Skip to main content

Full text of "MOSAIC EPIGENETIC DYSREGULATION OF ECTODERMAL CELLS IN AUTISM SPECTRUM DISORDER"

See other formats


SUPPLEMENTAL MATERIAL 



Berko et al. 



Mosaic ectodermal epigenetic dysregulation in autism spectrum disorder. 

Berko et al. 



Table of contents: 

Page Description 

2 Patient cohort 

3 Sample collection 

4 Genotyping and methylation microarrays 

5 Genotyping data preprocessing 
5 Local ancestry deconvolution 

7 Mosaicism detection 

9 Methylation data preprocessing 

9 Differentially-methylated region (DMR) identification 

14 Copy number variant (CNV) calling 

14 Massively-Parallel Bisulfite Sequencing to verify differential methylation 

16 Gene ontology enrichment analysis 

31 Weighted Gene Co-expression Network Analysis (WGCNA) 

32 Protein -Protein Interaction (PPI) analysis 
41 Code 

44 Supplemental references 
Illustrations 



Page 


Description 




2 


Supplemental 


Table S1 


3 


Supplemental 


Figure S1 


4 


Supplemental 


Table S2 


6 


Supplemental 


Table S3 


8 


Supplemental 


Figure S2 


10 


Supplemental 


Figure S3 


11 


Supplemental 


Figure S4 


12 


Supplemental 


Table S4 


13 


Supplemental 


Figure S5 


14 


Supplemental 


Table S5 


15 


Supplemental 


Figure S6 


16 


Supplemental 


Figure S7 


17 


Supplemental 


Table S6 


33 


Supplemental 


Figure S8 


34 


Supplemental 


Table S7 


37 


Supplemental 


Table S8 


40 


Supplemental 


Table S9 



1 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Patient Cohort 

All patient recruitment and sample collection was done with the appropriate human subjects 
protocol approval from the Institutional Review Board at the Albert Einstein College of Medicine. 

We enrolled two groups of subjects: descriptive traits of the two groups are summarized in 
Supplemental Table S1. 

We recruited subjects with an ASD by two primary methods: enrolling patients seen in our 
clinics, and recruiting patients nationally from advertisements and a posting on the Autism 
Speaks website (http://www.autismspeaks.org). Patients enrolled from our clinics and research 
labs (43 subjects) were diagnosed with an ASD by combinations of metrics including the ADOS, 
ADI-R, CARS, DSM-IV, and clinical assessment, by clinicians and researchers trained in ASD 
diagnostic measures. 

Other subjects enrolled in our study by internet and external recruitment (7 subjects) provided 
reports of an ASD diagnosis by ADOS and/or ADI-R from an accredited institution with one 
exception, a patient diagnosed with an ASD by the New York City Early Intervention Program. 

The control cohort consisted of typically developing individuals without any prior evidence of an 
autism spectrum disorder. 





ASD 


TD 


Ages 


Maternal age (mean, SD) 


37.58, 2.93 


38.1, 2.97 


(range) 


(35-48) 


(35-48) 


Paternal age 


40.19, 6.27 


40.72, 5.87 




(29-51) 


(30-52) 


Subject Age 


6.84, 3.58 


11.2, 7.31 




(2-17) 


(1-28) 


Gender 


Male (n) 


39 


22 


Female 


11 


28 


Genetic Ancestry 


% CEU (mean, SD) 


63.38, 29.37 


85.19, 16.38 


% YRI 


22.43, 30.74 


6.61, 17.02 


Total 


50 


50 



Supplemental Table S1 : Characteristics of study subjects. 



Age metrics reflect the mean age in years per group and the standard deviation, with the range 
included in parentheses. Maternal and paternal age refer to the age of the parents at the time of the 
subject's birth. All maternal ages are included, while information for paternal age was only available 
for -50% of subjects. Subject age refers to age when the sample was collected. Percent genetic 
ancestry based on subjects with quality filtered genotype data; 47 ASD and 46 TD genotypes were 
included. 

ASD: Autism Spectrum Disorder. 
TD: Typically Developing. 

CEU: Utah residents with Northern/Western European ancestry from the CEPH collection (European) 
YRI: Yoruba in Ibadan, Nigeria (African) 



2 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Sample Collection 

We optimized a method to collect DNA from buccal epithelium, using brushes to exfoliate the 
cheek and cheek gutters (area between the upper/lower gumline and cheeks). Brushes were 
then placed in CytoLyt (Hologic, MA), a methanol-based preservative. Upon arrival in the 
laboratory, brushes were removed from the solution, and cell pellets were retrieved from the 
sample, snap frozen in liquid nitrogen, and moved to -80°C for long term storage. 

Cell subtype composition was determined by preparing slides from each of the samples 
collected, performing haematoxylin and eosin (H&E) staining, and visualizing using light 
microscopy, counting the first 25 individual (not clumped) cells seen. The typical appearance of 
these cells plus examples of cell-like structures counted as contaminants in Supplemental 
Figure S1a-b. Comparably high purity of the squamous epithelial cells was found for both 
groups. 

We performed DNA extraction from these cells and confirmed the presence of high quality and 
high molecular weight DNA with UV spectrometry and gel electrophoresis. 



9 Examples of buccal epithelial samples (H&E stained, light microscopy) 







(ii> 




PT 




* 






i. 






* 












V 








% 


V 


« 








4 










* 



(iv) (V) (vi) 



D Examples of cells counted as contaminants (arrowed) 



■f 

% 



6) 


(ii) _ 


m c 








• 














it, 















(iv) » (V) 

9 



1 



9 



C Percentages of buccal epithelial cells in each cohort (25 cells counted per sample) 

Median Mean Standard deviation 
Typically developing (TD) 100 97.96 2.35 
Autism spectrum disorder (ASD) 100 97.86 2.52 

Supplemental Figure S1: H&E staining of buccal epithelial samples. 

The buccal brushings yielded the expected squamous epithelial cells (a) with very small proportions of 
possible contaminant cells types (b) of different sizes (b.i-b.iv) or nuclear characteristics (b.v). Counts 
of 25 individual cells per sample showed high purity in both ASD and TD groups (c). 



3 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Genotyping and Methylation microarrays 

For the epigenome-wide association study (EWAS), we used a microarray-based approach, as 
the presence of oral bacterial DNA in the extracted samples precluded cost-effective 
sequencing. 

Samples were randomized across arrays, and plates were submitted in batches of 96 to 
minimize potential for batch effect. Sufficient DNA was extracted to meet manufacturer's 
requirements of 200 ng and 500 ng for the genotyping and methylation arrays, respectively, 
Microarray hybridization and subsequent scanning was performed by a core facility according to 
manufacturer's protocols. The summary of samples run on each microarray is provided in 
Supplemental Table S2. 



The lllumina HumanOmni2.5-8 BeadChip genotyping platform was used for genotyping, and the 
lllumina Infinium HumanMethylation450 Beadchip platform for DNA methylation studies. 



ASD 



TD 



Parents 



Genotyping microarrays 



Batch 1 



24 



24 

(including 2 siblings) 



47 



Batch 2 



26 



26 

(including 5 siblings) 



45 



Methylation microarray 



Batch 1 



47 



48 



0 



Supplemental Table S2: Experimental batches of microarrays 



4 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Genotyping Data Preprocessing 

Raw data from the genotyping microarrays were loaded into BeadStudio, lllumina's proprietary 
chip analysis software. Genotype calls were then exported in plink format, specifying 
conversion to the forward strand, which refers to annotation based on dbSNP. 

We used plink to filter the data, with the following parameters for missingness and allele 
frequency: 

-mind 0.05 -geno 0.1 -maf 0.01 -hwe 0.001 

Filtering retained 184 out of 191 individual arrays, and 1,777,521 out of 2,379,855 SNPs, with 
genotyping rate of 0.992 in these remaining samples. 

