science webinar series slides_imaging in... · sanju ashraf mariana bexiga brendan bury george...
TRANSCRIPT
Webinar SeriesWebinar SeriesWebinar SeriesWebinar SeriesScienceScienceScienceScience
Discovering New Drugs for Disease: Advancing Imaging in the WellDiscovering New Drugs for Disease: Advancing Imaging in the WellDiscovering New Drugs for Disease: Advancing Imaging in the WellDiscovering New Drugs for Disease: Advancing Imaging in the Well
Change the size of any window by dragging the lower left corner. Use controls in top right corner to close or maximize each window.
shows slide window
What each widget does:
shows the audio media player
shows speaker bios
download slides and more info
shows slide window
opens the Ask a Question box
search Wikipedia
shows the audio media player
Facebook login
download slides and more info
LinkedIn login
search Wikipedia
to login to Twitter and send tweetsif you need helpTwitter login (#ScienceWebinar)
Webinar SeriesWebinar SeriesWebinar SeriesWebinar SeriesScienceScienceScienceScienceDiscovering New Drugs for Disease :Discovering New Drugs for Disease :
Brought to you by the Science/AAAS Custom Publishing Office
14 September, 201114 September, 2011
Discovering New Drugs for Disease :Discovering New Drugs for Disease :Advancing Imaging in the WellAdvancing Imaging in the Well
Participating Experts:
Brought to you by the Science/AAAS Custom Publishing Office
Jeremy Simpson, Ph.D.University College DublinDublin, Ireland
David W. Andrews, Ph.D.McMaster UniversityHamilton, Ontario, Canada
Sponsored by:
Di i N D f DiDiscovering New Drugs for Disease:Advancing Imaging in the Well
Jeremy C. Simpson
Professor of Cell Biology,School of Biology and Environmental ScienceSchool of Biology and Environmental Science,Conway Institute of Biomolecular and Biomedical Research,University College Dublin,Ireland
AAAS Science webinar, 14th September 2011
Cells contain spatial and temporal information
spatial informationspatial information
spatial and temporal information
The need for high content screening and analysis
Number of genes / compounds studied
Experimental detail obtained
Human genome sequencing
- comprehensive cDNA libraries- genome-wide RNAi reagents
- novel compound librariesnovel compound libraries
Imaging approaches are central to Systems Biology
Imaging modalities
Experimental consistencyp y
Quantification
Information
Pepperkok & Ellenberg (2006) Nature Rev. Mol. Cell Biol. 7:690‐696‘High‐throughput fluorescence microscopy for systems biology’
Cells Primary cells ‘conventional’ cells specialised cells stable cell lines
HCS experiments - so many decisions to make!
Cells
Plate format
Plating density
Primary cells, conventional cells, specialised cells, stable cell lines
Multi‐well plates (24/96/384 etc.), chambered slides
Sufficient for quantitative analysis but not impinging segmentationPlating density
Transfection
Reagents
Sufficient for quantitative analysis, but not impinging segmentation
Liquid‐based transfection, reverse transfection, viral mediated
Compound number concentration rangeReagents
Incubation time
Assay format
Compound number, concentration range
Sufficient to see phenotype, but minimising toxicity
Fixed cells (end‐point) or live cells (time‐lapse)Assay format
Assay readout
Replicates
Fixed cells (end point) or live cells (time lapse)
Fluorescence based ‐ number of channels required
Numbers of samples per plate, numbers of replicate platesReplicates
Controls
QC
Numbers of samples per plate, numbers of replicate plates
Positive controls (strong and weak), negative controls, plate position
Plating abnormalities, high toxicity, apoptosisQC
Analysis
Plating abnormalities, high toxicity, apoptosis
Cell segmentation and analysis, statistics, variance across screen
HCS experiments - so many decisions to make!
Basicassay
Primaryscreen
Secondaryscreen
Identifycandidates
Characterisecandidates
primary screen Primary screen
High sample number: eg: 80,000 siRNAses g p g ,High image number: eg: >1m images
High cell number: eg: >10m cellsLow parameter number: eg: 3 values/cell
er o
f sam
ple
secondary screen Secondary screen
Low sample number: eg: 2,000 siRNAsLow image number: eg: 20,000 images
Low cell number: eg: 0 5m cells
num
be
Low cell number: eg: 0.5m cellsHigh parameter number: eg: >50 values/cellnumber of parameters / cell
What can HCS / HCA tell us?
