genetic and phenotypic variation in sockeye salmon
DESCRIPTION
My Master's defense recapping the SNP assessment and sockeye senescence projects I worked on during my tenure as a grad student at the University of Washington.TRANSCRIPT
1
Genetic and phenotypic diversity in sockeye salmon, Oncorhynchus nerka
Caroline Storer
University of WashingtonSchool of Aquatic and Fishery Sciences
Committee:Thomas Quinn
Steven Roberts (Co-chair)James Seeb (Co-chair)
William Templin
2
Outline
• Introduction – Sockeye salmon
• Chapter 1:– Evaluating the performance of SNPs for individual
assignment • Chapter 2: – Characterizing differences in gene expression
patterns associated with variability in senescence
3
4
5
6
7
8
Sockeye Salmon
• Anadromous• Natal homing• Undergo rapid
senescence• Semelparous
9
Motivations
• Improved fisheries management– Developing new management tools
10
11
Fisheries Management
• Applying genetics to fisheries management
12
Fisheries Management
• Applying genetics to fisheries management
- Population structure
- Inferring population history
- Parentage analysis
- Fisheries forensics
- Estimating mixed stock
composition
13
Fisheries Management
• Applying genetics to fisheries management
Habicht et al. 2010
Alaska
Russia
Bering Sea
14
Molecular Markers, Today
• Single nucleotide polymorphisms (SNPs)
15
Molecular Markers, Today
A C T C G
A C A C G
SNP locus
• Single nucleotide polymorphisms (SNPs)
16
Molecular Markers, Today
A C T C G
A C A C G
SNP locus
• Single nucleotide polymorphisms (SNPs)
- Abundant
- The number of available
markers is growing
- Methods are robust and
automated
- Not all SNPs are equal
17
Chapter 1: Objectives
• Develop new SNP markers for sockeye salmon
18
Chapter 1: Objectives
• Develop new SNP markers for sockeye salmon
• Rank all SNPs in sockeye salmon based on
performance
19
Chapter 1: Objectives
• Develop new SNP markers for sockeye salmon
• Rank all SNPs in sockeye salmon based on
performance
• Evaluate the success of different ranking
methods
20
Measuring Genetic Variation
Russia
Bristol Bay
Alaska Peninsula
South-central Alaska
British Columbia
Washington
Genotyped 12 populations, 61- 93 fish per population, using 114 SNPs
21
Measuring Genetic Variation
RussiaPrin
cipa
l Coo
rdin
ate
2 (1
5.5%
)
Bristol Bay Alaska Peninsula
South-central Alaska
British Columbia
Washington
Principal Coordinate 1 (44.5%)
22
Russia
Bristol Bay
Alaska Peninsula
South-central Alaska
British Columbia Washington
23
SNP Ranking
• Performed using only half of available individuals– Remaining individuals reserved for panel testing
24
SNP Ranking
• Performed using only half of available individuals– Remaining individuals reserved for panel testing
• Each SNP ranked by 5 measures
25
SNP Ranking
• FST
- SNPs ranked by ability to measure population variance
26
SNP Ranking
• FST
- SNPs ranked by ability to measure population variance• Informativness (In)
- Potential for a genotype to belong to specific population versus a population average
27
SNP Ranking
• FST
- SNPs ranked by ability to measure population variance• Informativness (In)
- Potential for a genotype to belong to specific population versus a population average
• Locus contribution (LC)- Average contribution of each SNP to principal components
28
SNP Ranking
• FST
- SNPs ranked by ability to measure population variance • Informativness (In)
- Potential for a genotype to belong to specific population versus a population average
• Locus contribution (LC)- Average contribution of each SNP to principal components
• BELS- SNPs ranked by reduction in performance when removed
29
SNP Ranking
• FST
- SNPs ranked by ability to measure population variance • Informativness (In)
- Potential for a genotype to belong to specific population versus a population average
• Locus contribution (LC)- Average contribution of each SNP to principal components
• BELS- SNPs ranked by reduction in performance when removed
• WHICHLOCI- Algorithm for ranking SNPs based on power for individual assignment
30
SNP Ranking
0 10 20 30 40 50 60 70 80 90 100 110
1
21
41
61
81
101
SNPs ordered by average rank
Ave
rage
SN
P ra
nk
top ranked SNPs
31
Panel Design
• Created 48- and 96-SNP panels containing top ranked SNPs– for each of the five ranking measures– for average SNP rank– for randomly selected SNPs
32
Panel Design
96-SNP Panels
33
Panel Design
96-SNP Panels
34
Panel Design
96-SNP Panels
35
Panel Design
96-SNP Panels
36
Panel Design
48-SNP Panels
37
Panel Design
48-SNP Panels
38
Panel Design
48-SNP Panels
39
Panel Design
48-SNP Panels
40
Panel Design
48-SNP Panels
41
Panel Testing
• 2 panel testing methods
42
Panel Testing
• 2 panel testing methods– Empirical • Remaining individuals assigned to a baseline of
individuals used for SNP ranking• Assignment tests performed in ONCOR
43
