Yeast
Data analysis
Summary and conclusions
Hierarchical modelling of automated imaging data
Darren Wilkinson
http://tinyurl.com/darrenjw
School of Mathematics & Statistics,
Newcastle University, UK
Theory of Big Data 2
UCL, London
6th–8th January, 2016
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Overview
Background: Budding yeast as a model for genetics
High-throughput robotic genetic experiments
Image analysis and data processing
(Stochastic) modelling of growth curves
Hierarchical modelling of genetic interaction
Summary and conclusions
Joint work with Jonathan Heydari, Keith Newman, Conor Lawless
and David Lydall (and others in the “Lydall lab”)
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Budding yeast biology and genetics
Synthetic genetic array
Telomeres
HTP yeast SGA robotic screens
Saccharomyces cerevisiae
Saccharomyces cerevisiae, often known as budding yeast, and
sometimes as brewer’s yeast or baker’s yeast, is a single-celled
eukaryotic organism
Eukaryotic cells contain a nucleus (and typically other
organelles, such as mitochondria)
It is useful as a model for higher eukaryotes, having a great
deal of biological function conserved with humans
It is the most heavily studied and well-characterised model
organism in biology (eg. first fully sequenced eukaryote)
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Budding yeast biology and genetics
Synthetic genetic array
Telomeres
HTP yeast SGA robotic screens
Synthetic Genetic Array (SGA)
Possible to obtain a library of around 4,500 mutant strains,
each of which has one of the non-essential genes silenced
through insertion of a (kanMX) antibiotic resistance cassette
and tagged with a unique DNA barcode
These strains (stored frozen in 96-well plates) can be
manipulated by robots in 96-well plates (8×12), or on solid
agar in 96, 384 or 1536-spot format
Synthetic Genetic Array (SGA) is a clever genetic procedure
using robots to systematically introduce an additional
mutation into each strain in the library by starting from a
specially constructed query strain containing the new mutation
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Budding yeast biology and genetics
Synthetic genetic array
Telomeres
HTP yeast SGA robotic screens
Telomeres
The ends of linear chromosomes require special protection in
order not to be targeted by DNA damage repair machinery
(bacteria often avoid this problem by having just one
chromosome arranged in a single loop)
Telomeres are the ends of the chromosomal DNA (which have
a special sequence), bound with special telomere-capping
proteins that protect the telomeres
CDC13 is an essential telomere-capping protein in yeast
cdc13-1 is a point-mutation of cdc13, encoding a
temperature-sensitive protein which functions similarly to
wild-type CDC13 below around 25 ◦ C, and leads to
“telomere-uncapping” above this temperature
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Budding yeast biology and genetics
Synthetic genetic array
Telomeres
HTP yeast SGA robotic screens
Basic structure of an experiment
1
Introduce a mutation (such as cdc13-1) into an SGA query
strain, and then use SGA technology (and a robot) to cross
this strain with the single deletion library in order to obtain a
new library of double mutants
2
Inoculate the strains into liquid media, grow up to saturation
then spot back on to solid agar 4 times
3
Incubate the 4 different copies at different temperatures
(treatments), and image the plates multiple times to see how
quickly the different strains are growing
4
Repeat steps 2 and 3 four times (to get some idea of
experimental variation)
5
Repeat steps 2 to 4 with a “control” library that does not
include the query mutation
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Budding yeast biology and genetics
Synthetic genetic array
Telomeres
HTP yeast SGA robotic screens
Some numbers relating to an experiment
Initial SGA work (introducing mutations into the query and
the library) takes around 1 month of calendar time, and
several days of robot time
The inoculation, spotting and imaging of the 8 repeats takes
1 month of calendar time, and around 2 weeks of robot time
The experiment uses around £3,000 of consumables (plastics
and media)
The library is distributed across 72 96-well plates or 18 solid
agar plates (in 384 format, or 1536 in quadruplicate)
If each plate is imaged 30 times, there will be around 35k
high-resolution photographs of plates in 384 format,
corresponding to around 13 million colony growth
measurements (400k time series)
This is big data!
