Part 4

Transcription

Part 4
© 2011 J. Simonds
Tools and
approaches for
mass cytometry
data analysis
Erin Simonds, PhD
W. A. Weiss Lab, UCSF
May 23, 2013
Biaxial plots are not a scalable solution
Parameters: 4
32
14
8
Plots: 6
496
91
28
0
3
19
0
08
19
2
How can we
explore in 30+
dimensions?
The cytometrist’s toolbox
Transform
raw data
Histogram
Biaxial plot
3D plot
Radar
PCA
Gemstone
SPADE
viSNE
Wanderlust
Box plot
Heatmap
FLAME
SamSPECTRAL
Reduce
dimensionality
Summarize
statistics
Identify
clusters
FlowClust
FlowMerge
FLOCK
Vetting the CyTOF: In a fair fight, nearly identical data
Experimental design
7 Surface Markers (CD 3, 4, 8, 20, 33, 45RA, 56)
2 Functional Markers (pSTAT3 & 5)
Bendall, Simonds, et al. Science, May 2011
The cytometrist’s toolbox
Transform
raw data
Histogram
Biaxial plot
3D plot
Radar
PCA
Gemstone
SPADE
viSNE
Wanderlust
Box plot
Heatmap
FLAME
SamSPECTRAL
Reduce
dimensionality
Summarize
statistics
Identify
clusters
FlowClust
FlowMerge
FLOCK
Marker 1
Marker 1
SPADE
SPADE: Spanning-tree Progression
Analysis of Density-normalized Events
Marker 2
Marker 2
Qiu et al., Nature Biotechnology, Oct. 2011
SPADE projects bone marrow as a continuum of phenotypes
CD123
CD11b
CD45
CD34
CD33
CD20
CD19
CD8
CD4
CD3
0
Intensity (%max)
100
Bendall, Simonds, et al. Science, May 2011
Rediscovery of canonical signaling pathways
Basal
IL-7
pSTAT5
0 Intensity (%max) 100
Bendall, Simonds, et al. Science, May 2011
Dasatinib differentially affects immune cell subsets
IL7
(pSTAT5)
PVO4
(pSTAT5)
+ Dasatinib
Abl / Src-family
kinase inhibitor
Bendall, Simonds, et al. Science, May 2011
The cytometrist’s toolbox
Transform
raw data
Histogram
Biaxial plot
3D plot
Radar
PCA
Gemstone
SPADE
viSNE
Wanderlust
Box plot
Heatmap
FLAME
SamSPECTRAL
Reduce
dimensionality
Summarize
statistics
Identify
clusters
FlowClust
FlowMerge
FLOCK
viSNE preserves high-dimensional, non-linear relationships
viSNE map (in 2D)
viSNE-2
Raw data (in 3D)
z
y
x
viSNE-1
Amir et al., Nature Biotechnol. ePub 19 May 2013 viSNE maps healthy marrow as discrete populations
CD11b-hi
monocyte
CD4+ T
CD11b-lo
monocyte
CD20+ B
CD20- B
NK
CD8+ T
Amir et al., Nature Biotechnol. ePub 19 May 2013 viSNE can help detect rare subpopulations
Barcode A: Healthy bone marrow cells
Barcode B: ALL cells spiked into sample at 0.25% frequency (1/ 400 cells)
ALL cells
Amir et al., Nature Biotechnol. ePub 19 May 2013 Same data, different dimensionality reduction algorithms
Amir et al., Nature Biotechnol. ePub 19 May 2013 viSNE is still stochastic, but much more reproducible
SPADE’s stochasticity causes difference between runs
So why should I ever use SPADE?
à Clustering allows comparison between samples
Amir et al., Nature Biotechnol. ePub 19 May 2013 The cytometrist’s toolbox
Transform
raw data
Histogram
Biaxial plot
3D plot
Radar
PCA
Gemstone
SPADE
viSNE
Wanderlust
Box plot
Heatmap
FLAME
SamSPECTRAL
Reduce
dimensionality
Summarize
statistics
Identify
clusters
FlowClust
FlowMerge
FLOCK
Goal: Characterize the GCSF-responsive subpopulation
G-CSF
pSTAT3
pSTAT5
pSTAT5
Who are these
cells?
Are they the
same in every
patient?
pSTAT3
Surface marker profiling of G-CSF responders in AML
Differential gene expression
pSTAT5
Nonresponders
CD14 CD32 CD36 CD86 CD93 CD11a CD1d HLA-­‐DR pSTAT3
G-CSF
pSTAT5
pSTAT5
Basal
Responders
CD46 CD69 CD117 CD133 CD155 CD321 + markers from literature
LSCs
CD13 CD15 CD33 CD34 CD38 CD44 CD45 CD47 CD64 CD114 CD123 CD11b CXCR4 TIM3 Screen pediatric AML samples
(18 diagnosis, 3 relapse, 7 healthy)
Gate G-CSF responders
Identify enriched markers
CD46
pSTAT3
pSTAT3
with Faye Hsu, Yury Goltsev
pSTAT5
Leukemic GCSF responders have distinct surface profiles
Universal
markers
pSTAT3
Healthy
Growth
factor
receptors
CD34
LSC
markers
with Rob Bruggner & Sean Bendall
The cytometrist’s toolbox
Transform
raw data
Histogram
Biaxial plot
3D plot
Radar
PCA
Gemstone
SPADE
viSNE
Wanderlust
Box plot
Heatmap
FLAME
SamSPECTRAL
Reduce
dimensionality
Summarize
statistics
Identify
clusters
FlowClust
FlowMerge
FLOCK
Human B cell development is less understood than in the mouse
Surface
Markers
Intracellular
Markers
IgH Genes
CD34
CD38
CD10
IL7R
CD24
CD19
CD179a
Rag
TdT
DJH
VDJH
Adapted – Cobaleda, BioEssays 2009
The predicted trajectory ‘rediscovers’ human B cell development
TDT
CD3
4
CD24
Maturity Trajectory
“Wanderlust” algorithm: El-ad Amir, Dana Pe’er (Columbia)
TdT
CD38
Following the predicted B cell trajectory
3
4
2
5
1
CD34
CD24
•  Fluorescent panel design à Sorting à qPCR
•  VDJ rearrangement confirms the trajectory:
(unrearranged) 1 à 2 à 3 à 4 à 5 (rearranged)
Sean Bendall, Kara Davis
New insights and resolution in human B cell development
I Surface Markers Cytoplasmic Markers IgH Genes CD34 CD38 CD10 IL7R IIa IIb III IV Va Vb ✓ CD24 ✓ CD19 Ki67 CD179b CD179a Rag TdT pSTAT5 DJH VDJH Sean Bendall, Kara Davis
Summary
Transform
raw data
Histogram
Biaxial plot
3D plot
Radar
PCA
Gemstone
SPADE
viSNE
Wanderlust
Box plot
Heatmap
FLAME
SamSPECTRAL
Reduce
dimensionality
Summarize
statistics
Identify
clusters
FlowClust
FlowMerge
FLOCK
Prof. Garry P. Nolan
Sean Bendall
Kara Davis
Harris Fienberg
Astraea Jager
Rob Bruggner
Rachel Finck
Bernd Bodenmiller (à U. Zurich)
Eli Zunder
Greg Behbehani
Wendy Fantl
and the entire Nolan lab
stanford.edu/group/nolan
Columbia
El-Ad Amir
Jacob Levine
Dana Pe’er
St. Jude Children’s
Amanda Gedman
Ina Radtke
James Downing
Mount Sinai
Michael Linderman
Stanford
Peng Qiu (à MD Anderson)
Sylvia Plevritis

Similar documents