Panel Testing
• 2 panel testing methods– Empirical • Remaining individuals assigned to a baseline of
individuals used for SNP ranking• Assignment tests performed in ONCOR
– Simulated• 1000 individuals simulated using population allele
frequencies from remaining individuals• Assignment tests replicated 500 times
44
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
45
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
46
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
47
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
48
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
49
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
50
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
51
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
52
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
53
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
54
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
55
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
56
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
57
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
58
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
59
Panel Testing – Empirical data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.0
0.2
0.4
0.6
0.8
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
60
Panel Testing – Simulated data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.7
0.8
0.9
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
61
Panel Testing – Simulated data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.7
0.8
0.9
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
62
Panel Testing – Simulated data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.7
0.8
0.9
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
63
Panel Testing – Simulated data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.7
0.8
0.9
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
64
Panel Testing – Simulated data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.7
0.8
0.9
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
65
Panel Testing – Simulated data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.7
0.8
0.9
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
66
Panel Testing – Simulated data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.7
0.8
0.9
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
67
Panel Testing – Simulated data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.7
0.8
0.9
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
68
Panel Testing – Simulated data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.7
0.8
0.9
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
69
Panel Testing – Simulated data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.7
0.8
0.9
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
70
Panel Testing – Simulated data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.7
0.8
0.9
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
71
Panel Testing – Simulated data
96 Fst
96 In96 LC
96 BELS96 W
L
96 Random
96 AVG48 Fs
t48 In
48 LC
48 BELS48 W
L
48 Random
48 AVG0.7
0.8
0.9
1.0
Prob
abili
ty o
f cor
rect
ass
ignm
ent
72
Findings
• Greater variation and lower panel performance using empirical data
73
Findings
• Greater variation and lower panel performance using empirical data
• In general, 96-SNP panels performed better
74
Findings
• Greater variation and lower panel performance using empirical data
• In general, 96-SNP panels performed better• FST, In, and LC panels had the highest average
probability of correct assignment
75
Findings
• Greater variation and lower panel performance using empirical data
• In general, 96-SNP panels performed better• FST, In, and LC panels had the highest average
probability of correct assignment • Random SNP selection preforms nearly as well as
ranking when all available SNPs are used
76
Findings
• Greater variation and lower panel performance using empirical data
• In general, 96-SNP panels performed better• FST, In, and LC panels had the highest average
probability of correct assignment • Random SNP selection preforms nearly as well as
ranking when all available SNPs are used• BELS panels had the lowest average probability of
correct assignment
77
Conclusions
• Common ranking methods perform differently
78
Conclusions
• Common ranking methods perform differently • More SNPs is often better
79
Conclusions
• Common ranking methods perform differently • More SNPs is often better• When choosing a small proportion of available SNPs
the ranking approach is more important
80
Conclusions
• Common ranking methods perform differently • More SNPs is often better• When choosing a small proportion of available SNPs
the ranking approach is more important• Empirical panel tests performance on real (vs.