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Data analysis pipeline
Image processing (from images to colony size measurements)
Fitness modelling (from colony size growth curves to strain
fitness measures)
Modelling genetic interaction (from strain fitness measures to
identification of genetically interacting strains, ranked by
effect size)
Possible to carry out three stages separately, but benefits to joint
modelling through borrowed strength and proper propagation of
uncertainty. Not practical to integrate image processing step into
the joint model, but possible to jointly model second two stages.
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Automated image analysis (Colonyzer)
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Growth curve
A
his3Δ
htz1Δ
0
6
12
18
24
30
36
42
0.10
0.05
0.00
Normalised cell density (AU)
0.15
B
0
10
20
30
40
Time since inoculation (h)
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Growth curve modelling
We want something between a simple smoothing of the data
and a detailed model of yeast cell growth and division
Logistic growth models are ideal — simple semi-mechanistic
models with interpretable parameters related to strain fitness
Basic deterministic model:
dx
= rx(1 − x/K),
dt
subject to initial condition x = P at t = 0
r is the growth rate and K is the carrying capacity
Analytic solution:
x(t) =
Darren Wilkinson — UCL, 7/1/2016
KP ert
K + P (ert − 1)
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Statistical model
Model observational measurements {Yt1 , Yt2 , . . .} with
Yti = xti + εti
Can fit to observed data yti using non-linear least squares or
MCMC
Can fit all (400k) time courses simultaneously in a large
hierarchical model which effectively borrows strength,
especially across repeats, but also across genes
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Growth curve model
l m
n
Time Point
ylmn
ŷlmn
Repeat
rlm
Klm
orf ∆
τlK
τlK
σ K,o
Kp
νl
rlo
Klo
σ r,o
σ τ,K
τ K,p
rp
Darren Wilkinson — UCL, 7/1/2016
σν
σ τ,r
τ r,p
P
Population
νp
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Colony fitness
The results of model fitting are estimates (or posterior
distributions) of r and K for each yeast colony, and also the
corresponding gene level parameters
Both r and K are indicative of colony fitness — keep separate
where possible
Often useful to have a scalar measure of fitness — many
possibilities, including rK, r log K, or MDR×MDP, where
MDR is the maximal doubling rate and MDP is the maximal
doubling potential
Statistical summaries can be fed in as data to the next level of
analysis (or, ultimately, modelled jointly as a giant hierarchical
model)
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Epistasis
From Wikipedia:
“Epistasis is the interaction between genes. Epistasis takes
place when the action of one gene is modified by one or
several other genes, which are sometimes called modifier
genes. The gene whose phenotype is expressed is said to be
epistatic, while the phenotype altered or suppressed is said to
be hypostatic.”
“Epistasis and genetic interaction refer to the same
phenomenon; however, epistasis is widely used in population
genetics and refers especially to the statistical properties of
the phenomenon.”
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Multiplicative model
Consider two genes with alleles a/A and b/B with a and b
representing “wild type” (note that A and B could potentially
represent knock-outs of a and b)
Four genotypes: aa, Ab, aB, AB. Use [·] to denote some
quantitative phenotypic measure (eg. “fitness”) for each
genotype
Multiplicative model of genetic independence:
[AB] × [ab] = [Ab] × [aB]
no epistasis
[AB] × [ab] > [Ab] × [aB]
synergistic epistasis
[AB] × [ab] < [Ab] × [aB]
antagonistic epistasis
Perhaps simpler if re-written in terms of relative fitness:
[AB]
[Ab] [aB]
=
×
[ab]
[ab]
[ab]
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Genetic independence and HTP data
Suppose that we have scaled our data so that it is consistent
with a multiplicative model — what do we expect to see?
The independence model [AB] × [ab] = [Ab] × [aB] translates
to
[query : abc∆] × [wt] = [query] × [abc∆]
In other words
[query : abc∆] =
[query]
× [abc∆]
[wt]
That is, the double-mutant differs from the single-deletion by
a constant multiplicative factor that is independent of the
particular single-deletion
ie. a scatter-plot of double against single will show them all
lying along a straight line
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Statistical modelling
Assume that Fclm is the fitness measurement for repeat m of
gene deletion l in condition c (c = 1 for the single deletion
and c = 2 for the corresponding double-mutant)
Model:
Fclm ∼ N (F̂cl , 1/νcl )
log F̂cl = αc + Zl + δl γcl
δl ∼ Bern(p)
δl is a variable selection indicator of genetic interaction
Then usual Bayesian hierarchical stuff...