We then removed all ambiguously mapping single nucleotide polymorphisms (SNPs), 
designated by lllumina as "chrO". 

Since we wanted to perform local ancestry deconvolution based on the 1000 Genomes data, 
which by convention is annotated on the "+" strand of hg19, we had to ensure our lllumina data 
likewise all mapped to the "+" strand. We therefore identified which of the dbSNP 
polymorphisms represented on the array actually mapped to the "-" strand, and used plink to flip 
those bases exclusively. We subsequently further removed the handful of SNPs on each 
chromosome that still possessed two variant alleles. We confirmed correct strandedness by 
merging our genotyping data with 1000 Genomes data and performing principal components 
analysis using vcftools (Danecek et al. 2011). 



Local Ancestry Deconvolution 

Since principal components analysis of our dataset revealed many individuals of mixed 3-way 
genotype admixture, we chose to perform local ancestry deconvolution with simulated mixed 
ancestry using HapMix, an approach previously shown to be accurate (Price et al. 2009). 
HapMix is an algorithm which uses a hidden Markov model to estimate the probability of 
ancestry at each data point along a chromosome, based on the input of 2 parental ancestral 
populations and rates of recombination. To model 3-way admixture, we ran two iterations of 
HapMix with different mixed parental populations, and calculated the joint probability of 
homozygous or heterozygous genotypes at each point (Price et al. 2009). 

As in Price et al., we constructed 2 sets of reference populations using genotype data from the 
1000 Genomes project from Caucasians (CEU), Africans (YRI) and East Asians (CJ: CHB + 
JPT). After randomly selecting 80 unrelated individuals from each group, we constructed two 
mixed sets: a CEU/(CHB+JPT) set with 40 individuals from each group (80 total), and a 
CEU/YRI set with 40 individuals from each group (also 80 total). We then ran HapMix twice for 
each chromosome; first with the 2 parental populations as the CEU/(CHB+JPT) mixed set and 
homogeneous YRI, and then with the 2 parental populations as the CEU/YRI mixed set and 
homogeneous (CHB+JPT). 



5 



SUPPLEMENTAL MATERIAL 



Berko et al. 



For each HapMix run, at every locus, HapMix estimates the probability of a homozygous call of 
the genotype of parental population 1, a heterozygous call of one allele from each parental 
population, and a homozygous call of the genotype of parental population 2. To calculate the 
final joint probability from both runs, we employed the following calculations (Price et al. 2009), 
explained in Supplemental Table S3: 

RUN 1: CEU/(CHB+JPT) versus YRI as the two potential parental populations 

RUN 2: CEU/YRI versus (CHB+JPT) as the two potential parental populations 

The probabilities of the six potential states at each locus were normalized to 1, and final 
ancestry calls were made as the maximum probability at the locus only if it exceeded 0.5. 



Call 


Probability Calculation (run number) 


Homozygous YRI 


rrODaDllliy Ynl (TrOITI run \) A 

Probability CEU/YRI (from run 2) 


Homozygous CJ 


Probability CEU/(CHB+JPT)(1) X 
Probability (CHB+JPT)(2) 


Homozygous CEU 


Probability CEU/(CHB+JPT)(1) X 
Probability CEU/YRI(2) 


Heterozygous YRI/CJ 


Probability heterozygous CEU/(CHB+JPT )& YRI(1) X 
Probability heterozygous CEU/YRI&(CHB+JPT)(2) 


Heterozygous YRI/CEU 


Probability heterozygous CEU/(CHB+JPT)&YRI(1) X 
Probability CEU/YRI(2) 


Heterozygous CJ/CEU 


Probability CEU/(CHB+JPT)(1) X 
Probability heterozygous CEU/YRI&CJ(2) 



Supplemental Table S3: HAPMIX local ancestry probability calculation 



Abbreviations: 

CHB: Han Chinese in Beijing, China. 

JPT: Japanese in Tokyo, Japan 

CJ: CHB/JPT (East Asian) 

YRI: Yoruba in Ibadan, Nigeria (African) 

CEU: Utah residents with Northern and Western European ancestry from the CEPH collection 
(European) 



6 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Mosaicism Detection 

We used the Mosaic Alteration Detection (MAD) algorithm, implemented in the package GADA, 
to identify potential chromosomal mosaic events (Gonzalez et al. 2011). To confirm our ability 
to use MAD to discover detectable whole chromosome mosaicism, we prepared a test dataset 
of 4 samples. The samples consisted of DNA from 3 cell lines, known to harbor mosaic trisomy, 
from Coriell Cell Repository, and one DNA sample extracted from buccal epithelium of a patient 
with a clinical diagnosis of mosaic trisomy 18. Samples were hybridized on the lllumina Infinium 
Omni1-Quad (prior to lllumina's release of the 2.5-8 BeadChip) and genotype calls were made 
with lllumina Genome Studio. We plotted the B-Allele Frequencies (BAF) of each known 
mosaic chromosome (Supplemental Figure S2) confirming the presence of visually detectable 
mosaic trisomy. From the BAF plots we were further able to characterize the mosaicism as 
resulting from either a meiosis I or meiosis II non-disjunction, based on loss of the 5 th and 6 th 
genotype bands at either the telomere of centromere, respectively. We were also able to 
identify segments of uniparental disomy (UPD), based on loss of heterozygous bands (Conlin et 
al. 2010). We then analyzed the genotypes with MAD, employing the default suggested 
parameters (aAlpha=0.8, T=9, and MinSegLen=75). Although it also called mosaic events on 
other chromosomes, MAD correctly identified the whole chromosome mosaic trisomy in all 4 
cases, either with an over-abundance of mosaic calls or by calling exceptionally long mosaic 
regions on the affected chromosome. 

To prepare our genotyping dataset for MAD analysis, we first re-clustered the data in 
GenomeStudio to improve the quality of some poorly performing SNPs. After exporting the data 
and applying the same plink filters shown previously, we retained 186 out of 191 individual 
arrays and 1,820,947 SNPs. We ran MAD with the same default parameters as our test 
dataset, obtaining 376 called mosaic segments across the 186 individuals. No whole 
chromosome mosaicism events were detected in either the ASD or the TD subjects. 



7 



SUPPLEMENTAL MATERIAL 



Berko et al. 



AG 13074 Cell Line 




2e+07 



4e+07 
Chromosome 18 Position 



GM00503 Cell Line 




2e+07 



6e+07 8e+07 
Chromosome 13 Position 




F44P110 



Oe+00 



2e+07 4e+07 

Chromosome 18 Position 



6e+07 



Supplemental Figure S2: B allele frequency plots of known mosaic cases 

The MAD output correctly highlights an abnormality along chromosomes; analysis of BAF patterns 
allows determination of the source of error. The pattern in AG13074 results from meiosis I non- 
disjunction, and F44 P110 demonstrates a meiosis II non-disjunction with mosaic UPD. The lack of 
array probes on the p arm of chromosome 13 precludes definitive assessment of the meiotic source of 
the GM00503 error (I or II); the pattern of GM00682 could result from a variety of situations. 



8 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Methylation Data Preprocessing 

Supplemental Figure S3 provides an overview of the preprocessing pipeline. 

Raw idat data were loaded in the R (RCoreTeam 2012) package minfi (Hansen and Aryee) and 
assessed for basic quality control metrics, including determination of poorly performing probes 
with insignificant detection p-values above background control probes. Since our cohort 
includes both males and females, we removed the X and Y chromosomes from the raw 
methylation values, and performed SWAN normalization to correct for intra-array differences 
between lllumina Type I and Type II probes (Dedeurwaerder et al. 2011; Maksimovic et al. 
2012). We excluded any probes with detection p-value greater than 0.01 (12,590 probes), and 
then corrected for batch (microarray chip) effect using the ComBat function in the R package 
SVA (surrogate variable analysis) (Leek et al.). 



