health is a product of genome & exposomebetterbones.org/wp...personalized-genomics-ncibh.pdf ·...
TRANSCRIPT
Using Omics and Big Data to Manage Health and
Disease
Michael Snyder
Stanford University
May 19, 2017
Conflicts: Personalis, Genapsys, SensOmics, QbioFood
Health Disease
Genome
PathogensStress
Exercise
Health Is a Product of Genome & Exposome
Drivers of Big Data
Human Genome Cost <$2K
126.90 127.00 127.10 127.20m/z
0
10
20
30
40
Rela
tive A
bu
nd
an
ce
127.0613
1) DNA Sequencing
2) Mass Spectrometry 3) Wearables
Genome
Transcriptome
Proteome
Metabolome
Lipidomics
Autoantibody-ome
Personal Omics Profiling
Cytokines
Epigenome
Billions of
Measurements!
Microbiome (Gut, Urine, Nasal,
Tongue, Skin)
Medical Tests, Questionaires
Omics
Measurements
4
Year 2 …Year 1 Viral infection
Biosensors
1) Understand how individuals change over time and
during periods of health and disease at high resolution
1) Understand how different “omes” (microbiome,
metabolome, proteome, genome) relate to one another
dynamically
1) Understand how individual responses are similar and
differ from one another when faced with specific
perturbations
2) Identify factors that can affect and help manage the
health of an individual
General Goals
5
Year 2 …Year 1Viral infection
912
Adenovirus Infection
694679 683 688 700680
711 735
796 840
Adenovirus Infection
944 948 984945 959 966
HRV Infection
1030 10381029 1032 1045 1051 1060
400186185
255
116
369
380
329322
Day from 1st HRV
Infection (D)
RSV Infection
297 301289 292 294 307 311290
HRV Infection
4 210
476 546532
HRV Infection
625615 618 620 630616
602 647-123
Day from 1st HRV
Infection (D)
14151453
1487
15161526
HRV Infection
1714
1720 17431716 1723 1729
Infection
19081906
110911241164 12001227128413161319 132313311338136013811536
1564
1589
1612
1628
1631
1643
1680
1695
1702
1709
17
50
17
57
17
64
17
71
17
74
17
79
17
92
18
15
18
28
18
35
18
42
18
49
18
52
18
59
18
71
18
94
Personal Omics Profile
84 months; >250 Timepoints; 10 Viral Infections
Chen et al., Cell 2012, unpublished
Skin Rash
Genome Sequence (Ilumina, Complete Genomics)
Predict Type 2
Diabetes
Rong Chen
and Atul Butte
0% 100%
+)"# "# )"# %""# %)"# 200 250 300 350 400 450 500 550 600 650+%)"# +%""# +)"# 0 )"# %""# %)"# &""# &)"# '""# ')"# (""# ()"# )""#
HRV Infection (Day 0-21)
RSV Infection (Day 289-311)
Life Style Change (Day 380-Current)
Glu
co
se
(m
g/d
L)
Day Number (Relative to 1st Infection)
80
90
100
110
120
130
140
150
160
-150
Glycated HgA1c (%):
(Day Number)
6.4 (329)
6.7 (369)
4.9 (476)
5.4 (532)
5.3 (546)
4.7 (602)HbA1c (%) 6.4 6.7 4.9 5.4 5.3 4.7
(Day Number) (329) (369) (476) (532) (546) (602)
RSVHRVLIFESTYLE
CHANGE
Glucose levels
*
*
**
*Previously known
HRV
Exercise
RSVSkinRash/Itch
HRV
Adenovirus
HRVHRV
Changed life style
}
Norm
al R
ange 3
.8-5
.7%
{
Norm
al R
ange 7
0-9
9 m
g/d
L
Extended Time Line
Molecules and Biochemical Pathways that Change
During Acquisition of Diabetes
george miasRSV 18 days
Platelet Plug
Formation
Glucose Regulation
of Insulin Secretion
Insulin
Biosynthetic
Pathway• Affected by nutrition, lifestyle factors,
aging, and environment
• Causes gene silencing
Map all the methylated sites using
whole genome bisulfite sequencing
Epigenetics: DNA Methylation
5 methylC
Father
Mother
T CInactivated by
mutation
Inactivated by
DNA
MethylationM MM
Methylated
CpGs
Few RNAs
Lots of RNAPDE4 DIP Gene
Gene Inactivation by Mutation and
Methylation: PDE4 involved in eosinophilia
Kim Kukurba
Longitudinal Profiling of 100 individuals (Prediabetics &
Healthy) over periods of health, stress and disease
Year 2 …Year 1
Viral infection
Stress
Diet change
Cell Host & Microbiome 2014
100>1500 collections thus far
Most Datasets are Open Access!
Stu
dy p
art
icip
an
t
Genome Sequencing – First 60 People
• Eleven have important pathogenic mutations:• SHBD (2X): high freq. of neuroendocrine tumors• PROC: Affects coagulation• HNF1A: MODY mutation• ABCC8: Hyperinsulinemic hypoglycemia• MUTYH: Colon cancer• SLC7A9: Cystinuria• RBM20: Dilated cardiomyopathy• CHEK2: Breast cancer• APC: Colon cancer• BRCA1: Breast & ovarian cancer
• All have reportable carrier mutations and/or pharmacogenetic variants
Personalis, IncShannon Rego et al.