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Genetic interaction model
c
l
n
Repeat
αc
σcγ
Fclm
F̂clm
γcl
Condition
νcl
orf ∆
δl
Zl
σZ
Zp
Darren Wilkinson — UCL, 7/1/2016
σν
Population
νp
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
SAN1
60
NMD2
UPF3
EXO1
NAM7
RAD24
RAD17
DDC1
BMH1
LST4
50
IPK1
RPN4
VIP1
UBC4
RPL16B
SCS2
YDL119C
SAC7
TMA22
30
40
TMA20A
CKA2
YER119C−
MPH1
MTC5
VPS8 FYV10
EBS1
ARX1 DOA1 SRN2
VID28
YNR005C
MCK1
RPL13ARPL2A KRE1
GTR1
OGG1
MCX1
RAD9
YDL118W
YIL055C
MEH1
CHK1
RPL8A RPL35B
RIF2
VPS4
MUM2
BUD27
RTC1
YBL104C
NRG2
EDE1
FTR1
RIC1
RPL9A
BAS1
NOP16
UFD2
KIN3
RPL37B
VPS9
NPR1 PET122
YIL057C
ELM1
SIM1
DBP3
AIR1
FET3
BST1
UBP6
SHE4
ARO8
GEF1
RPL35A
RPL8B
RPL43A
SLA1
YMR057C
SSF1
RPO41
UTH1
UFD4CRP1
ERG3
TKL1
POT1
RPL42A
TAL1
DID2
SNX4
NUP2
BUD14
PIB2
BNR1
NPT1
CST6
DIE2
CYK3
YPT6
GUF1
RVS167
YLR261C
OCA6
UBX4
RPL27B
YIL161W
GUP1
GPA2
RPL11B
UBX7
COX23
QCR9
VPS60
MNI1
AYR1
PHO2
RPL37A
CCC2
YML010C−
B
RAD23
RPP1B
YLR111W
TRP1
RIM1 RPL16A YKR035C
CAP2
CTP1
YMR153C−
HCM1
YKE4
SIW14
SUB1
GZF3
GPD2
XBP1
SAS4 A
ARO2
RPL6B
YDL176W
MRT4 SUR4
CBP4
RHR2
PHB2
PFK26
OST3
KEX1
RNR3
FCY2
TNA1
OAC1
YNL226W
VPS24
COX7
VPS51 RPL24ACCW12
RPL4A
VPS21
RPL36A
VID24
PUS7
COX12
YJ
L185C
SKI8
YPL080C
RPL34A
SAS5
NUP53
AVT5
LRP1 YPR044C
YLR218C
SYS1
YER093C−
AYBL083C
ALG3
RPL24B
RPL43B
RPL26A
LIA1
YMR193C−
APBP2
PER1
OST4
MNI2
YDR271C
MSS18
COQ10
BMH2
SKI2
CRD1
TRM44
A BUD28
RPS4A
ICT1
CLG1
YBL059W
VID30
KAP122
STM1
CKB2 YLR338W YML013C−
YLR290C
HPT1
NAP1
SFH5
SER2
DBF2
YDL050C
FMP35
RPL21B VPS35
GDH1
CBF1
YJ GID8
L206C
HSE1
IMG2
FEN1
ALG12
YBR144C
MDM34
YGR259C
RPL29
VAM10
SKI7
VRP1 MET18
KEX2
VAC14
VPS17
DOT1
OCA4
OTU2
LAC1
VHS2
YPT7
SPT2
CBT1
EOS1
ERJ
5RBS1
MON1
RPL33B
CAT5
OMS1
NBP2
MNE1
YLR407W
VPS5
RAV1
SAS2
TEF4VPS38
CYT2
YGL057C
BEM4
RPL40B
AEP2
ARC18
RSM25
OYE2
DSK2
UPS1
RPP2B
UBP2
FUS3
YDR269C
TOM70
APT1
MTC3
CKB1
ERG6
RPL17B
RPL23A
YTA7
PSD1
ACE2
YUR1
RPL9B
LRG1
SWI4
CYT1
QCR2
YLR402W
RMD5
YDR203W
CYC2
CKA1
YBR266C
LEU3
MDM38
SWF1
IRC3
HSP12
YGL218W
YGL149W
GOS1
ALG9
PHO87
YPL062W
TPS3
PTC1