Differentially-methylated region (DMR) Identification 

To understand the relative effects of known technical, biological, and microarray-based 
covariates acting on methylation data variability, we performed principal components analysis 
(PCA) on the M values (logit-transformed lllumina-defined beta values) obtained from the 
previous preprocessing. We accounted for the possible known confounders, including technical 
(date of DNA extraction, microarray chip, position on chip), microarray-based (all categories of 
control probes designed by lllumina) and biological (ASD status, age, gender, and ancestry 
percentage). Ancestry percent was calculated as the proportion for each population of all allele 
genotyping positions called by HapMix. We fit a linear model for each of the top 10 principal 
components as a function of each covariate, and summarized the data with a heatmap of the 
negative log™ p-values for each regression. We identified the significant confounding 
covariates and corrected for them in all subsequent analysis. 

Bump-hunting provides significant advantages over typical individual probe statistical test: it 
smoothes data over a region, obliterating the need for arbitrary genomic cutoffs, it applies a 
rigorous analysis and correction for confounding factors, and it incorporates false discovery rate 
(FDR) considerations as part of the algorithm. Bump-hunting utilizes known covariates input 
into a model matrix as components in regression analyses. Although we did not see a strong 
effect of gender on PCA variability, we included gender in our model matrix since methylation 
patterns are known to vary by gender (Sarter, Long et al. 2005). Based on our PCA data, we 
input age, percent CEU (European), and percent YRI (African) ancestry as known biological 
covariates. Bump-hunting was used in the dmrFind function in the R package charm (Aryee et 
al. 2011). 



9 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Raw Data 



Load raw data 
into R package minfi 
Assess array-based 

probe controls 



Transform lllumina Beta 
values to M values 
M value= Logit Beta= 
Log2(Beta/1-Beta) 



Remove X and Y 
chromosomes 



Subset Within-Array 
Normalization (SWAN) 

y 



Removed failed probes, 
ie, with detection 
P-value >0.01 
(based on values from 
raw intensity data) 



COMBAT normalization to 
correct for batch effect 
using R package SVA 




1 r 

10000 20000 
Intensity 



Transformed Autosomes Data 



30000 



0J J± 




-5 0 

M value/Logit Beta 



SWAN Normalized & Filtered Data 




COMBAT adjusted SWAN Normalized & Filtered Data 




r 

-4 -2 0 

M value/Logit Beta 



Supplemental Figure S3: Preprocessing of lllumina 450K data 

The colors in the density plots correspond to different chips (microarrays), each run to contain 12 
samples. The normalization procedures correct for both intra-array and inter-array differences. 



10 



SUPPLEMENTAL MATERIAL 



Berko et al. 



We excluded two TD samples from our analysis, since the genotyping arrays performed on their 
DNA did not pass quality control thresholds. 

The dmrFind algorithm corrects for unidentified confounders, and returns probe methylation 
values that preserve the effects of the known covariates input into the model matrix. PCA and 
linear regression on this data confirmed that SVA properly identified and corrected for all 
unwanted technical sources of variation, Supplemental Figure S4. 



Biological 
Covariates 



Technical 
Covariates 



Control 
Probes 



ASDfTD 
Gender 
Age 

African % 
Caucasian % 
Asian % 
Microarray 
Position 

Date DNA Extracted 

BSI Green 

BSI Red 

BSII Green 

BSI I Red 

NPC Green 

NPC Red 

i-CMootLncor^oooo 

OOOOOOOOOTT 
clclclclclclclclclO 

Q_ 



o 

o 



o 
00 



o 



1 1 r 



5 10 



20 



-Log 10 P-value 
of association 



Supplemental Figure S4: Associations between principal components and known covariates 

Heatmap of -log 10 P-values for the association of each principal component with each known 
covariate demonstrates that variation due to technical artifact has been removed, while variation due 
to known biological covariates has been preserved for subsequent analysis. 



11 



SUPPLEMENTAL MATERIAL 



Berko et al. 



The bump-hunting ASD-associated DMR output is provided in Supplemental Table S4. Stable 
DMRs are visualized in Supplemental Figure S5. 

The differentially methylated regions were associated with the nearest overlapping gene, shown 
in Table 1. We investigated autism gene databases (Xu et al. 2012) and published literature for 
previous evidence linking the gene to ASD. We used UCSC genome browser (Kent et al. 2002) 
to determine the cytoband, and the SFARI gene CNV database (Basu et al. 2009) for a 
summary of CNV research implicating the cytoband in ASD. Gene functions were gleaned from 
UCSC genome browser and the GeneCards database (Rebhan et al. 1997). 



chromosome 


start 


end 


value 


area 


pns 


indexStart 


indexEnd 


n probes 


avg 


max 


area.raw 


gene name 


lllumina annotation 


1 


248,100,183 


248,100,614 


0.427 


4.273 


23,221 


45,365 


45,374 


10 


0.080 


0.150 


0803 


OR2L13 


1st exon 5'UTR TSS200 and TSS 
1500, CpG island and north shore 


2 


241 ,290,446 


241,291,070 


0 381 


1.904 


137,222 


274,789 


274,793 


5 


0.086 


0,104 


0.428 


GPC1 


north shelf 


2 


114,033,360 


114,033,830 


0.331 


1.323 


128,333 


258,706 


258,709 


4 


0.075 


0.105 


0.301 


PAX8 


gene body, CpG island and north 
shore 


4 


3,748,154 


3,748,554 


0.465 


2.324 


163,348 


326,311 


326,315 


5 


0 119 


0.148 


0 595 ADRA2C 


CpG island and north shore 


5 


139,227,979 


139,228,242 


-0.158 


0.790 


180,529 


358,423 


358,427 


5 


-0.076 


-0.127 


0.378 


NRG2 


gene body, CpG island 


5 


16,508,920 


16,509,123 


0.373 


1.491 


174,420 


347,241 


347,244 


4 


0.078 


0.095 


0.312 


FAM134B 


TSS200, 5'UTR, 1st exon, body, 
enhancer 


6 


73,329,988 


73,330,358 


-0.209 


1.253 


192,434 


389,727 


389,732 


6 


-0.077 


-0.088 


0.461 


KCNQ5 


TSS1500, CpG island north shore 


7 


28,452,066 


28,452,289 


-0.202 


0.806 


202,381 


410,259 


410,262 


4 


-0.076 


-0.083 


0302 


CREB5 


TSS200, 5'UTR, 1st exon, CpG 
island south shelf 


10 


135,341,870 


135,342,620 


-0491 


2.947 


35,126 


69,284 


69,289 


6 


-0.076 


-0.148 


0.469 


CYP2E1 


gene body, CG island and south 
shore 


10 


135,342,936 


135,343,280 


-0.462 


1.850 


35,127 


69,290 


69,293 


4 


-0.078 


-0.098 


0.310 


CYP2E1 


gene body, south shore 


12 


117,797,056 


117,797,635 


-0.213 


1.067 


57,778 


115,485 


115,489 


5 


-0.086 


-0.109 


0.432 


NOS1 


5'UTR, north shore 


16 


53,407,678 


53,407,808 


0.174 


0.696 


87,732 


174,199 


174,202 


4 


0.089 


0.115 


0.357 LOC643802 


CpG island, south shore 


16 


2,879,944 


2,880,326 


-0.265 


1.060 


83,603 


166,459 


166,462 


4 


-0.101 


-0.125 


0.402 


ZG16B 


TSS200, TSS1500, 1st exon 


,. 