How much / how many ? Where ? What effects ?
Intact organelleNot internalisedNone
I t li d DisruptedInternalisedMany
Automated quantitative image analysis
Image Processing
• background correction and subtraction• border objects (to keep or not to keep)
• object identification and counting• object segmentation
• object dilation
Object intensity
• mean intensity of fluorescence
Object texture
• Granularity
Object morphology
• area y• variation of intensity (stdev)
• object distribution• radial intensity distribution
• o lo alisation
y• Haralick texture features• Gabor texture features• SER (Spot‐edge‐ridge)
• width• length
• roundness / circularity• idth / len th ratio • co‐localisation• width / length ratio
raw data mask generation
Automated quantitative image analysis
raw data mask generation
subtractive mask segment cells
‘good cells’
‘bad cells’
extract cellular features- morphology- intensity
t t
classify and analyse
- texture
Cell segmentation
fibroblast astrocyte neuron
Physiological relevance of y gcells for particular assay
Ease of segmentation and quantification during analysis
Cell segmentation
Primary cilia Nuclear segmentation and dilation
Bacterial infection Cytoplasm segmentationBacterial infection Cytoplasm segmentation
Analysis of internalisation
Pathogenslow
Plasma membrane
PathogensToxins
NanoparticlesDrugs
low
Early Recycling
medium
yendosome
y gendosome
LatehighTGN
Lateendosome
Golgi
high
Lysosome
Golgi
ens
/ cel
l
Endoplasmic reticulum
path
oge
pathogen load
Analysis of internalisation0 mins
Plasma membrane
Earlyendosome
Recyclingendosome
5 mins
TGNLate
endosome
Lysosome
Golgi
30 mins
Endoplasmic reticulum
Image texture analysis
Intensity
Spot Hole
Ridge Valley
Saddle Edge
Bright Dark
SER texture analysis Image courtesy of Perkin Elmer Columbus / Acapella software)
Image texture analysis
HomogeneityCorrelation
Spearman correlationSubcellular localisations
IntensityGabor minGabor maxSum varianceHoleHoleDarkValleySpotEdgeBright RidgeSaddleContrast
y y x e e k y t e t e e t
Hom
ogen
eity
Cor
rela
tion
Inte
nsit y
Gab
or m
inG
abor
ma x
Sum
var
ianc
eH
ole
Dar
kVa
lley
Spo
tE
dge
Brig
htR
idge
Sad
dle
Con
trast
High differenceHigh similarity
H S
Image texture analysis
nt 2
All featuresSubcellular localisations
Com
pone
Component 1O f t
ent 2
One feature removed
Com
pone
Component 1
0 i 5 i
Analysis of internalisation
Plasma membrane0 mins 5 mins 30 mins
Earlyendosome
Recyclingendosome
Control RNAi
TGNLate
endosome
0.4
0.6
0.8
1
GolgiEndosomes and Golgi
E d
Lysosome
Golgi0
0.2
RNAi
Endosomes
5 15 30 60 90 120 min
Endoplasmic reticulum
0.4
0.6
0.8
1
RNAi
GolgiEndosomes and Golgi
0
0.2
0.4Endosomes
5 15 30 60 90 120 min
Acknowledgements
Collaborators
Oliver Blacque (UCD, Dublin)Kenneth Dawson (UCD, Dublin
Wim Meijer (UCD, Dublin)Rainer Pepperkok (EMBL, Heidelberg)
John Presley (McGill, Montreal)
UCD-Dublin
S j A h fIrish Research Council for Science,
Engineering & Technology
Sanju AshrafMariana BexigaBrendan BuryGeorge GaleaKenan Handzic
Nora LieggiElaine O’Neill
Angela PanarellaVasanth Singan
Juan Varela