MRPL1
RNH202
ERD1
YDR266C
HAP2
UBP3
CAC2HAP5
MMM1
YDR348C
RVS161
YLR217W
MGR2
SYF2
ALD6
PHO80
QCR6
ALG5
PHB1
FYV1
PEX32
VPS1
YLR091W
PMS1
HAP3
YBR232C
YNR029C
PMP3
PKR1
STP1
YML090W
GET1
YDR537C
PEX8ALG6
MNN11
MIS1
YAP1801
DRS2
MIR1
YLR184W
YBR226C
MLH1
YDR049W
PEX15
YJ
L211C
CPS1
BNA2
PPZ1
PEX13
URE2
PEX2
BUL1
OCA5
RTC4
CYM1
EMI1
YOR309C
ARO1
NBA1
RGS2
PTP3
VTS1 YKL121W
KRE11
TOM7
YHL005C
GEM1
TOM6
TOP1
HAP4IMP2
JJJ1
RPP1A YJ L120WBRE5
VAM3
PEP8
VAM6
RPE1
VAM7
PUF6
ASC1
AAH1
ATP10
PAH1
ZWF1
RPL22A
REI1
MDM10
20
DEG1
CCZ1
YGL024W
SPE2
10
BTS1
RTT109
MMS22
MRE11
RAD50
CTK1
SIR3
LSM6
YBR028C
YBR099C
YDR262W
ALD3
GPH1
KNS1
EMI5
YPR004C
FIT2
YJ
R154W
YDL012C
NAT1
ASR1
YPL035C
YMR310C
PGM2
ARA1
YMR206W
HIS3
YOR052C
PNG1
MNT4
TCM62
HSP82
YBL071C
AZF1
STD1
ECM5
UME1
ZRT3
CSM3
KAR9
YPS6
PAN3
PUF3
SYT1
AKL1
TOS3
REH1
PET130
YBR246W
YNL011C
EFT2
ESBP6
YDL109C
GRS1
BFA1
YDR149CSNF1
DIP5
YPR098C
MET16
HIR3
HAT1
YPR039W
YBR025C
YGL217C
MRN1
YAP1
NCS2 YPR084W
SWC5 SET2
ERG24
LDB18
CHL1
CLB2 PAN2
URM1
DEP1
RAD27
PAC1EAF3 MBP1
APQ12 SHE1YGL042C
YMR074C
NTO1
YBR277C
THR4 UBA4
MCM16
SBP1
YPL102C
CSN12
YOR251C
BCK1
DBR1
DPB3
SLT2
IMP2'
TMA19
ELG1
DYN1
BUB2
FKH2
J NM1
PPH3
YLR143W
MAK3
DYN3
DPB4
PTC2
KTI12 YNL120C
PPQ1
ARP1 IRA2 PHO13
HGH1
ELP6RTF1
TEL1 HSP26
JTIP41
JPUF4
JEST3
3 HMT1
HSP104
RRD1
YPR050C
YKR074W
DPH5 FKH1
FPS1
:::RTT103
GIM3
PTC6
ELP4
NUP133
NPL4
HPR5 IRC21
EST1 PPT1
IKI3
BEM2
CDC73
RMD11
LDB19
CTF4
XRS2
MFT1
DPH1
DCC1 THP2
INP52
MAK10
SWM1
PAT1
LTE1 DPH2
MAK31
ELP2
SRB2
MRC1
POL32 CLB5
RIF1
ELP3
YKU70
STI1
:::MRC1
0
Fitness (F) of orfΔ cdc13-1 double mutants at 27°C (doublings2 / day)
Genetic interaction results
0
50
100
150
Fitness (F) of orfΔ ura3Δ double mutants at 27°C (doublings2 / day)
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Joint modelling of growth curves and genetic interaction
We can integrate together the hierarchical growth curve model
and the genetic interaction model into a combined joint model
This has usual advantages of properly borrowing strength,
proper propagation of uncertainty, etc.