1 ,796,832 


1,797,383 


-0.515 


2.060 


83,090 


165,290 


165,293 


4 


-0.090 


-0.139 


0.361 


MAPK8IP3 


gene body, CpG island and south 
shore 


19 


12,876,846 


12,877,188 


0.401 


1.603 


111,948 


225,403 


225,406 


4 


0.079 


0.086 


0.315 HOOK2 


gene body, CpG island, north 
shore and south shore 



Supplemental Table S4: DMRs Associated with ASD from Bump-Hunting. 

Unstable DMRs are shaded in gray. 



12 



SUPPLEMENTAL MATERIAL 



Berko et al. 



DMR;-rtiM:374B154-374B55i 




DMR5 Ml r2: 24 1290446-241 281 070 








1 j 


i 


i 
| 

! 











[)MRin-mriar1?B7B«4B-i;B771«1 





ASD 
TD 




Supplemental Figure S5: Stable DMRs defined by the bump-hunting algorithm 

DNA methylation values are displayed along with -log 10 p-values for each probe. 
Left column, from top to bottom, gene names: ADRA2C, GPC1, CREB5, HOOK2. 
Right column, top to bottom: PAX8, NOS1, MAPK8IP. 



13 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Copy number variant (CNV) Calling 

The CNVision algorithm (Sanders et al. 2011) integrates three CNV-calling platforms: GNOSIS, 
PennCNV, and QuantiSNP (Colella et al. 2007), increasing accuracy of calls with this 
complementary assessment of each CNV. Since we wanted to exclude any loci with potential 
CNVs from methylation analysis, we retained all called CNVs, even those called with low 
potential probability. 

When DMRs were found to overlap CNVs, we re-ran bump-hunting with these loci excluded to 
confirm that the local DMR was not solely due to the presence of CNVs in the region. 



Massively-parallel bisulfite sequencing to validate differential DNA methylation 

We bisulphite converted 500 ng of DNA using the Zymo EZ-96 Methylation-Lightning Kit. After 
separate PCR amplification of 4 target regions, we pooled the amplicons in averaged equal 
ratios and generated lllumina libraries using Tecan automation. Two sets of 48 libraries each 
were multiplexed on the MiSeq. Using bsmap (Bisulphite Sequencing Mapping Platform) (Xi 
and Li 2009) we checked for bisulphite conversion efficiency (C->T in CH contexts) and 
quantified the percent methylation for each person at every CpG in the amplicons. 

We validated differential methylation for 3 predicted DMRs: the loci in NOS1, FAM134B, and 
OR2L13, and tested one of the loci with unstable DMR prediction (at the KCNQ5 gene). The 
loci tested were chosen on the basis of their potential functional relevance and technical ability 
to design bisulphite PCR primers covering CpGs that overlapped with array probes. 
Supplemental Table S5 shows the primers used. 



Locus 


Forward Primer 


Reverse Primer 


chrl2 : 117797179- 
117797517 


GGGGAAAAAATTTATGTTTTAGAGAG 


AAAATTCTTCCCTCTACTCCCATAAC 


chr6: 73330181- 
73330482 


GGTTTTTGTTGGTGATTAGGAGTAG 


AAAAACAAAAC TAAACTTCCACCAC 


chrl : 248100298- 
248100643 


TTTTATTGTTTTTGGGGTTAATTAT 


C AC CAAT AT AT AAAACAAAAC CC T T C 


chr5: 16508806- 
16509201 


TATTTTAATGTTGAATATTAGGAGGAAAA 


AAC CACTCCACCCT T AAAT AAAT AC 



Supplemental Table S5: Primers for bisulphite-converted DNA 

We showed a concordance of DNA methylation changes between microarray and sequencing- 
based approaches for all loci, but the degrees of difference of DNA methylation at the DMRs at 
NOS1 and KCNQ5 are insufficient to allow significant associations to be assigned statistically 
(Supplemental Figure S6). 



14 



SUPPLEMENTAL MATERIAL 



Berko et al. 



NOS1 



P=0 .177 KCNQ5 



p=0.0S588 





CpGs 



CpGs 



Supplemental Figure S6: Results of massively-parallel bisulphite sequencing of the NOS1 and 
KCNQ5 DMRs. The p values were calculated using t tests combining all of the loci tested in the 
putative DMR and comparing between the ASD and TD groups. 



15 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Gene Ontology Enrichment Analysis 

For age-associated DMRs obtained from bump-hunting, we plotted only the enrichment 
categories with a threshold of the p-value with a significance of 10" 7 or lower. Enrichment 
category results are displayed in Supplemental Figure S7 and the DMRs themselves in 
Supplemental Table S6. 




Supplemental Figure S7: Gene ontology analysis of genes associated with age-related DMRs. 

Panel A shows connectivity of the gene ontology categories significantly enriched for genes related to 
age-associated DMRs, plotted by corrected p-value in Panel B. 



16 



SUPPLEMENTAL MATERIAL 

Supplemental Table S6: Age-associated candidate DMRs. 



Berko et al. 



chr 



start 



indexEnd nprobes 

end value area pns indexStart avg max area. raw 




132,157 

6,986,460 176,987,465 0.029 0.318 265,348 265,358 



11 0.732 0.855 



chM 231,155,632 231,156,204 -0.039 0.467 21,845 42,875 42,886 



chr19 58,220,295 58,220,837 0.027 0.271 119,212 241,574 241,583 



8 054 
8.054 



12 -0.609 -0.707 7.312 




10 0.697 0.817 6.965 



Chr8 23,563,970 23,564,717 0.027 0.246 215,946 

12,843 



436,377 436,385 



chr19 9,473,565 9,473,781 0.032 0.285 111,176 223,482 223,490 



chr10 22,634,038 22,634,226 0.046 0.320 25,480 49,991 49,997 

63,849 127,718 
189,003 380,630 



chr6 32,078,398 32,078,624 -0.028 0.225 



380,637 



9 0.729 0.859 6.557 




9 0.691 0.797 6.217 




7 0.832 0.864 



5.825 
5.74 




8 -0.691 -0.752 5.525 



chr14 


29,234,890 


29,235,196 


0.029 


0.229 


67,445 


135,202 


135,209 


8 


0.666 


0.827 


5.332 




chr3 


147,126,638 


147,127,097 


0.035 


0.244 


158,563 


316,905 


316,911 


7 


0.752 


0.852 


5.264 


5.251 


chr4 


155,661,691 


155,662,795 


0.029 


0.259 


170,611 


339,758 


339,766 


9 


0.556 


0.630 


5.004 




86,383,182 


86,383,430 


0.041 


0.286 


44,708 


90,004 


90,010 


7 


0.707 


0.738 




chr22 


24,890,690 


24,890,833 


0.043 


0.301 


145,700 


292,479 


292,485 


7 


0.678 


0.766 


4.744 



17 



SUPPLEMENTAL MATERIAL 



Berko et al. 