simulated) populations
81
Conclusions
• Common ranking methods perform differently • More SNPs is often better• When choosing a small proportion of available SNPs
the ranking approach is more important• Empirical panel tests performance on real (vs.
simulated) populations• Simulated data highlights performance based on
SNP composition
82
Implications
• 43 new SNPs are now available for sockeye salmon
83
Implications
• 43 new SNPs are now available for sockeye salmon- Already in use
83Year
Cat
ch (
mill
ions
of
sock
eye
salm
on)
1960 1970 1980 1990 2000 2010
010
2030
40 Stock
TogiakIgushikWoodNushagakKvichakAlagnakNaknekEgegikUgashik
84
Implications
• 43 new SNPs are now available for sockeye salmon- Already in use
85
Implications
• 43 new SNPs are now available for sockeye salmon- Already in use
• Methods outlined are important for developing SNP panels for any system or question
86
Motivations
• Improved fisheries management– Developing new management tools
87
Motivations
• Improved fisheries management– Developing new management tools
• Understanding salmon mortality– Characterizing variability in senescence
88
89
Salmon Senescence
• Undergo rapid senescence
90
91
Salmon Senescence
• Undergo rapid senescence
92
Salmon Senescence
• Undergo rapid senescence• Rates of senescence vary:– in the same populations (Perrin & Irvine 1990)– between populations (Carlson et al. 2007)
• Characterized by physiological trade-offs
93
Salmon Senescence
• Characterized by physiological trade-offs
• Increased energetic investment in reproduction
• Starvation and stress
Finch 1994; Gotz et al. 2005; Maldonado et al. 2002
94
Salmon Senescence
• Characterized by physiological trade-offs
• Increased energetic investment in reproduction
• Starvation and stress
• Decreased immune function• Increased oxidative stress• Central nervous system disintegration
Finch 1994; Gotz et al. 2005; Maldonado et al. 2002
95
Objectives
• Uncover driving mechanisms of senescence– Develop quantitative gene expression assays for
genes associated with aging – Characterize senescent specific expression
patterns in sockeye salmon
96
Assay Design
• Selected genes based on physiological responses of interest
97
Assay Design
• Selected genes based on physiological responses of interest
• Developed 5 successful assays
Gene Acesion # Response Amplicon sizeViperin (vig1) NM_001124253.1 immune 244NMDA-type glutamate receptor 1 subunit AB292234.1 memory 239olfactory marker protein 1 AB490250.1 olfactory 169telomerase reverse transcriptase (TERT) CX246542 aging 151GnRH Precursor D31868 reproduction 226
98
Measuring Gene Expression
• 25 sockeye salmon– 11 pre-senescent– 14 senescent
• Expression measured in brain tissue
99
Measuring Gene Expression
100
NMDA
• Involved in synaptic plasticity and memory
• Linked to neurodegenerative disorders
101
NMDA
• Involved in synaptic plasticity and memory
• Linked to neurodegenerative disorders
• No significant difference
50
40
30
20
10
0Pre-
senescentSenescent
Gen
e ex
pres
sion
P = 0.12
102
OMP1
• Olfactory marker proteins (OMP) necessary for the function of olfactory receptor neurons
103
OMP1
• Olfactory marker proteins (OMP) necessary for the function of olfactory receptor neurons
• No significant difference
150
100
50
0
Pre-senescent
Senescent
Gen
e ex
pres
sion
P = 0.32
104
GnRHp
• Part of the GnRH axis which plays a critical role in reproduction
105
GnRHp
• Part of the GnRH axis which plays a critical role in reproduction
• No significant difference
1000
600
400
0
Pre-senescent
Senescent
Gen
e ex
pres
sion
P = 0.15
800
200
106
Viperin
• Anti-viral protein involved in the innate immune response
107
10000
30000
25000
20000
15000
5000
0Pre-
senescentSenescent
Gen
e ex
pres
sion
P = 0.017
Viperin
• Anti-viral protein involved in the innate immune response
• Significant difference
108
Viperin
• Anti-viral protein involved in the innate immune response
• Significant difference• Immune response
attempted in senescent salmon
10000
30000
25000
20000
15000
5000
0Pre-
senescentSenescent
Gen
e ex
pres
sion
P = 0.017
109
TERT
• Catalytic subunit of the enzyme telomerase
• Responsible for telomere repair and extension
110
TERT
• Catalytic subunit of the enzyme telomerase
• Responsible for telomere repair and extension
• Significant difference
80
40
60
20
0Pre-
senescentSenescent
Gen
e ex
pres
sion
P = 0.03
111
TERT
• Catalytic subunit of the enzyme telomerase
• Responsible for telomere repair and extension
• Significant difference• Maintaining telomere length
critical to survival till spawning
80
40
60
20
0Pre-
senescentSenescent
Gen
e ex
pres
sion
P = 0.03
112
Gene Expression
-2 -1 0 1 2 3 4 5 6 7 8-3
-2
-1
0
1
2
3
4Pre-senescentSenescent
Principal component 1 (61.06 %)
Prin
cipa
l com
pone
nt 2
(19.