Also very convenient to avoid requiring a scalar measure of
“fitness”
If yclmn is the colony size at time point n in repeat m of gene
l in condition c, then
yclmn ∼ N (ŷclmn , 1/νcl )
ŷclmn = X(tclmn ; Kclm , rclm , P )
log Kclm ∼ N (αc + Klo + δl γcl , 1/τclK )
log rclm ∼ N (βc + rlo + δl ωcl , 1/τclr )
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
Joint model
c
l
m
n
σcγ
γcl
Time Point
yclmn
ŷclmn
Repeat
rclm
Kclm
αc
σcω
Condition
ωcl
νcl
βc
Klo
δl
rlo
σ r,o σ τ,r
σ K,o σ τ,K
Kp
τ K,p
orf ∆
τlr
τlK
rp
Darren Wilkinson — UCL, 7/1/2016
τ r,p
P
Population
σν
νp
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Data analysis pipeline
Growth curve modelling
Modelling genetic interaction
80
SAN1
NMD2
EXO1
UPF3
RAD24
NAM7
RAD17
DDC1
BMH1
IPK1
LST4
SCS2
YDL119C
VIP1 UBC4
SAC7
TMA22
RPL16B
YER119C−
A
TMA20
CKA2
MPH1FYV10
MTC5
VPS8
EBS1
ARX1
RPL2A SRN2 RAD9
DOA1
MCK1
YNR005C
RPL13A
GTR1
CHK1
OGG1
YDL118W
VID28
YIL055C
KRE1
RIF2
MUM2
MEH1 RPL8A VPS4
RPL35B
BUD27
MCX1
FTR1
EDE1
RTC1
RPL9A
RIC1
YBL104C
BAS1
NOP16
KIN3
UFD2
RPL37B
NPR1 FET3
NRG2
YIL057C
VPS9
DBP3
PET122
ELM1
UBP6
AIR1
BST1
ARO8
SIM1 CRP1
GEF1
SSF1SHE4
RPL35A
YMR057C
RPL8B
ERG3
SLA1
TKL1
TAL1
UTH1
POT1
RPO41 BUD14
NUP2
UFD4
RPL42A
BNR1
SNX4
RPL43A
DID2
PIB2
DIE2
YLR261C
CYK3
RPL27B
CST6 YPT6
NPT1
UBX4
RVS167
GUF1
YIL161W
OCA6
RPL11B
COX23
GUP1
QCR9
AYR1
PHO2
MNI1
CCC2
RPL37A
RPP1B
VPS60 UBX7
YLR111W
RAD23
YML010C−
YMR153C−
GPA2
CAP2
RPL16A B CTP1
RIM1
HCM1
GZF3
SAS4 A
XBP1
YKE4
TRP1
GPD2
SIW14
RPL6BSUB1
YKR035C
YDL176W
ARO2
PHB2
KEX1
RNR3
MRT4 SUR4
CBP4
FCY2
PFK26
RHR2
COX12
TNA1
RPL4A
OST3
VID24
OAC1
PUS7
COX7
VPS21
RPL34A
YJ
L185C
SKI8
SAS5
RPL36A
AVT5
SYS1
NUP53
YBL083C
YPL080C
YER093C−
A
VPS51 RPL24ACCW12
YLR218C
ALG3
RPL24B
LRP1 YPR044C
LIA1
RPL26A
PER1
BMH2
HPT1
YDR271C
YNL226W
VPS24RPL43B
YMR193C−
APBP2
YBL059W
COQ10
SFH5
VID30
MSS18
CRD1
MNI2
YDL050C
SKI2
YLR290C
TRM44
ICT1
KAP122
CLG1
BUD28
FMP35
HSE1
STM1
NAP1
YJ
L206C
YBR144C
RPS4A
CKB2
OCA4
SER2
CBF1
GDH1
GID8
DBF2 IMG2
RBS1
ALG12
VPS35
ERJ
5LRG1
DOT1
SAS2
YGR259C
YML013C−
AVAM10
OMS1
VAC14
OTU2
LAC1
NBP2
SKI7
VHS2
RPL29
DSK2
CBT1
KEX2
YLR407W
MDM34
SPT2
RPL21B
VPS17
MNE1
RSM25
RAV1
OST4
YDR203W
IRC3
MET18
RPL33B
YGL057C
RMD5
YLR338W
FEN1
OYE2
MIS1
VPS38
CAT5
YUR1
APT1
MTC3
YBR232C
UBP2
RNH202
CYT2
TOM70