chr7 


99,775,422 


99,775,558 


0.034 


0.240 


207,428 


419,896 


419,902 


7 


0.666 


0.745 


4.661 




57,275,967 


57,276,789 




0.204 


68,821 


137,800 






0.773 




4.638 


chr7 


27,225,058 


27,225,143 


0.029 


0.172 


202,219 


409,908 


409,913 


6 


0.772 


0.848 


4.631 










0.255 


94,610 














chM 


248,020,436 


248,020,812 


0.043 


0.258 


23,207 


45,341 


45,346 


6 


0.749 


0.833 


4.492 




chr16 


87,864,324 


87,865,062 


-0.050 


0.300 


91,530 


181,608 


181,613 


6 


-0.721 


-0.750 


4.325 




chr2 


74,875,227 


74,875,387 


0.042 


0.254 


125,794 


254,086 


254,091 


6 


0.713 


0.788 


4.276 




24,772,137 


24,772,350 


0.039 


0.232 


216,010 


436,484 


436,489 


6 








chr14 


54,413,218 


54,413,931 


0.052 


0.310 


68,535 


137,302 


137,307 


6 


0.704 


0.806 


4.221 






103,603,869 


0.029 


0.205 


31,479 


61,239 


61,245 


7 


0.602 


0.702 




chM 


91,301,204 


91,301,962 


0.029 


0.172 


1 1 ,874 


23,499 


23,504 


6 


0.695 


0.766 


4.170 












156,324 


312,596 


312,600 


5 


0.830 


0.849 


4.148 


chr2 


10,182,878 


10,183,227 


0.043 


0.258 


120,711 


244,696 


244,701 


6 


0.690 


0.742 


4.140 




chr11 2,292,751 2,292,914 0.036 0.253 36,185 72,072 72,078 

391,945 391,950 

chr6 85,474,028 85,474,209 0.030 0.178 192,887 390,662 390,667 

45,079 90,675 
177,666 353,160 



chr5 87,980,882 87,981,253 0.026 0.158 

0.190 134,1 



353,165 



7 0.588 0.751 4.116 

6 0.686 0.807 4.113 

6 0.680 0.859 4.083 
4.C 




6 0.670 0.802 4.020 



18 



SUPPLEMENTAL MATERIAL 



Berko et al. 



chr22 


30,476,089 


30,476,285 


0.033 


0.230 


146,109 


293,345 


293,351 


7 


0.565 


0.685 


3.953 




chr6 


29,943,268 


29,943,480 


0.038 


0.308 


188,133 


374,577 


374,584 


8 


0.489 


0.583 


3.911 
























3.895 






79,472,452 


0.030 


0.152 


11,278 


22,419 


22,423 


5 


0.779 


0.847 














106,733 


214,413 












chr18 


44,526,430 


44,527,026 


0.031 


0.246 






214,420 


8 


0.485 


0.657 


3.883 










0.128 


64,614 
















100,624,279 


100,624,373 


0.026 






129,099 




5 


0.776 


0.832 




chr5 


43,017,982 


43,018,629 


0.045 


0.227 


175,429 


349,043 


349,047 


5 


0.760 


0.817 


3.802 




chr22 


24,181,191 


24,181,270 


0.034 


0.169 


145,604 


292,299 


292,303 


5 


0.754 


0.818 


3.771 


chr5 


172,672,390 


172,672,817 


0.027 


0.136 


183,336 


363,882 


363,886 


5 


0.749 


0.773 














14,673 














chM 


151,810,586 


151,810,904 


0.031 


0.187 




29,289 


29,294 


6 


0.621 


0.691 


3.729 




chr7 


100,463,416 


100,464,145 


-0.059 


0.353 


207,617 


420,346 


420,351 


6 


-0.615 


-0.639 


3.692 




chr4 


85,402,870 


85,403,409 


0.027 


0.137 


167,592 


334,286 


334,290 


5 


0.732 


0.761 


3.658 














44,685 














chr11 


86,085,623 


86,086,005 


0.026 


0.158 




89,972 


89,977 


6 


0.609 


0.731 


3.652 




chr2 


223,164,831 


223,164,925 


0.027 


0.134 


135,087 


270,748 


270,752 


5 


0.727 


0.805 


3.637 




Chr3 147,125,712 147,125,782 0.031 0.157 158,563 316,888 316,892 

0.032 0.159 192,892 390,680 390,684 

186,057 369,233 




chr6 11,044,877 11,044,974 0.067 0.269 



369,236 



5 0.717 0.828 3.586 
3.573 

5 0.715 0.793 

4 0.892 0.905 3.568 



19 



SUPPLEMENTAL MATERIAL 



Berko et al. 



chM 91,190,366 91,190,891 0.028 0.139 11,865 23,457 23,461 



chr16 1,593,152 1,593,766 -0.034 0.202 83,002 165,085 165,090 



chr11 2,891,065 2,891,118 0.031 0.154 



72,730 72,734 



5 0.705 0.795 3.527 




6 -0.583 -0.705 3.499 
3.4 




5 0.691 0.737 3.456 



chr10 93,805,441 93,805,870 0.028 0.139 30,293 58,645 58,649 



chr14 100,069,535 100,069,840 0.055 0.222 72,269 144,115 144,118 

chr22 40,417,285 40,417,869 -0.043 0.214 147,204 295,668 295,672 

57,370 



chr12 113,916,473 113,916,664 0.030 0.122 



114,617 114,620 




.128 220,041 



chM 164,545,553 164,546,143 0.035 0.174 16,595 33,260 33,264 



chr6 101,846,779 101,846,872 0.026 0.130 193,554 392,047 392,051 



5 0.669 0.772 3.344 




4 0.827 0.851 3.307 

3.264 

5 -0.653 -0.769 

4 0.811 0.828 3.242 




5 0.646 0.810 3.229 




5 0.638 0.692 3.191 



192,434 389,727 

chr6 73,329,988 73,330,358 0.036 0.217 389,732 



6 0.529 0.610 3.176 



chM 119,535,693 119,535,986 0.038 0.153 13,653 27,185 27,16 



4 0.794 0.872 3.175 



chr11 123,066,529 123,067,275 0.030 0.181 47,241 94,847 94,852 



6 0.529 0.683 3.174 



chr8 56,015,399 56,015,785 0.026 0.105 218,073 440,284 440,287 



4 0.784 0.808 3.135 
3." 



20 



SUPPLEMENTAL MATERIAL 



Berko et al. 



216,106 436,652 

chr8 25,898,191 25,898,539 0.031 0.126 436,655 




chr11 134,147,143 134,147,634 0.034 0.136 48,559 97,232 97,235 



17,958,339 17,958,736 0.032 0.159 113,027 227,806 227,810 



chr19 57,182,844 57,183,268 0.031 0.124 119,037 241,073 241,076 

5,631 28,035,894 0.038 0.267 25,811 50,667 50,673 

63,854 



4 0.778 0.803 3.111 




4 0.773 0.855 3.091 



chr13 79,177,877 79,177,925 0.027 0.107 



127,748 127,751 



5 0.618 0.744 3. 



4 0.768 0.814 3.070 
3.070 

7 0.439 0.487 
4 0.759 0.790 3.037 



chr11 64,146,487 64,146,822 -0.034 0.172 41,583 83,185 83,189 



5 -0.606 -0.642 3.030 



chr7 153,584,416 153,584,609 0.034 0.136 211,900 428,260 428,263 



4 0.757 0.787 3.029 



chM 75,595,919 75,596,336 0.029 0.118 11,085 22,047 22,050 




chr12 54,071,090 54,071,194 0.033 0.163 



106,359 106,363 



chr5 87,968,528 87,968,749 0.029 0.115 177,652 353,113 353,116 



5 0.605 0.704 3.024 




4 0.752 0.830 3.008 



chM 18,959,268 18,959,625 0.025 0.101 4,757 9,510 9,513 



86,317 



chr16 29,625,216 29,625,259 0.036 0.145 



171,434 171,437 



4 0.742 0.756 2.966 




4 0.739 0.794 2.958 





147,111,120 


147,111,308 


0.024 


0.098 


158,547 


316,846 


316,849 


4 


0.737 






chr11 


35,441,558 


35,441,900 


0.029 


0.144 


39,114 


78,143 


78,147 


5 


0.588 


0.723 


2.942 
















267,331 


4 


0.735 


0.837 




chr5 


132,083,532 


132,084,068 


0.032 


0.127 


179,674 


356,716 


356,719 


4 


0.734 


0.835 


2.938 



21 



SUPPLEMENTAL MATERIAL 



Berko et al. 