A subset of individuals undergo a dietary
perturbation.
30
days
60
days7
days
23 participants:
• 13 Insulin resistant
• 10 healthy controls (BMI matched)
Brian Piening, Wenyu Zhou, Gucci Gu, Kevin Contrepois
Metabolic differences between IR and IS
Univariate analysis: Wilcoxon ttest pvalue < 0.05 and fold change > 1.5
Me
tab
olite
s
Participants at T1
Example data: Short-term weight gain
bio
mo
lecu
le
s
Integrative c-means clustering: pattern
recognition across RNA-seq, proteome,
metabolome, microbiome, cytokines
PATTERN1: UP AT PEAK
WEIGHT THEN DOWN
KEGG: HYPERTROPHIC
CARDIOMYOPATHY (q<0.001)
Normalized log2 plasma concentration
Blood cytokine profiles: 20 subjects at baseline
Insulin Sensitive
Insulin Resistant
Microbial abundance
pattern group by
individual, not by dietary
supplement
-- Distance matrix by
Manhattan methods and
Hierarchical clustering by Ward
method
Basis B1
Basis Peak
Apple Watch
Dexcom Constant Glucose Monitor
Withthings Smart Scale
Qardio Blood Pressure Cuff
Scanadu Scout
Autographer – Life Logger
iHealth Pulse Oximeter
Athos – Smart Shorts
RadtargeRadiation
Sensors: Measure Many
ThingsOverview of the project
Li, Dunn et al.
PloS Biol 2017
Circadian and diurnal patterns
Li, Dunn et al.
PloS Biol 2017
Δ S
kin
Tem
p (
Nig
ht –
Day
)
Δ H
R
(Nig
ht –
Day
)
43 People
Activity Phenotypes: 4 Patterns
h
Accele
rom
ete
r M
ag
nit
ud
e
Hour of Day
Li, Dunn et al.
PloS Biol 2017
SpO2
Levels
Drops During
Airline Flights
Finger device
91-96%: 65.4%
90 or less: 5.1%
SpO2 levels drops on airplane flights:
18 Subjects
Li, Dunn et al.
PloS Biol 2017
SpO2
Measurements
Associated
With Fatigue
Li, Dunn et al.
PloS Biol 2017
SpO2
Measurements
Can Adapt on
Flights >7hrs
SpO2 Measurements: Unusually
Low on One Flight to Norway
Li, Dunn et al.
PloS Biol 2017
Elevated Heart Rate and Skin
Temperature During Lyme DiseaseLyme Disease
Early detection of Lyme disease
% o
f o
utlie
r (H
R)
% o
f o
utlie
r (S
kin
te
mp
.)
458 470 474462 466 478 482 486 490 494Day
Detects All Days of IllnessChange-of-Heart
Algorithm
Detects Periods of Illness
at High Resolution using
Wearable Device
HR differences in IR and Insulin Sensitivity Smart Phone = Control Center
Share Information With
Physician
The Future?
Genomic Sequencing
1. Predict risk
2. Early Diagnose
3. Monitor
4. Treat
GGTTCCAAAAGTTTATTGGATGCCGT
TTCAGTACATTTATCGTTTGCTTTGG
ATGCCCTAATTAAAAGTGACCCTTTC
AAACTGAAATTCATGATACACCAATG
GATATCCTTAGTCGATAAAATTTGCG
AGTACTTTCAAAGCCAAATGAAATTA
TCTATGGTAGACAAAACATTGACCAA
TTTCATATCGATCCTCCTGAATTTAT
TGGCGTTAGACACAGTTGGTATATTT
A….
Amanda Mills
Omes & Sensors: Personal Devices
Overall Summary
1) Personal genome sequencing is here. It can be used to predict disease risk and manage health
2) Multi-omics analyses are valuable for determining pathways and biochemical activities involved in human disease.
3) Longitudinal profiles are very valuable for understanding personal disease states
4) Everyone’s profile is different
5) Wearables will be useful for managing health
6) Individuals will be responsible for their own health
Acknowledgements
40
Snyder Lab
Wenyu Zhou
Brian Piening
Kevin Contrepois
Tejaswini Mishra
Kim Kukurba
Shannon Rego
Jessica Sibal
Hannes Rost
Varsha Rao
Liang Liang
Tejas Mishra
Christine Yeh
Hassan Chaib
Eric Wei
Wearables
Xiao Li
Jessie Dunn
Sophia Miryam …
Denis Salins
Heather Hall
Weinstock Lab
George Weinstock
Erica Sodergren
Shana Leopold
Daniel Spakowicz
Blake Hanson
Eddy Bautista
Lauren Petersen
Lei Chen
Benjamin Leopold
Sai Lek
Purva Vats
Jon Bernstein
NIH
Lita Proctor
Salvatore Sechi
Jon LoTempio
And otherMcLaughlin Lab
Tracy McLaughlin
Colleen Craig
Candice Allister
Dalia Perelman
Elizabeth Colbert
Genomics and
Personalized Medicine
What Everyone Needs to
Know®
Michael Snyder
Available from Amazon