YLL029W
PHO87
MON1YPT7
BEM4
LEU3
VPS5TEF4
CKA1
ALG9
HSP12
YDR266C
ALG5
CYC2
YDR348C
ERG6
RPL40B
TPS3
YLR091W
QCR6
ERD1
ARC18
YBR226C
ACE2
YDR537C
LHP1
PMP3
FYV1
YLR217W
MMS2
GOS1
FUS3
BNA2
MDM38
RTC4
RPL23A
NMD4
CYT1
PHB1
SWI4
MNN11
YAP1801
YTA7
AEP2UPS1
CPS1
YDR269C
RPL9B
PKH3
QCR2
YML090W
OCA5
ISU2
YBR285W
MDL1
RPL17B
MRPL1
HAP2
PPZ1
YGL149W
GET1
YKL121W
MIR1
ALG6
PEX8
PSD1 CKB1
ASM4
CYM1
URE2
RPP2B
VRP1 YLR402W
PTP3
YOR131C
SWF1
HAP5
HAP3
PEX13
ROT2
YDR049W
ILV6
BUL1
YML108W
UBC13
NBA1
MLH1
PMS1
CAC2
YJ
L211C
KRE11
PEX2
DCW1
MGR2
IOC3
PTC1 UBP3 RVS161
YNR029C
PEX15
APL1
ARL1
NFI1
YDR467C
MMM1
YML053C
TOS1
MVB12
CWH41
YBR266C EOS1
YPS7
YPL062W
COX8
RAX2
ALG8
YOR008C−
ABP1
TOM6
IMP2
YGL218W
VTS1
PEX10
RIT1
SRL1
TOM7
CTI6A
FIS1
YJ
L215C
PEX32
TPS1
PEX14
ARL3
RGS2
YAL004W
RGA1
LEM3
YNL105W
YDR506C
SYF2
FMP36
VPS1
RNH201
RSM28
YBR224W
PSR2
YDR029W
ADE1
YOR082C
SKG3
MSC6
POC4
MFB1
OCA1
DCR2
PHO80
YLR184WALD6
STP1
OSH3
PEP8 TMA16
YLR118C
SSA1
SWA2
PET127
YBR134W
RPL41B
COG7
OST5
GEM1
YPL041C
INP53
ASH1
VPS75
RNH203
AHP1
COX5B
EMI1
MET3
YER077C
PKR1DRS2
YNL109W
BCH2
YHL005C
YBR238C
BRE5
VTC4
YML119W
YCR051W
VPS30
MAD1
TOP1VPS27
ITR1
MEP3
APE2
YGR242W
COX5A
MGA2
APP1
YML102C−
A
RTC6
NDL1
YGL235W
REX4
YAL058C−
A
TFP1
YKL158W
YAF9
YIL166C
ERF2
MLH2
J
SN1
SER1
PCT1
GGA2
RTG2 RPP2A
PRS5
YHL044W
KES1
IRC8
PEX12 YGL046W
VAM6
SYC1
YIA6
BEM3
BUG1
ARO1
MGR1
YOL079W
PTH1
YHM2
YLR282C
GIS3
YIL064W
YER078C
YGR226C
APL3
FMP25
YER066W
YSA1
OPI10
YOR309C
ATP10
SLM1
PIN4
ENT2
ECM22
RTT102
YMC2
YGL132W
:::GAL11
SPO11
SRF6
SBA1
RPP1A YJ L120W
IES2VPS29
YKL077W
YOR022C
YLR334C
MSH2
IMD4
YDR445C
PRX1
SRF4
ALB1
VAM3
RHO5
COG8
ATG27
HOS2
PEX5
HAP4
RPE1
PAM17
OCA2
PMT2
YPT32
ASC1
PEX30ATG5
KEL1
JVAM7
JJ1
MRP49
INP2
PEX25
YNR020C
APS2
VPS13
PAH1
PEX4
HOM2
SIF2
YER084W
YPL105C
GTR2
MNN2
GDS1
LYS20
SKI3
GIS4
ERV14
TLG2
AAH1
PEX3
SNC2
PEX6PIH1
RPS24A
MDM10
KEM1 RPL22A
TPM1PUF6 YLR404W
UBA3SSH1
FSF1
HCR1 DEG1
RPL19A
REI1
ATP18
RPS14A
MOG1
ZWF1
YNL198C
CHZ1
ICE2
CCZ1
PHO86
BUD13
FKS1
YNG1
BUB3
MLS1
HUL5
PRM4
HST2
SSK1
RHO2
CCS1
STB5GAS1
MET31
PCP1
YIL102C
GCV2
HMX1
ATH1
BSP1
NRM1
POM34
MMR1
YDL071C
NUP60
VPS53
ZRT1
YLR434C
PRE9
PUS4
MCR1
BIK1
NCE102
:::GUK1
YOL153C
BUB1
YBR028C
FYV12
PBY1
SCS7
RTG3
OM45
AAD3
YDR262W
ARR1
AGA1
GSH2
LRE1
HPC2GPH1
KIP2
RPS30B
SHE9
CSR2
SKT5