Science Foundation Ireland
Perkin ElmerJuan Varela
Webinar SeriesWebinar SeriesWebinar SeriesWebinar SeriesScienceScienceScienceScienceDiscovering New Drugs for Disease :Discovering New Drugs for Disease :
14 September, 201114 September, 2011
Discovering New Drugs for Disease :Discovering New Drugs for Disease :Advancing Imaging in the WellAdvancing Imaging in the Well
Brought to you by the Science/AAAS Custom Publishing Office
Participating Experts:
Brought to you by the Science/AAAS Custom Publishing Office
Jeremy Simpson, Ph.D.University College DublinDublin, Ireland
David W. Andrews, Ph.D.McMaster UniversityHamilton, Ontario, Canada
Sponsored by:
Live Cell imaging: Measuring apoptosis and other cellular responses to drugsusing automated image analysis and machine learning
Drugs
Cellular responses to drugs: Apoptosis, Stress, Autophagy
Damaged Cell
One potential outcome:
Bcl‐2 Proteins Regulate Programmed Cell Death
Bax tBid
Bcl-XL tBid
Caspases execute the cell
Caspases 3 & 7
Using Annexin V to measure apoptosis
Measures externalization of phosphatidylserineve
0.4
0.6nn
V positiv
0.2
entage An
0
CDMSO 0.1
0.3 1 3 10 0.1
0.3 1 3 10 0.1
0.3 1 3 10 0.1
0.3 1 3 10
Perce
TNF Thaps BFA DTTp
Using Annexin V to measure apoptosis
Measures externalization of phosphatidylserineve
0.4
0.6nn
V positiv
0.2
entage An
0
CDMSO 0.1
0.3 1 3 10 0.1
0.3 1 3 10 0.1
0.3 1 3 10 0.1
0.3 1 3 10
Perce
TNF Thaps BFA DTT
Expensive, slow, not very sensitive
p
Control 100 mM NAO 10x DTT
Measuring apoptosis usingNonylacridine Orange (NAO)
Control 100 mM NAO 10x DTT
10x BFA 10x TNFa +CHX
Tony Collins, Fei Geng, Sudeepa Dixit
Measuring apoptosis usingNonylacridine Orange (NAO)
Control 100 mM NAO 10x TG
Analysis of Cell textures
Measuring apoptosis usingNonylacridine Orange (NAO)
Control – 100 nM NAO 10x DTT
Tony Collins, Fei Geng, Sudeepa Dixit, Sean Cianflone
Measuring apoptosis usingNonylacridine Orange (NAO)
Control – 100 nM NAO 10x DTT
Tony Collins, Fei Geng, Sudeepa Dixit, Sean Cianflone
Measuring apoptosis usingNonylacridine Orange (NAO)
Control – 100 nM NAO 10x DTT
Tony Collins, Fei Geng, Sudeepa Dixit, Sean Cianflone
Measuring apoptosis usingNonylacridine Orange (NAO)
Control – 100 nM NAO 10x DTT
010x BFA
Tony Collins, Fei Geng, Sudeepa Dixit, Sean Cianflone
Texture analysis M hi l i d l ifi i d l i i l iMachine learning, automated classification and multiparametric clustering
Support Vector Machine
Sean Cianflone
Using texture analysis to measure apoptosis
Measures changes in subcellular membranes
0.75
1fe
d tr
eate
d Treated’
0.25
0.5
ctio
n cl
assi
fFraction
‘T
0
CD
MSO
0.1x
TNF
0.3x
TNF
1xTN
F3x
TNF
10xT
NF
0.1x
Tg0.
3xTg
1xTg
3xTg
10xT
g
0.1x
BFA
0.3x
BFA
1xB
FA3x
BFA
10xB
FA
0.1x
DTT
0.3x
DTT
1xD
TT3x
DTT
10xD
TT
Frac
CDMSO 0.1
0.3 1 3 10 0.1
0.3 1 3 10 0.1
0.3 1 3 10 0.1
0.3 1 3 10
TNF Thaps BFA DTT
F
p
Tony Collins, Fei Geng, Sudeepa Dixit
Using texture analysis to measure apoptosis
Measures changes in subcellular membranes
0.75
1fe
d tr
eate
d Treated’
0.25
0.5
ctio
n cl
assi
fFraction
‘T
0
CD
MSO
0.1x
TNF
0.3x
TNF
1xTN
F3x
TNF
10xT
NF
0.1x
Tg0.