216,114 436,684 
202,239 409,982 




chr7 27,245,018 27,245,747 0.026 0.103 



409,985 



chr18 70,534,298 70,535,005 0.027 0.135 107,430 215,904 215,908 



chr14 95,239,381 95,239,751 0.026 0.131 71,852 143,386 143,390 



4 0.724 0.765 2.896 




5 0.578 0.744 2.888 




5 0.576 0.861 2.882 



193,460 391,824 

chr6 100,054,585 100,054,817 0.026 0.104 391,827 



chr6 100,442,105 100,442,151 0.037 0.150 193,479 391,871 391,874 



9,868,552 0.027 0.108 106,986 214,941 214,944 




chr3 62,354,991 62,355,443 0.030 0.118 153,1 



308,533 308,536 



chr3 187,387,555 187,387,734 0.031 0.124 160,929 321,238 321,241 



chr6 134,213,992 



134,214,307 0.027 0.109 195,293 395,191 395,194 



4 0.714 0.855 2.855 




4 0.710 0.720 2.841 



4 0.708 0.75 



4 0.708 0.780 2.832 




0.841 2.823 




4 0.699 0.753 2.794 



chr4 52,942,997 52,943,247 0.042 0.169 166,053 331,543 331,546 



4 0.696 0.757 2.784 



Chr2 223,161,771 223,162,128 0.032 0.127 135,085 270,727 270,730 

Chr11 111,385,450 111,385,659 0.027 0.109 45,879 92,155 92,158 

chr15 60,288,082 60,288,404 0.026 0.103 77,285 154,102 154,105 

287,544 

chr21 36,041,605 36,041,699 0.033 0.134 143,100 287,541 

397,938 



4 0.695 0.792 2.780 

4 0.695 0.805 2.779 
2.7 




4 0.694 0.716 2.775 



22 



SUPPLEMENTAL MATERIAL 



Berko et al. 



chr8 98,290,310 98,290,372 0.034 0.137 220,135 444,065 444,068 

chr15 



4 0.692 0.826 2.767 





Chr4 48,486,087 48,486,472 0.026 0.106 165,951 331,335 331,338 

4 0.172 97,771 194,856 194,860 

chrl 19,665.070 19,665.240 0.040 0.161 4,926 9,817 9,820 

1.059 0.237 33,326 64,892 64,895 

0.097 27,342 



4 0.687 0.752 2.748 

5 0.549 0.726 2.744 

4 0.686 0.731 2.744 
0.755 




chr10 50,604,330 50,604,569 0.024 



53,347 53,350 



4 0.680 0.779 2.721 



chr19 10,397,612 10,397,780 0.031 0.125 111,331 223,899 223,902 



4 0.678 0.766 2.713 



chr17 61,778,366 61,778,813 0.025 0.100 101,437 202,761 202,764 




chr20 55,964,998 55,965,497 0.035 0.138 141,209 283,686 283,689 

157,177,345 0.035 
143,545,949 0.034 




chr8 143,545,478 



0.170 223,102 449,175 449,179 



4 0.675 0.730 2.701 

2.688 
2.682 

4 0.671 0.781 
0.671 

5 0.536 0.619 2.680 



chr12 66,627,900 66,628,232 -0.025 0.099 54,519 109,239 109,242 



chM 41,119,634 41,119,988 0.030 0.121 8,190 16,458 16,461 



chr4 151,500,631 151,501,298 0.037 0.148 170,283 339,171 339,174 



4 -0.667 -0.715 2.668 




4 0.660 0.741 2.638 








85,932,853 


0.031 0.123 


90,989 


180,519 


180,522 


4 


0.658 


0.752 




chr2 


177,053,274 


177,053,292 


0.028 0.113 


132,211 


265,539 


265,542 


4 


0.658 


0.724 


2.631 



23 



SUPPLEMENTAL MATERIAL 



Berko et al. 



chr12 56,414,442 56,414,533 0.025 0.099 53,637 107,388 107,391 



chr19 18,260,330 18,260,515 -0.033 0.131 113,106 227,997 228,000 



chr6 110,736,772 110,737,053 -0.033 0.163 194,095 393,045 



4 0.654 0.701 2.616 




4 -0.650 -0.685 2.599 
2.5 




5 -0.517 -0.584 2.586 



chr8 61,777,711 61,778,137 -0.036 0.146 218,407 440,878 440,881 



4 -0.643 -0.724 2.572 




chr11 17,803,160 17,803,421 -0.036 0.145 38,057 76,022 76,025 



,919 46,619,555 0.026 0.104 99,835 199,378 199,381 



chr10 99,734,416 99,734,912 0.034 0.169 30,914 59,888 59,892 




4 -0.642 -0.684 2.567 



4 0.641 



5 0.513 0.616 2.564 

















341,094 


4 


0.640 


0.744 




chM 


200,009,830 


200,010,283 


0.024 


0.097 


18,525 


36,715 


36,718 


4 


0.638 


0.685 


2.551 




54,518,745 


54,519,159 


0.028 


0.113 
















chr6 


30,070,059 


30,070,403 


0.038 


0.339 


188,216 


375,015 


375,023 


9 


0.283 


0.533 


2.543 


















































chr7 


94,284,865 


94,284,900 


0.030 


0.121 


206,839 


418,369 


418,372 


4 


0.634 


0.835 


2.535 




175,208,588 


175,208,761 


0.026 


0.103 


131,985 


264,936 


264,939 


4 


0.631 


0.743 


2.526 


chr14 


94,392,718 


94,392,932 


-0.030 


0.121 


71,715 


143,118 


143,121 


4 


-0.630 


-0.671 


2.519 




50,426,531 


50,427,095 














0.628 






ChM 


17,085,860 


17,086,071 


0.030 


0.121 


4,431 


8,941 


8,944 


4 


0.627 


0.684 


2.508 



24 



SUPPLEMENTAL MATERIAL 



Berko et al. 



chr12 


11,653,278 


11,653,827 


0.024 


0.098 


50,245 


100,630 


100,633 


4 


0.626 


0.842 


2.505 




chr7 


142,494,148 


142,494,244 


0.028 


0.113 


210,754 


426,012 


426,015 


4 


0.622 


0.724 


2.490 




74,663,416 


74,663,698 


0.026 


0.104 


125,719 


253,871 


253,874 


4 


0.622 


0.725 


2.487 


Chr1 


91,300,288 


91,300,446 


0.025 


0.102 


1 1 ,873 


23,494 


23,497 


4 


0.622 


0.653 


2.486 




42,733,527 


42,733,600 


0.033 


0.130 


99,202 


198,076 


198,079 


4 


0.611 






chr10 


110,225,900 


110,226,387 


0.025 


0.099 


31,992 


62,298 


62,301 


4 


0.611 


0.627 


2.443 




chr13 


28,545,214 


28,545,566 


0.025 


0.098 


61,363 


123,182 


123,185 


4 


0.607 


0.697 


2.429 


chr3 


194,408,516 


194,408,901 


0.036 


0.143 


161,474 


322,164 


322,167 


4 


0.607 


0.695 


2.427 


chM 


33,231,272 


33,231,382 


0.035 


0.140 


7,123 


14,255 


14,258 


4 


0.603 


0.623 


2.412 










0.145 


93,375 


185,564 








0.611 




Chr1 


115,881,130 


115,881,259 


0.029 


0.114 


13,375 


26,633 


26,636 


4 


0.601 


0.698 


2.403 


168,519 


chr2 


186,603,398 


186,603,639 


0.030 


0.120 


132,626 


266,293 


266,296 


4 


0.592 


0.691 


2.369 




chr7 


101,512,529 


101,513,100 


0.025 


0.099 


207,835 


420,826 


420,829 


4 


0.592 


0.650 


2.367 




chr10 


94,451,351 


94,451,736 


0.026 


0.105 


30,344 


58,759 


58,762 


4 


0.589 


0.644 


2.358 


















































chr17 


4,648,566 


4,648,949 


0.034 


0.135 


93,932 


186,796 


186,799 


4 


0.586 


0.755 


2.346 






11,530,065 


0.029 


0.145 


1 1 1 ,665 


224,649 


224,653 


5 


0.468 


0.552 




chr8 


10,261,972 


10,262,221 


0.029 


0.117 


214,638 


433,984 


433,987 


4 


0.585 


0.697 


2.338 



25 



SUPPLEMENTAL MATERIAL 



Berko et al. 