J ID1
ALD3
CWH43
STE50
APC9
PET494
KNS1
IOC2
WWM1
UME1
YPR096C
YCF1
SPE2
SCD6
RPS1BVPS41
PHO88
EMI5
SEM1
MPA43
NAT1
RFM1
RTG1
ARA1
YPL183C
AKL1
YBR099C
MRPS9
YDL012C
YJ
R154W
YMR206W
YMR144W
AUA1
YPL225W
EAF6
MDM12
RPL14A
YPR148C
RPS23B
AKR2
ARF1
SPO14
PGM2
CUE3
ADY4
PNG1
SPE1
IRC15
IBD2
HIS3
RCE1
YPR090W
FIT2
SYT1
MSN5
YGL024W
ZRT3
DIA2 THR1
LAS21
VIK1
ECM5
TSA1
MNT4
BUD19
STB1
YPS6
RBL2
YBR246W
PET130
REH1
RPA14 ELF1
STD1
YMR310C
IXR1YDR149C
NIF3
AHA1
EFT2
TOS3
YPL035C
YNL011C
CHS3
HSP82
AZF1
YPR097W
PUF3
IOC4
SPE3
GRS1
BFA1
TCM62
ESBP6
YPR004C
SWC5
YOR052C
MRN1
KAR9
PAN3
YDL109C
APL4
ASR1
RPS27A
NCS6
MID2
YNL140C
ERG24
YNR004W
CLB2
YBL071C
DIP5
YBR025C
MBP1
CSN12
MKS1
CSM3 PAN2
YAP1
SET2
KRE28
CPR7 SLX9 URM1LDB18
PAC1
CHL1
YGL042C
SLT2
YPR098C
SHE1 LEO1
HAT1
BUD21
RAD52
YBR277C
YGL217C
BCK1
EAF3
PPH3
DEP1 YPR084W MET16
SNF1
DPB4
YMR074C
DPB3
YOR251C
DYN3
ELG1
YPR039W
FKH2
DYN1
IMP2'
YLR143W
MCM16
MMS1
SBP1
UBA4
NTO1
HGH1
BUB2
RAD27
PDE2
HSP104
PTC6
JJNM1
NCS2
IRA2 DBR1
JRRN10
JIRC21
3 MAK31
PHO13
YNL120C
GIM4
PPQ1
THR4 KTI12ARP1
DPH5
TIP41
SNF5
YPL205C
NPL4
FKH1
TEL1
CTF18 YPL102C
HMT1
EST3
MAK3PTC2
BEM2
PPT1HSP26
SOH1
LDB19
MMS22
IKI3 FPS1
ELP6
NUP133
TMA19RRD1 SWM1
CDC73 HPR5
YKU80
ELP4ELP2 YPR050C
RTF1
DPH1
KAR3
:::RTT103
PUF4
CPA2
YNL171C YKR074W
MFT1 STI1
THP2
EST1
RMD11
GIM3
MAK10
POL32
CTF4
LSM1
MRE11
DPH2
DCC1 CLB5
YKU70
MRC1
SRB2
XRS2
:::MRC1
TGS1
INP52
LSM6 RAD50 CTK1
SRB8
ELP3PAT1
LTE1
OPI1
RIF1
40
60
RPN4
20
BTS1
GSH1
RPL20B
SGF29
CSE2
RPS9B
BUD31
STE2
YMD8
IES5
TRP5
GRR1
SIN4 MFA1
0
Fitness (F) of orfΔ cdc13-1 double mutants at 27°C (doublings2 / day)
Joint model results
YDL041W
SFL1
NHP10 SIR4
ARC1
STE4
SSD1
ARP8
STE11
MED1
0
20
40
60
80
Fitness (F) of orfΔ ura3Δ double mutants at 27°C (doublings2 / day)
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Summary
References
Big data issues
Understanding conflict between model and data in big data
contexts — does more data demand more complex models?
Model simplifications and improvements to MCMC (block
updates and proposals, reparameterisations, GVS, etc.)
Basic parallelisation strategies (parallel chains, parallellelised
single chain)
Investigation of data parallel strategies (consensus Monte
Carlo, etc.)
Novel representations and implementations of Bayesian