35 %
)
113
Gene Expression
-2 -1 0 1 2 3 4 5 6 7 8-3
-2
-1
0
1
2
3
4Pre-senescentSenescent
Principal component 1 (61.06 %)
Prin
cipa
l com
pone
nt 2
(19.
35 %
)
114
Gene Expression
-2 -1 0 1 2 3 4 5 6 7 8-3
-2
-1
0
1
2
3
4
1) Viperin
GnRHp
OMP1
NMDA
2) TERT
Pre-senescentSenescent
Principal component 1 (61.06 %)
Prin
cipa
l com
pone
nt 2
(19.
35 %
)
1 2
115
Findings
• Greater expression in senescent salmon
116
Findings
• Greater expression in senescent salmon• Greater variation in expression of senescent
salmon
117
Findings
• Greater expression in senescent salmon• Greater variation in expression of senescent
salmon
118
Findings
• Greater expression in senescent salmon• Greater variation in expression of senescent
salmon• Significant differences detected for two genes:
TERT (aging) and Viperin (immune function)
119
Conclusions
• Strong response detected in immune function– Driving mechanism or associated process?
120
Conclusions
• Strong response detected in immune function– Driving mechanism or associated process?
• Telomerase activity represents senescence specific signal
121
Implications
• New assays can be used at any stage of the sockeye salmon life cycle
122
Implications
• New assays can be used at any stage of the sockeye salmon life cycle
• Telomere dynamics important for understanding variation in rates of senescence
123
Telomere Dynamics
Population 1 Population 2
124
Telomere Dynamics
Population 1 Population 2
• Fast senescence • Slow senescence
125
Telomere Dynamics
Population 1 Population 2
• Fast senescence• Low telomerase
expression
• Slow senescence• High telomerase
expression
126
Implications
• New assays can be used at any stage of the sockeye salmon life cycle
• Telomere dynamics important for understanding variation in rates of senescence
127
Implications
• New assays can be used at any stage of the sockeye salmon life cycle
• Telomere dynamics important for understanding variation in rates of senescence– Measure of life history diversity (rate of
senescence)
128
Motivations
• Improved fisheries management– Developing new management tools
• Understanding salmon mortality– Characterizing variability in senescence
129
Acknowledgments Roberts Lab:• Sam White• Steven Roberts• Emma Timmins-Schiffman• Dave Metzger• Mackenzie Gavery
Seeb Lab:• Jim Seeb• Lisa Seeb• Carita Pascal• Eleni Petrou• Meredith Everett• Wes Larson• Marissa Jones• Sewall Young• Ryan Waples
Funding:• Alaska Sustainable Salmon Fund• Bristol Bay Regional Seafood
Development Group• The Gordon and Betty Moore
Foundation• The School of Aquatic and Fishery
Sciences• OACIS NSF GK12
Committee:• Thomas Quinn• Steven Roberts (Co-chair)• James Seeb (Co-chair)• William Templin
FRIENDS and FAMILY Cohort ‘09
130THANK YOU!
131
132
133
134