chM 


4,770,676 


4,771,201 


0.034 0.136 


2,015 


4,574 


4,577 


4 0.583 


0.707 


2.333 




58,003,774 


58,003,965 


0.025 0.127 




108,274 






0.629 




chr11 


9,025,730 


9,026,308 


0.035 0.140 


37,246 


74,531 


74,534 


4 0.580 


0.758 


2.322 


chr11 












69,535 


4 

























chr17 



7,350,001 7,350,413 0.026 0.103 94,440 188,076 188,079 



4 0.574 0.627 2.297 



chr10 8,097,331 8,097,689 0.032 0.128 24,556 48,236 48,239 



4 0.573 0.672 2.293 





chr13 


112,717,207 


112,717,707 


0.031 


0.126 


65,571 


130,968 


130,971 


4 


0.571 


0.630 


2.284 












21,403 


41,970 


41,973 


4 


0.570 


0.623 


2.282 


chr22 


17,083,412 


17,083,727 


0.025 


0.101 


144,702 


290,507 


290,510 


4 


0.570 


0.658 


2.280 






145,106,582 


0.039 


0.155 


223,787 


450,753 


450,756 










chr6 


125,855,124 


125,855,421 


0.027 


0.110 


194,820 


394,326 


394,329 


4 


0.565 


0.583 


2.261 


chr6 


170,597,326 


170,597,588 


-0.036 


0.145 


198,758 


401,914 


401,917 


4 


-0.562 


-0.663 





chr5 172,110,211 172,110,579 0.028 0.113 183,213 363,608 363,611 



4 0.559 0.695 2.237 



chr12 113,913,695 113,914,222 0.028 0.112 57,369 114,609 114,612 



I 



chM 236,557,182 236,557,682 -0.032 0.127 22,345 

155,320 



chr3 100,712,058 100,712,345 -0.024 0.097 



76,437 



43,778 43,781 
310,850 



4 0.557 0.748 2.229 



4 -0.557 -0.739 2.229 



310,853 



4 -0.555 -0.560 2.221 



chr15 37,387,304 37,387,577 0.032 0.160 75,453 150,422 150,426 



5 0.443 0.507 2.213 



26 



SUPPLEMENTAL MATERIAL 



Berko et al. 



chr19 3,097,565 3,097,728 -0.028 0.111 109,601 220,187 220,190 



4 -0.549 -0.559 2.197 




chM 228,225,533 228,225,687 0.030 0.119 21,398 41,960 41,963 



15,530,606 15,530,870 -0.025 0.100 112,528 226,742 226,745 

0.024 




chr3 42,307,519 42,307,866 



0.097 151,540 303,761 303,764 

0.028 0.112 



59,825,552 159,825,761 
chr8 54,569,668 54,570,293 0.031 0.124 217,968 440,066 440,069 



4 0.549 0.571 2.1S 



4 -0.547 -0.653 



4 0.544 0.672 2.175 




4 0.536 0.569 2.143 



chr6 


29,795,501 


29,795,595 


0.034 


0.136 


188,094 


374,395 


374,398 


4 


0.525 


0.586 


2.101 




80,289,500 


80,289,701 


-0.027 


0.107 


104,929 


210,429 


210,432 


4 


-0.520 


-0.603 


2.081 


chr10 


70,321,770 


70,321,959 


0.031 


0.124 


28,175 


54,833 


54,836 


4 


0.519 


0.636 


2.075 


chr12 


117,797,056 














5 








chr19 


11,353,961 


11,354,240 


0.028 


0.113 


111,615 


224,525 


224,528 


4 


0.514 


0.652 


2.055 


26,624,865 


chr16 


56,696,748 


56,697,229 


0.028 


0.111 


88,038 


174,840 


174,843 


4 


0.511 


0.680 


2.045 




chr19 


1,387,394 


1,387,894 


-0.028 


0.110 


108,823 


218,581 


218,584 


4 


-0.509 


-0.687 


2.036 




chr6 


10,883,895 


10,884,314 


0.033 


0.130 


186,038 


369,196 


369,199 


4 


0.508 


0.553 


2.034 




56,565,286 


56,565,644 


0.027 


0.109 


100,960 


201,742 


201,745 






















chr20 


44,746,392 


44,747,006 


-0.028 


0.285 


140,615 


282,425 


282,434 


10 


-0.202 


-0.249 


2.018 


chr6 


146,755,301 


146,755,900 


0.027 


0.110 


196,115 


396,635 


396,638 


4 


0.500 


0.767 




























chM 


11,708,792 


11,709,271 


0.033 


0.131 


3,510 


7,279 


7,282 


4 


0.496 


0.660 


1.983 



27 



SUPPLEMENTAL MATERIAL 



Berko et al. 



chM 17,215,834 17,216,201 0.037 0.147 4,452 8,979 8,982 




272,015 
5,630 11,094 



chM 23,884,703 23,885,086 0.035 0.139 



11,097 



223,182 



chr8 145,008,957 145,009,406 -0.030 0.118 223,731 450,584 450,587 



4 0.494 0.646 1.975 
0.516 




4 0.482 0.511 1.930 




4 -0.481 -0.548 1.925 



chr20 61,732,467 61,732,608 0.027 0.106 141,955 285,308 285,311 



chr19 52,996,083 52,996,617 0.029 0.117 118,026 238,929 238,932 



21,322 41,802 41,805 




0.025 0.098 



chr3 50,487,955 50,488,230 



152,798 306,618 306,621 




chM 228,346,014 228,346,347 0.028 0.112 21,442 42,079 42,082 



4 0.476 0.561 1.904 




4 0.469 0.534 1.877 
0.588 



4 0.466 



12,712,424 112,712,795 0.025 0.101 65,563 130,953 130,956 4 0.458 0.570 



1.863 



4 0.464 0.620 1.856 



4 0.458 0.540 1.830 




chr22 46,449,498 46,449,821 -0.031 0.124 147,948 297,262 297,265 4 -0.453 -0.486 1.811 



1.787 



chr20 5,485,144 5,485,294 -0.030 0.181 138,494 277,554 277,559 



6 -0.298 -0.320 





56,791,576 


56,791,798 


-0.033 


0.132 


218,114 


440,360 


440,363 


4 


-0.446 


-0.476 


1.785 






















chr6 


31,148,404 


31,148,483 


-0.025 


0.124 


188,606 


377,487 


377,491 


5 


-0.349 


-0.373 


1.747 




chr17 


73,584,029 


73,584,111 


0.028 


0.141 


102,695 


205,201 


205,205 


5 


0.335 


0.522 


1.673 



28 



SUPPLEMENTAL MATERIAL 



Berko et al. 