3xTg
1xTg
3xTg
10xT
g
0.1x
BFA
0.3x
BFA
1xB
FA3x
BFA
10xB
FA
0.1x
DTT
0.3x
DTT
1xD
TT3x
DTT
10xD
TT
Frac
CDMSO 0.1
0.3 1 3 10 0.1
0.3 1 3 10 0.1
0.3 1 3 10 0.1
0.3 1 3 10
TNF Thaps BFA DTT
F
Tony Collins, Fei Geng, Sudeepa Dixit
Using texture analysis of cellular unhappiness
% SVM classified as treated highest TNF as positive control training group.
Dose response curves
Ctr
highest TNF as positive control training group.
E t i i A t iST
ActDTN
TNF
Intrinsic Apoptosis
Extrinsic Apoptosis
ER Stress/UPR TNTAMCtrCtr
ER Stress/UPRAutophagic cell death
Ctr
Increasing dose of compounds
Tony Collins, Fei Geng, Sudeepa Dixit
But can we tell them apart???
% SVM classified as treated highest TNF as positive control training group.
Dose response curves
Ctr
highest TNF as positive control training group.
E t i i A t iST
ActDTN
TNF
Intrinsic Apoptosis
Extrinsic Apoptosis
ER Stress/UPR TNTAMCtrCtr
ER Stress/UPRAutophagic cell death
Ctr
Increasing dose of compounds
Tony Collins, Fei Geng, Sudeepa Dixit
Dose Response Curves 3 way classification
90100 Ctr
TAMTN
60708090
cells
(%)
reated
’
405060
Trea
ted
cPe
rcen
t ‘Tr
102030P
-2 -1 0 1 2 3 4 50
TN log(conc.)
Tony Collins, Fei Geng, Sudeepa Dixit
Dose Response Curves 3 way classification
90
100 CtrTAMTN
6070
8090
cells
(%)
reated
’
3040
5060
Trea
ted
cPe
rcen
t ‘Tr
1020
30P
-1 0 1 2 3 4 5 60
TAM log(conc.)
Tony Collins, Fei Geng, Sudeepa Dixit
Measuring cellular functions in un-engineered live cellse.g. primary cells, stem cells, cell lines etc.
Use a single assay system to measure: ApoptosisER StressAutophagyAutophagyIntoxication____________________________________Differentiaion of stem cellsDifferentiaion of stem cellsRadiation damage Membrane biogenesisInsulin secretionProtein traffickingEtc.
WWW.ANDREWSLAB.CA
Canada Research Chairswww.chairs-chaires.gc.ca
Webinar SeriesWebinar SeriesWebinar SeriesWebinar SeriesScienceScienceScienceScienceDiscovering New Drugs for Disease :Discovering New Drugs for Disease :
14 September, 201114 September, 2011
Discovering New Drugs for Disease :Discovering New Drugs for Disease :Advancing Imaging in the WellAdvancing Imaging in the Well
Brought to you by the Science/AAAS Custom Publishing Office
Participating Experts:
Brought to you by the Science/AAAS Custom Publishing Office
Jeremy Simpson, Ph.D.University College DublinDublin, Ireland
David W. Andrews, Ph.D.McMaster UniversityHamilton, Ontario, Canada
Sponsored by:
Webinar SeriesWebinar SeriesWebinar SeriesWebinar SeriesScienceScienceScienceScienceDiscovering New Drugs for Disease :Discovering New Drugs for Disease :
14 September, 201114 September, 2011
Discovering New Drugs for Disease :Discovering New Drugs for Disease :Advancing Imaging in the WellAdvancing Imaging in the Well
Brought to you by the Science/AAAS Custom Publishing OfficeBrought to you by the Science/AAAS Custom Publishing Office
Look out for more webinars at:www.sciencemag.org/webinar
To provide feedback on this webinar, please e‐mailp p
your comments to [email protected]
For related information on this webinar topic, go to:
www.perkinelmer.com/imaging
Sponsored by: