ontology, pathways, drug targets, cancer gene sets
Unique anatomical patterning of 48 high DS genes. GNB4 PRSS23 TES 265 high DS genes uncorrelated with major patterns
Module preservation from human to mouse. (A) Mouse-human module preservation index measuring average within-module gene correlation in an anatomyindependent fashion, showing highest preservation of the most neuronal modules (M1, M2, M4). (B) Conservation of anatomical patterning, defined as the proportion of mouse genes correlated at > 0.4 to the corresponding human module eigengene (green bars). A subset of genes in each module are both poorly correlated to the human eigengene (gray bars), but instead very highly correlated to a different human module eigengene (> 0.8). (C-H) Correspondence of module eigengene anatomical patterning between human and mouse. Histogram representation of ME pattern in human (blue) and mouse (red), with overlap in green, demonstrating highly conserved patterns for M4, M10, M12 and M19.
Human Connectome Project www.humanconnectome.org Van Essen et al., 2013; Smith et al., 2013
Cortex DS and Functional Connectivity Vilas Menon
Taxonomies of Neocortical Cell Types Large-scale quantitative phenotyping of single neurons in adult neocortex Transcriptomic Physiological Anatomical/Morphology Data-driven taxonomy of neocortical neuron types
Cell Types and Hierarchical Organization GABA Glutamatergic Glia
Genetics of Cell Type based on Cre Lines Tasic, Menon et al., Nat. Neuroscience, 2016
Acknowledgments External Collaborators Anil G. Jegga, Bruce J. Aronow, Kenneth A. Berman, Cincinnati Children s Hospital Medical Center Matthew F. Glasser, Donna L. Dierker, David C. Van Essen, Washington University Pascal Grange, Xi an Jiaotong- Liverpool University Albert-László Barabási, Jörge Menche, Northeastern University, Central European University We wish to thank the Allen Institute founders, Paul G. Allen and Jody Allen, for their vision, encouragement, and support. Jay Schulkin, Georgetown University David R. Haynor, Lance Stewart, University of Washington