Lecture 10: Multiparameter Data Analysis

advertisement
BMS 631 - LECTURE 10x
Flow Cytometry: Theory
J. Paul Robinson
Professor of Immunopharmacology
Professor of Biomedical Engineering
Purdue University
Multiparameter Data Analysis
3rd Ed. Shapiro p 207-214
Bindley Bioscience Center
Purdue University
Office: 494 0757
Fax 494 0517
email; robinson@flowcyt.cyto.purdue.edu
WEB http://www.cyto.purdue.edu
Page 1
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Data Analysis
• Gating
• Data displays
–
–
–
–
–
–
histogram
dot plot
isometric display
contour plot
chromatic (color) plots
3 D projection
Page 2
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Gating
•Real-time gating vs. software gating
•Establishing regions
•Gating strategies
•Quadrant analysis
•Complex or Boolean gates
•Back gating
Page 3
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Real-Time vs. Software Gating
Real-time or live gating:
-restrict the data that will be accepted by a
computer (some characteristic must be met
before data is stored)
Software or analysis gating:
-excludes certain stored data from a
particular analysis procedure
Page 4
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Establishing Regions
•Establishing regions:
-objective or subjective?
-training/skill/practice
•Possible shapes:
-rectangles
-ellipses
-free-hand
-quadrants
•Statistics
Page 5
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
log PE
Using Gates
Region 1 established
Gated on Region 1
Page 6
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Quadrant Analysis
(+ +)
log PE
( - +)
(- -)
(+ -)
Page 7
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Drawing Regions: Sample Preparation
Sample Quality
Data removed
From analysis
Spores
Debris
Spores
Vegetative
Debris
B.subtilis spores
B.subtilis veg. + spores
Page 8
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Complex or Boolean Gating
With two overlapping regions, several
options are available:
Page 9
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Boolean Gating
Not Region 1:
Page 10
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Boolean Gating
Not Region 2:
Page 11
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Boolean Gating
Region 1 or Region 2:
Page 12
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Boolean Gating
Region 1 and Region 2:
Page 13
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Boolean Gating
Not (Region1 and Region 2):
Page 14
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Light Scatter Gating
1000
600
800
Scatter
Forward
Side Scatter Projection
Forward Scatter Projection
Neutrophils
Forward Scatter Projection
200
400
Monocytes
0
Lymphocytes
0
200
400
600
800
1000
90 Degree Scatter
Human white blood cells
Page 15
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
log PE
Back-Gating
Back gate
Region 4 established
Back-gating using Region 4
Page 16
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
3 Parameter Data Display
Isometric Display
Page 17
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Methods that can change results:
1. Doublet discrimination
2. Time as a quality control parameter
Example: DNA content
-need to eliminate debris & clumps
-need to gate out doublets
-maintain constant flow rate
Page 18
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
DNA Histogram
Gating out bad data
G2-M
S
A
B
C
Counts
# of Events
G0-G1
Fluorescence Intensity
Time
Page 19
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Multi-color studies
generate a lot of data
++
--
+-
++
--
+-
++
--
+-
+-
++
--
+-
++
--
+-
QUADSTATS
Log Fluorescence
QUADSTATS
-+
-+
++
--
+-
++
--
+-
QUADSTATS
Log Fluorescence
QUADSTATS
-+
-+
++
--
+-
-+
++
--
+-
QUADSTATS
Log Fluorescence
QUADSTATS
-+
10
Log Fluorescence
--
QUADSTATS
Log Fluorescence
QUADSTATS
-+
++
Log Fluorescence
+-
-+
Log Fluorescence
--
QUADSTATS
Log Fluorescence
QUADSTATS
-+
++
Log Fluorescence
+-
-+
Log Fluorescence
--
QUADSTATS
Log Fluorescence
QUADSTATS
-+
++
9
QUADSTATS
-+
++
--
+-
-+
++
--
+-
Log Fluorescence
QUADSTATS
Log Fluorescence
+-
+-
Log Fluorescence
-+
8
Log Fluorescence
--
--
QUADSTATS
7
Log Fluorescence
++
++
6
Log Fluorescence
-+
-+
5
Log Fluorescence
+-
Log Fluorescence
--
+-
QUADSTATS
-+
++
--
+-
Log Fluorescence
Log Fluorescence
QUADSTATS
QUADSTATS