hierarchical models using functional programming languages
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Summary
References
Summary
Modern bioscience is generating large, complex data sets
which require sophisticated modelling in order to answer
questions of scientific interest
Big data forces trade-offs between statistical accuracy and
computational tractability
Stochastic dynamic models are much more flexible than
deterministic models, but come at a computational cost —
the LNA can sometimes represent an excellent compromise
Notions of genetic interaction translate directly to statistical
models of interaction
Big hierarchical variable selection models are useful in
genomics, but can be computationally challenging
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Summary
References
Funding acknowledgements
Major funders of this work:
BBSRC
MRC
Wellcome Trust
Cancer Research UK
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data
Yeast
Data analysis
Summary and conclusions
Summary
References
References
Addinall, S. G., Holstein, E., Lawless, C., Yu, M., Chapman, K., Taschuk, M.,
Young, A., Ciesiolka, A., Lister, A., Wipat, A., Wilkinson, D. J., Lydall, D. A.
(2011) Quantitative fitness analysis shows that NMD proteins and many other
protein complexes suppress or enhance distinct telomere cap defects. PLoS
Genetics, 7:e1001362.
Heydari, J. J., Lawless, C., Lydall, D. A., Wilkinson, D. J. (2016) Bayesian
hierarchical modelling for inferring genetic interactions in yeast, Journal of the
Royal Statistical Society, Series C, early view.
Heydari, J. J., Lawless, C., Lydall, D. A., Wilkinson, D. J. (2014) Fast Bayesian
parameter estimation for stochastic logistic growth models, BioSystems,
122:55-72.
Lawless, C., Wilkinson, D. J., Addinall, S. G., Lydall, D. A. (2010) Colonyzer:
automated quantification of characteristics of microorganism colonies growing on
solid agar, BMC Bioinformatics, 11:287.
Wilkinson, D. J. (2009) Stochastic modelling for quantitative description of
heterogeneous biological systems, Nature Reviews Genetics. 10(2):122-133.
Wilkinson, D. J. (2011) Stochastic Modelling for Systems Biology, second
edition. Chapman & Hall/CRC Press.
Darren Wilkinson — UCL, 7/1/2016
Hierarchical modelling of automated imaging data