The document discusses the need for informatics methods, databases, and standards to support exposome-driven discovery research in a similar way that informatics has supported genomic research. Specifically, it notes that estimates of heritability from twin studies indicate that environmental factors likely play an equally important role as genetics in many traits/diseases. However, the chemical space of the exposome is large and heterogeneous, posing challenges to integrate exposome, genome, and phenome data through approaches like exposome-wide association studies.

1. Informatics and data analytics to support
exposome-based discovery
Perspectives from a NIEHS workshop
Chirag J Patel
International Society of Exposure Science
Henderson, NV (by way of Boston, MA)
10/20/15
chirag@hms.harvard.edu
@chiragjp
www.chiragjpgroup.org

2. Arjun Manrai (Harvard)*
Yuxia Cui (NIEHS)
Pierre Bushel (NIEHS)
Molly Hall (Penn State, now U Penn)*
Spyros Karakitsios(Aristotle U, Greece)
Carolyn Mattingly (NCSU)
Marylyn Ritchie (Geisinger Health/Penn State)
Charles Schmitt (NIEHS)
Denis Sarigiannis (Aristotle U, Greece)
Duncan Thomas (USC)
David Wishart (U Alberta, Canada)
David Balshaw (NIEHS)
The workgroup discussed informatics capability for
high-throughput exposome research
(late 2014 to early 2015)

3. We are now in the era of high-throughput
biology and biomedicine.
(now possible to assay thousands to millions of datapoints today)

4. We are now in the era of high-throughput
biology and biomedicine: examples of genomic advances
genetic arrays
gene expression
common genetic variants
epigenome (methylation)
whole genome sequencing (WGS)
full genome sequencing
mRNA-seq
epigenome (3D, histone)
3 x 109 nucleotidebases
3-4 x 104 genes
106 to 107 variants

5. Informatics has enabled discovery in genomics investigations.
1. infrastructure/standards,
2. analytics,
3. databases

6. Information infrastructure has enabled discovery in genomics
(example: UCSC genome browser)

7. Analytic methods have enabled discovery in genomics
(example: genome-wide association [GWAS])
A search engine for genetic influence in phenotypes
Genome-wide association studies (GWASs)
A RT I C L E S
13 autosomal loci exceeded the threshold for genome-wide significance (r2 < 0.05), and conditional analyses (see below) establish these SNPs
50 Locus established previously
Locus identified by current study
Locus not confirmed by current study
BCL11A
THADA
NOTCH2
ADAMTS9
IRS1
IGF2BP2
WFS1
ZBED3
CDKAL1
HHEX/IDE
KCNQ1 (2 signals*: )
TCF7L2
KCNJ11
CENTD2
MTNR1B
HMGA2 ZFAND6
PRC1
FTO
HNF1B DUSP9
Conditional analysis
Unconditional analysis
TSPAN8/LGR5
HNF1A
CDC123/CAMK1D
CHCHD9
CDKN2A/2B
SLC30A8
TP53INP1
JAZF1
KLF14
PPAR
40
30
–log10(P)–log10(P)
20
10
10
1 2 3 4 5 6 7 8
Chromosome
9 10 11 12 13 14 15 16 17 18 19 20 21 22 X
0
0
Suggestive statistical association (P < 1 10
–5
)
Association in identified or established region (P < 1 10
–4
)
Figure 1 Genome-wide Manhattan plots for the DIAGRAM+ stage 1 meta-analysis. Top panel summarizes the results of the unconditional meta-
analysis. Previously established loci are denoted in red and loci identified by the current study are denoted in green. The ten signals in blue are those
taken forward but not confirmed in stage 2 analyses. The genes used to name signals have been chosen on the basis of proximity to the index SNP and
should not be presumed to indicate causality. The lower panel summarizes the results of equivalent meta-analysis after conditioning on 30 previously
established and newly identified autosomal T2D-associated SNPs (denoted by the dotted lines below these loci in the upper panel). Newly discovered
conditional signals (outside established loci) are denoted with an orange dot if they show suggestive levels of significance (P < 10−5), whereas
secondary signals close to already confirmed T2D loci are shown in purple (P < 10−4).
Voight et al, Nature Genetics 2012
N=8K T2D, 39K Controls
GWAS in Type 2 Diabetes

8. 758,000 individuals
>400 studies
>>1B datapoints (genotypes and phenotypes)
>950 manuscripts (Paltoo et al., Nature Genetics 2014)
Accessible data repositories have enabled discovery in genomics
investigation:
(ex: Databases of Genotypes and Phenotypes)

9. We claim that there is need for informatics analytic methods,
databases, and standards for the exposome-driven discovery.
EWAS akin to GWAS?