QUADSTATS
QUADSTATS
QUADSTATS
QUADSTATS
QUADSTATS
QUADSTATS
QUADSTATS
QUADSTATS
-+
++
--
+-
Log Fluorescence
-+
++
--
+-
Log Fluorescence
-+
++
--
+-
Log Fluorescence
-+
++
--
+-
Log Fluorescence
-+
++
--
+-
Log Fluorescence
-+
++
--
+-
Log Fluorescence
-+
++
--
+-
Log Fluorescence
-+
++
--
+-
Log Fluorescence
-+
++
--
+-
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
Log Fluorescence
++
--
QUADSTATS
Log Fluorescence
QUADSTATS
-+
++
4
Log Fluorescence
+-
Log Fluorescence
-+
Log Fluorescence
--
QUADSTATS
Log Fluorescence
++
Log Fluorescence
Log Fluorescence
QUADSTATS
3
Log Fluorescence
2
Log Fluorescence
1
Log Fluorescence
3 color
-+
5 color
4 color
-+
++
--
+-
Log Fluorescence
Page 20
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Contour plots
Dot plots
Page 21
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Typical phenotypic analysis histograms
•
This figure shows two examples of simultaneous 2 color immunophenotyping. In figure 3 (a)
the directly labeled MABs used were CD4-PE / CD8-FITC. In this example approximately 50%
of the cells were positive for CD4 and 23% positive for CD8. These percentages were
calculated based upon the settings of the negative control for 2% positivity. Right figure shows
a similar situation for CD2-PE / CD19-FITC.
Page 22
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Kinetic Analysis
50 ng PMA
Stimulated
0 ng PMA
Unstimulated
0
450
900
1350
1800
TIME (seconds)
0
450
900
1350
1800
TIME (seconds)
Figure: This figure shows an example of stimulation of
neutrophils by PMA (50 nm/ml). On the left the unstimulated
cells show no increase in DCF fluorescence . On the right, activated
cells increase the green DCF fluorescence at least 10 times the
initial fluorescence.
Page 23
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
80
70
60
50
40
10
20
30
SSC-Height -->
90
100
120
Color Coded Dot Plots
10
20
30
40
50
60
70
80
90
100
120
10
4
FSC-Height -->
3
R4
2
10
10
1
FL2-Height -->
10
R3
R1
R2
10
1
10
2
10
3
10
4
FL4-Height -->
Key to understanding this figure, is the notion that different
populations can be identified by colors and the relationship of
these populations to one another can be monitored.
Page 24
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
50
40
30
0 10
20
Forward Scatter
60
Contour Plot with Projection
0 10
20
30
60
90 Light Scatter
40
50
Page 25
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
3 Color Combinations
4+4+4=12
PE
Positives
4+4=8
FITC
Negatives
Page 26
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
3 Color Combinations
10
100
1000
10
100
1000
1
10
1
Texas Red
10
.1
CD3
CD45
Negative
100
FITC
1000
CD20
IgG
CD8 (dim)
CD2
L
.1
1
K
CD4
Negative
10
CD5
100
1000
Phycoerytherin
PE
FITC
CD7
CD8
CD20
.1
1
CD8
CD3
CD20
CD2
.1
.1
CD3
CD45
CD8
CD3
CD20
1
Texas Red
1
IgG
100
1000
1000
100
10
CD3
CD4
Negative
TR-PE Pattern
TR-FITC Pattern
CD38
.1
4+4+4=12
Phycoerytherin
PE-FITC Pattern
FITC
Page 27
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Page 28
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Lasers used for multicolor studies
Page 29
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Page 30
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Page 31
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Innovative Data Analysis
Page 32
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Page 33
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Some advanced ways of showing data relationships
20
60
100
Classification
Enrico Lugli et al, Università di Modena e Reggio Emilia
Oral Presentation Immunology section 15.30-17.30 today
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
?
Page 34
Spectral analysis allows classification
Page 35
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Conclusions
• The more parameters you have, the more complex the
analysis will be
• But…when you have more parameters (variables) you
have more opportunities for population discrimination
• Display of data in histogram and dotplot formats assists
the analysis process
• Displays in 3D are nice but not particularly useful for
analysis.
• Multiple parameter displays such as PCA or LDA are
more useful for high content data sets
•
Page 36
© 1988-2006 J.Paul Robinson, Purdue University BMS 602 LECTURE 9.PPT
Download