10. Why?
courtesy: colabria.com

11. P = G + E

12. σ2
P = σ2
G + σ2
E

13. σ2
G
σ2
P
H2 =
Heritability (H2) is the range of phenotypic variability
attributed to genetic variability in a population

14. Eye color
Hair curliness
Type-1 diabetes
Height
Schizophrenia
Epilepsy
Graves' disease
Celiac disease
Polycystic ovary syndrome
Attention deficit hyperactivity disorder
Bipolar disorder
Obesity
Alzheimer's disease
Anorexia nervosa
Psoriasis
Bone mineral density
Menarche, age at
Nicotine dependence
Sexual orientation
Alcoholism
Lupus
Rheumatoid arthritis
Crohn's disease
Migraine
Thyroid cancer
Autism
Blood pressure, diastolic
Body mass index
Depression
Coronary artery disease
Insomnia
Menopause, age at
Heart disease
Prostate cancer
QT interval
Breast cancer
Ovarian cancer
Hangover
Stroke
Asthma
Blood pressure, systolic
Hypertension
Osteoarthritis
Parkinson's disease
Longevity
Type-2 diabetes
Gallstone disease
Testicular cancer
Cervical cancer
Sciatica
Bladder cancer
Colon cancer
Lung cancer
Leukemia
Stomach cancer
0 25 50 75 100
Heritability: Var(G)/Var(Phenotype) Source: SNPedia.com
H2 estimates for complex traits are low and variable:
massive opportunity for high-throughput E research
Type 2 Diabetes (25%)
Heart Disease (25-30%)
Autism (50%???)
Gaugler et al, Nature Genetics (2014)

15. Eye color
Hair curliness
Type-1 diabetes
Height
Schizophrenia
Epilepsy
Graves' disease
Celiac disease
Polycystic ovary syndrome
Attention deficit hyperactivity disorder
Bipolar disorder
Obesity
Alzheimer's disease
Anorexia nervosa
Psoriasis
Bone mineral density
Menarche, age at
Nicotine dependence
Sexual orientation
Alcoholism
Lupus
Rheumatoid arthritis
Crohn's disease
Migraine
Thyroid cancer
Autism
Blood pressure, diastolic
Body mass index
Depression
Coronary artery disease
Insomnia
Menopause, age at
Heart disease
Prostate cancer
QT interval
Breast cancer
Ovarian cancer
Hangover
Stroke
Asthma
Blood pressure, systolic
Hypertension
Osteoarthritis
Parkinson's disease
Longevity
Type-2 diabetes
Gallstone disease
Testicular cancer
Cervical cancer
Sciatica
Bladder cancer
Colon cancer
Lung cancer
Leukemia
Stomach cancer
0 25 50 75 100
Heritability: Var(G)/Var(Phenotype) Source: SNPedia.com
H2 estimates for complex traits are low and variable:
massive opportunity for high-throughput E research
H2 < 50%

16. ©2015NatureAmerica,Inc.Allrightsreserved.
Despite a century of research on complex traits in humans, the
relative importance and specific nature of the influences of
genes and environment on human traits remain controversial.
We report a meta-analysis of twin correlations and reported
variance components for 17,804 traits from 2,748 publications
including 14,558,903 partly dependent twin pairs, virtually
all published twin studies of complex traits. Estimates of
heritability cluster strongly within functional domains,
and across all traits the reported heritability is 49%. For a
majority (69%) of traits, the observed twin correlations are
consistent with a simple and parsimonious model where twin
resemblance is solely due to additive genetic variation. The
data are inconsistent with substantial influences from shared
environment or non-additive genetic variation. This study
provides the most comprehensive analysis of the causes of
individual differences in human traits thus far and will guide
future gene-mapping efforts. All the results can be visualized
using the MaTCH webtool.
Specifically, the partitioning of observed variability into underlying
genetic and environmental sources and the relative importance of
additive and non-additive genetic variation are continually debated1–5.
Recent results from large-scale genome-wide association studies
(GWAS) show that many genetic variants contribute to the variation
in complex traits and that effect sizes are typically small6,7. However,
the sum of the variance explained by the detected variants is much
smaller than the reported heritability of the trait4,6–10. This ‘missing
heritability’ has led some investigators to conclude that non-additive
variation must be important4,11. Although the presence of gene-gene
interaction has been demonstrated empirically5,12–17, little is known
about its relative contribution to observed variation18.
In this study, our aim is twofold. First, we analyze empirical esti-
mates of the relative contributions of genes and environment for
virtually all human traits investigated in the past 50 years. Second, we
assess empirical evidence for the presence and relative importance of
non-additive genetic influences on all human traits studied. We rely
on classical twin studies, as the twin design has been used widely
to disentangle the relative contributions of genes and environment,
across a variety of human traits. The classical twin design is based
on contrasting the trait resemblance of monozygotic and dizygotic
twin pairs. Monozygotic twins are genetically identical, and dizygotic
twins are genetically full siblings. We show that, for a majority of traits
(69%), the observed statistics are consistent with a simple and parsi-
monious model where the observed variation is solely due to additive
genetic variation. The data are inconsistent with a substantial influence
from shared environment or non-additive genetic variation. We also
show that estimates of heritability cluster strongly within functional
domains, and across all traits the reported heritability is 49%. Our
results are based on a meta-analysis of twin correlations and reported
variance components for 17,804 traits from 2,748 publications includ-
ing 14,558,903 partly dependent twin pairs, virtually all twin studies of
complex traits published between 1958 and 2012. This study provides
the most comprehensive analysis of the causes of individual differences
in human traits thus far and will guide future gene-mapping efforts. All
Meta-analysis of the heritability of human traits based on
fifty years of twin studies
Tinca J C Polderman1,10, Beben Benyamin2,10, Christiaan A de Leeuw1,3, Patrick F Sullivan4–6,
Arjen van Bochoven7, Peter M Visscher2,8,11 & Danielle Posthuma1,9,11
1Department of Complex Trait Genetics, VU University, Center for Neurogenomics
and Cognitive Research, Amsterdam, the Netherlands. 2Queensland Brain
Institute, University of Queensland, Brisbane, Queensland, Australia. 3Institute
for Computing and Information Sciences, Radboud University Nijmegen,
Nijmegen, the Netherlands. 4Center for Psychiatric Genomics, Department
of Genetics, University of North Carolina, Chapel Hill, North Carolina, USA.
5Department of Psychiatry, University of North Carolina, Chapel Hill, North
Carolina, USA. 6Department of Medical Epidemiology and Biostatistics,
Karolinska Institutet, Stockholm, Sweden. 7Faculty of Sciences, VU University,
Insight into the nature of observed variation in human traits is impor-
tant in medicine, psychology, social sciences and evolutionary biology.
It has gained new relevance with both the ability to map genes for
human traits and the availability of large, collaborative data sets to do
so on an extensive and comprehensive scale. Individual differences in
human traits have been studied for more than a century, yet the causes
of variation in human traits remain uncertain and controversial.
Nature Genetics, 2015
17,804 traits of the phenome
2,748 publications
14,558,903 twin pairs
Average H2 (genome): 0.49
Exposome plays an equal role.

17. What is the potential chemical (external and internal) space of the
exposome?: perhaps on the order of thousands.
>84,000
TSCA and EPA Inventory
(2014)
>13,000
Davis et al
Comparative Tox DB (2015)
3,600 + 1,634
Toxic Exposome Database
Wishart et al (2015)
toxicants drugs
100-1,000?
uBiome

18. What will the exposome data structure look like?:
a high-dimensioned 3D matrix of (1) exposure measurements
on (2) individuals as a function of (3) time
tim
e
exposome
pollutants
diet
m
etabolites . . .
gut flora
CVD
xenobiotics
individuals
GWAS, RVAS,
pathway
analysis..etc.
EWAS,
PheWAS..etc.
genome(static)
mixtures of
exposures
drugs
integrative
(A) (C)
(B)
exposome
factors
nutrient value for
individual i
individual i

19. What will the exposome data structure look like?:
a high-dimensioned 3D matrix of (1) exposure measurements
on (2) individuals as a function of (3) time
tim
e
exposome
pollutants
diet
m
etabolites . . .
gut flora
CVD BP
can
xenobiotics
individuals
GWAS, RVAS,
pathway
analysis..etc.
EWAS,
PheWAS..etc.
genome(static)
mixtures of
exposures
drugs
integrative
(A) (C)
(B)
longitudinal
system
genome

20. Data-driven investigation for novel exposome factors in the phenome:
Exposome-wide, phenome-wide, and genome-exposome-wide discovery
tim
e
exposome phenome
pollutants
diet
m
etabolites . . .
gut flora
height
w
eight
CVD BP
T2D
cancer
xenobiotics . . .
individuals
GWAS, RVAS,
pathway
analysis..etc.
EWAS,
PheWAS..etc.
genome(static)
mixtures of
exposures
tim
e
drugs
integrative
mixtures of
phenotypes
(A) (C)
(B)
Informatics methods to integrate heterogeneous data (E, G, and P)
and to conduct EWAS, GxEWAS, and PheWAS
EWAS
PheWAS

21. Integration challenges in conducting
data-driven investigation for novel exposome factors in the phenome:
The exposome is heterogenous and G does not equal E.
platform
scale
time-dependent
type
correlation
mass-spec: targeted vs. untargeted
external vs. internal
sampling and life trajectories
continuous vs. categorical
dense!

22. Interdependencies of the exposome:
Correlation globes paint a dense and complex view of exposure
JAMA 2015
Pac Symp Biocomput. 2015

23. σ2
P = σ2
G + σ2
E
σ2
E ???

24. Alpha-carotene
Alcohol
VitaminEasalpha-tocopherol
Beta-carotene
Caffeine
Calcium
Carbohydrate
Cholesterol
Copper
Beta-cryptoxanthin
Folicacid
Folate,DFE
Foodfolate
Dietaryfiber
Iron
Energy
Lycopene
Lutein+zeaxanthin
MFA16:1
MFA18:1
MFA20:1
Magnesium
Totalmonounsaturatedfattyacids
Moisture
Niacin
PFA18:2
PFA18:3
PFA20:4
PFA22:5
PFA22:6
Totalpolyunsaturatedfattyacids
Phosphorus
Potassium
Protein
Retinol
SFA4:0
SFA6:0
SFA8:0
SFA10:0
SFA12:0
SFA14:0
SFA16:0
SFA18:0
Selenium
Totalsaturatedfattyacids
Totalsugars
Totalfat
Theobromine
VitaminA,RAE
Thiamin
VitaminB12
Riboflavin
VitaminB6
VitaminC
VitaminK
Zinc
NoSalt
OrdinarySalt
a-Carotene
VitaminB12,serum
trans-b-carotene
cis-b-carotene
b-cryptoxanthin
Folate,serum
g-tocopherol
Iron,FrozenSerum
CombinedLutein/zeaxanthin
trans-lycopene
Folate,RBC
Retinylpalmitate
Retinylstearate
Retinol
VitaminD
a-Tocopherol
Daidzein
o-Desmethylangolensin
Equol
Enterodiol
Enterolactone
Genistein
EstimatedVO2max
PhysicalActivity
Doesanyonesmokeinhome?
Total#ofcigarettessmokedinhome
Cotinine
CurrentCigaretteSmoker?
Agelastsmokedcigarettesregularly
#cigarettessmokedperdaywhenquit
#cigarettessmokedperdaynow
#dayssmokedcigsduringpast30days
Avg#cigarettes/dayduringpast30days
Smokedatleast100cigarettesinlife
Doyounowsmokecigarettes...
numberofdayssincequit
Usedsnuffatleast20timesinlife
drink5inaday
drinkperday
days5drinksinyear
daysdrinkinyear
3-fluorene
2-fluorene
3-phenanthrene
1-phenanthrene
2-phenanthrene
1-pyrene
3-benzo[c]phenanthrene
3-benz[a]anthracene
Mono-n-butylphthalate
Mono-phthalate
Mono-cyclohexylphthalate
Mono-ethylphthalate
Mono-phthalate
Mono--hexylphthalate
Mono-isobutylphthalate
Mono-n-methylphthalate
Mono-phthalate
Mono-benzylphthalate
Cadmium
Lead
Mercury,total
Barium,urine
Cadmium,urine
Cobalt,urine
Cesium,urine
Mercury,urine
Iodine,urine
Molybdenum,urine
Lead,urine
Platinum,urine
Antimony,urine
Thallium,urine
Tungsten,urine
Uranium,urine
BloodBenzene
BloodEthylbenzene
Bloodo-Xylene
BloodStyrene
BloodTrichloroethene
BloodToluene
Bloodm-/p-Xylene
1,2,3,7,8-pncdd
1,2,3,7,8,9-hxcdd
1,2,3,4,6,7,8-hpcdd
1,2,3,4,6,7,8,9-ocdd
2,3,7,8-tcdd
Beta-hexachlorocyclohexane
Gamma-hexachlorocyclohexane
Hexachlorobenzene
HeptachlorEpoxide
Mirex
Oxychlordane
p,p-DDE
Trans-nonachlor
2,5-dichlorophenolresult
2,4,6-trichlorophenolresult
Pentachlorophenol
Dimethylphosphate
Diethylphosphate
Dimethylthiophosphate
PCB66
PCB74
PCB99
PCB105
PCB118
PCB138&158
PCB146
PCB153
PCB156
PCB157
PCB167
PCB170
PCB172
PCB177
PCB178
PCB180
PCB183
PCB187
3,3,4,4,5,5-hxcb
3,3,4,4,5-pncb
3,4,4,5-tcb
Perfluoroheptanoicacid
Perfluorohexanesulfonicacid
Perfluorononanoicacid
Perfluorooctanoicacid
Perfluorooctanesulfonicacid
Perfluorooctanesulfonamide
2,3,7,8-tcdf
1,2,3,7,8-pncdf
2,3,4,7,8-pncdf
1,2,3,4,7,8-hxcdf
1,2,3,6,7,8-hxcdf
1,2,3,7,8,9-hxcdf
2,3,4,6,7,8-hxcdf
1,2,3,4,6,7,8-hpcdf
Measles
Toxoplasma
HepatitisAAntibody
HepatitisBcoreantibody
HepatitisBSurfaceAntibody
HerpesII
Albumin, urine
Uric acid
Phosphorus
Osmolality
Sodium
Potassium
Creatinine
Chloride
Total calcium
Bicarbonate
Blood urea nitrogen
Total protein
Total bilirubin
Lactate dehydrogenase LDH
Gamma glutamyl transferase
Globulin
Alanine aminotransferase ALT
Aspartate aminotransferase AST
Alkaline phosphotase
Albumin
Methylmalonic acid
PSA. total
Prostate specific antigen ratio
TIBC, Frozen Serum
Red cell distribution width
Red blood cell count
Platelet count SI
Segmented neutrophils percent
Mean platelet volume
Mean cell volume
Mean cell hemoglobin
MCHC
Hemoglobin
Hematocrit
Ferritin
Protoporphyrin
Transferrin saturation
White blood cell count
Monocyte percent
Lymphocyte percent
Eosinophils percent
C-reactive protein
Segmented neutrophils number
Monocyte number
Lymphocyte number
Eosinophils number
Basophils number
mean systolic
mean diastolic
60 sec. pulse:
60 sec HR
Total Cholesterol
Triglycerides
Glucose, serum
Insulin
Homocysteine
Glucose, plasma
Glycohemoglobin
C-peptide: SI
LDL-cholesterol
Direct HDL-Cholesterol
Bone alkaline phosphotase
Trunk Fat
Lumber Pelvis BMD
Lumber Spine BMD
Head BMD
Trunk Lean excl BMC
Total Lean excl BMC
Total Fat
Total BMD
Weight
Waist Circumference
Triceps Skinfold
Thigh Circumference
Subscapular Skinfold
Recumbent Length
Upper Leg Length
Standing Height
Head Circumference
Maximal Calf Circumference
Body Mass Index
-0.4 -0.2 0 0.2 0.4
Value
050100150
Color Key
and Histogram
Count
http://bit.ly.com/pemap
phenotypes
exposures
+- EWAS-derived phenotype-exposure association map:
A 2-D view of 86 phenotype by 252 exposure associations

25. Triglycerides
Total Cholesterol
LDL-cholesterol
Trunk Fat
Albumin, urine
Insulin
Total Fat
Head Circumference
Blood urea nitrogen
Albumin
Homocysteine
C-peptide: SI
C-reactive protein
Body Mass Index
Ferritin
Thigh Circumference
Maximal Calf Circumference
Direct HDL-Cholesterol
Total calcium
Total bilirubin
Red cell distribution width
Gamma glutamyl transferase
Mean cell volume
Mean cell hemoglobin
White blood cell count
Uric acid
Protoporphyrin
Hemoglobin
Total protein
Alkaline phosphotase
Waist Circumference
Hematocrit
Weight
Standing Height
1/Creatinine
Creatinine
Trunk Lean excl BMC
Methylmalonic acid
Triceps Skinfold
Lymphocyte number
Subscapular Skinfold
Total Lean excl BMC
Segmented neutrophils number
Lactate dehydrogenase LDH
Bone alkaline phosphotase
TIBC, Frozen Serum
Aspartate aminotransferase AST
Phosphorus
Lumber Pelvis BMD
Glycohemoglobin
Globulin
Chloride
Bicarbonate
Alanine aminotransferase ALT
60 sec. pulse:
Upper Leg Length
Total BMD
Potassium
Glucose, serum
Glucose, plasma
Red blood cell count
Lumber Spine BMD
Platelet count SI
MCHC
Osmolality
Monocyte number
mean systolic
Lymphocyte percent
Segmented neutrophils percent
Recumbent Length
Eosinophils number
Monocyte percent
Head BMD
mean diastolic
Prostate specific antigen ratio
60 sec HR
Basophils number
Sodium
PSA, free
Mean platelet volume
Eosinophils percent
PSA. total
Basophils percent
0 10 20 30 40
R^2 * 100
1 to 66 exposures identified for 81
phenotypes
Additive effect of E factors:
Describe less than 10% of variability in P
(On average: 8%)
Stan Shaw, Hugues Aschard, JP Ioannidis
σ2
E?
Exposome may enable
realization of
remainder of P (>40%)
Recall: H2 <= 50%

26. What do we do now?
Recommendations from the workgroup

27. Data workgroup recommendation highlights
Comprehensive catalog of documented environmental associations
(e.g., risk, variance explained) to strengthen case for exposome.
Where is evidence robust (e.g., air pollution and CVD)?
Where do we see non-replication?
Where is heritability low and ripe for exposome?
Identify technologies that can measure the exposome.
Targeted and untargeted metabolomics.

28. Develop high-throughput data analytic capability.
Statistical methodologies for the 3D matrix!
Encourage a shift from 1 E to many Es.
Link external and internal exposome measures.
Data workgroup recommendation highlights
tim
e
exposome phenome
pollutants
diet
m
etabolites . . .
gut flora
height
w
eight
CVD BP
T2D
cancer
xenobiotics . . .
individuals
GWAS, RVAS,
pathway
analysis..etc.
EWAS,
PheWAS..etc.
genome(static)
mixtures of
exposures
tim
e
drugs
integrative
mixtures of
phenotypes
(A) (C)
(B)
Develop data repositories to house and disseminate individual-level
exposome data.
Assess the variability of the exposome in diverse populations

29. Data workgroup recommendation highlights
Identify data standards for exposome research.
Develop data standards to enable the re-use of research to build
large exposome-rich cohorts.
Identify analytics standards for reproducible research.
Software libraries and tools to share methods and findings.
Incentivize other parties (e.g., researchers, funders, and industry) to
integrate the exposome in their existing programs.

30. Data workgroup recommendation highlights
Educate.
Identify example datasets (e.g., NHANES, DEMOCOPHES).
Hackathons and challenges to recruit data scientists.
Develop big data training support (e.g., K awards) directed at
exposome-related research

31. google:“niehs chear”

32. Informatics will enable us to decipher the role of the emerging
exposome in phenotypes to capture the missing σ2
P
σ2
P = σ2
G + σ2
E

33. Arjun Manrai (Harvard)*
Yuxia Cui (NIEHS)
Pierre Bushel (NIEHS)
Molly Hall (Penn State, now Penn)*
Spyros Karakitsios(Aristotle U, Greece)
Carolyn Mattingly (NCSU)
Marylyn Ritchie (Geisinger/Penn State)
Charles Schmitt (NIEHS)
Denis Sarigiannis (Aristotle U, Greece)
Duncan Thomas (USC)
David Wishart (U Alberta, Canada)
David Balshaw (NIEHS)
Thanks again to the group:
Funded in part by the NIEHS.

34. chirag@hms.harvard.edu
@chiragjp
www.chiragjpgroup.org
Thank you.