chr15 


81,410,745 


81,411,066 


0.031 


0.126 


79,970 


158,935 


158,938 


4 


0.416 


0.507 


1.662 




chr6 


32,055,135 


32,055,316 


0.027 


0.107 


188,977 


380,461 


380,464 


4 


0.411 


0.627 


1.643 




6,486,598 


6,486,709 


-0.024 


0.098 


49,523 


99,109 


99,112 


4 


-0.411 






chr12 


53,358,946 


53,359,506 


0.030 


0.118 


53,044 


105,984 


105,987 


4 


0.410 


0.466 


1.640 




132,939,657 


132,939,992 


0.028 


0.114 


60,038 


120,334 


120,337 


4 








chr2 


66,659,348 


66,659,590 


0.025 


0.098 


124,840 


252,149 


252,152 


4 


0.404 


0.514 


1.615 




chr17 


43,716,423 


43,716,617 


-0.024 


0.098 


99,443 


198,608 


198,611 


4 


-0.393 


-0.451 


1.570 




496,069 


496,476 


0.035 


0.141 


172,645 


343,618 


343,621 


4 


0.392 


0.628 


1.568 


chr9 


96,715,687 


96,716,209 


0.025 


0.101 


226,639 


455,043 


455,046 


4 


0.386 


0.424 


1.543 


333,738 


chr6 


37,616,410 


37,616,803 


0.027 


0.107 


190,315 


385,855 


385,858 


4 


0.378 


0.556 


1.512 




chr16 


55,794,456 


55,794,910 


0.033 


0.164 


87,927 


174,589 


174,593 


5 


0.272 


0.363 


1.361 




chr22 


26,875,499 


26,875,652 


-0.025 


0.102 


145,844 


292,777 


292,780 


4 


-0.336 


-0.374 


1.342 




chr7 


4,901,337 


4,901,628 


0.026 


0.105 


200,602 


406,489 


406,492 


4 


0.330 


0.366 


1.318 




81,045,495 


81,045,863 


0.025 


0.101 


105,350 


211,445 


211,448 


4 


0.320 


0.590 




chr6 


292,329 


292,823 


-0.031 


0.157 


184,828 


366,824 


366,828 


5 


-0.250 


-0.260 


1.252 


chr7 


4,832,112 


4,832,359 


0.030 


0.119 


200,565 


406,401 


406,404 


4 


0.309 






chr5 


139,227,979 


139,228,242 


0.038 


0.192 


180,529 


358,423 


358,427 


5 


0.247 


0.311 


1.236 



29 



SUPPLEMENTAL MATERIAL 



Berko et al. 




The table shows the results of the dmrFind algorithm. DMR positions are shown in the 
chr/start/end co-ordinates, with probe indices and numbers represented by indexStart, indexEnd 
and nprobes, and area_raw the significance calculation following permutation analysis, allowing 
ranking of these DMRs by significance, as shown. 



30 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Weighted Gene Co-expression Network Analysis (WGCNA) 

We performed weighted gene co-expression network analysis (WGCNA) (Langfelder and 
Horvath 2008) to assess networks of co-methylated CpGs associated with ASD status. We 
used the CpG values output from the surrogate variable analysis (SVA) algorithm called in 
bump-hunting, since SVA corrected for all known technical artifacts. We used the WGCNA 
package in R and built an unsigned co-methylation network. Correlation matrices were raised to 
the power of 5, as calculated by the scale-free topology criterion on data subsets, and 
thresholds were set of minimum module connectivity (kME) of greater than 0.7, and minimum 
height for module merging of 0.1 (Voineagu et al. 201 1). We ran the algorithm with block sizes 
of 40,000 CpGs. 

We assessed module relevance to case/control status with a t-test (two tailed, unequal 
variance) of module eigengene values with case/control and gender categories. For 
relationship with the continuous variables of age and percent YRI and CEU ancestry, we used 
Pearson correlation coefficients and their Student asymptotic p-values. 

For analysis of methylation changes associated with ASD, we selected the 2 modules ("light 
green", and "dark olive green2") that showed significant correlation only with ASD status and not 
with any other covariate, to avoid confounding effects. 

A full list of the modules obtained and their Bonferroni-corrected p-values is provided in 
Supplemental Table S7. 

The genes associated with "light green" and "dark olive green 2" two ASD-associated modules 
are listed in Supplemental Tables S8 and S9. 

Gene ontology enrichment was performed using the same Cytoscape method described 
previously. For the light green module, gene ontology showed significant enrichment for 
negative regulation of smooth muscle cell migration (p = 1.34 X 10" 2 ), regulation of cell 
proliferation (p = 3.25 X 10" 2 ), negative regulation of cell migration (p = 3.27 X 10" 2 ), negative 
regulation of locomotion (p = 3.27 X 10" 2 ), negative regulation of cellular component movement 
(p = 3.27 X 10" 2 ), and negative regulation of metabolic process (p = 4.64 X 10" 2 ). All p-values 
shown are after FDR correction. 

It has been recently demonstrated that gene ontology analysis can be affected by unequal 
probe distribution in microarrays, with bias towards genes represented by greater numbers of 
probes (Geeleher et al. 2013). To address this concern, we calculated the number of probes 
corresponding to every gene by using annotations assigned by lllumina in the 450K array 
manifest. We used the number of probes per gene to generate a weight for every gene and 
recalculated enriched gene ontologies, using the R package GoSeq (Young et al. 2010). After 
accounting for unequal probe distribution, we obtained candidate ontology categories, but these 
did not remain significant after correcting the associated p-values for multiple testing. 

We also interrogated some of the modules most significantly associated with age and ancestry, 
the "dark turquoise" and "red" modules, respectively. Using the Lists2Networks software 
(Lachmann and Ma'ayan 2010), we found that the most significant gene ontology category 
enriched with genes from the age module was brain development (p=0.008). We then asked 
whether the ancestry-associated modules were enriched for CpGs with annotated SNPs on 
either the C or G. Out of 247 CpGs, 154 contained a reported variant (based on 1000 Genomes 
data), with 11 of those possessing variants on both the C and the G. This confirms the 



31 



SUPPLEMENTAL MATERIAL 



Berko et al. 



importance of both polymorphism on the CpG, as well as the presence of ancestry-associated 
methylation changes resulting from effects other than direct polymorphism. 



Protein -Protein Interaction (PPI) Analysis 

To assess the functional impact of our WGCNA ASD-associated co-methylated gene modules, 
we interrogated the genes' relevance in protein-protein interaction (PPI) networks. We 
combined the genes in the light green and dark olive green 2 modules with a list of previously 
curated known ASD risk genes, with the addition of exome sequencing candidates (KATNAL2 
and CHD8) (Neale et al. 2012). We used GeneMania (Warde-Farley et al. 2010) to build a PPI 
network of this combined list, using only data from physical protein interaction databases. We 
visualized this network in Cytoscape. 

To test the significance of these PPI connections, we performed Degree Aware Disease Gene 
Prioritization (DADA) (Erten et al. 2011). We used the ASD seed list mentioned previously, a 
combined candidate list of the genes related to the ASD-associated WGCNA CpGs, and the 
physical interaction database from the Human Protein Reference Database (HPRD) available 
through GeneMania. We evaluated if our genes were significantly enriched in ranking using the 
Mann-Whitney test. 

We also assessed whether the candidate genes related to ASD-associated CpG modules were 
functionally connected to known risk genes for Intellectual Disability (ID). We used a previously 
curated list of ID-implicated genes (Neale et al. 2012) and combined that with our 2 module 
genes to build a PPI network. Using the same parameters as above, we generated a network 
showing many functional connections, displayed in Supplemental Figure S8. 



32 



SUPPLEMENTAL MATERIAL 



Berko et al. 




Supplemental Figure S8: PPI Network of candidate genes and intellectual disability (ID) genes 

Purple nodes indicate genes from a curated list of previously implicated genes in ID. Light green and 
dark green nodes refer to genes from our WGCNA analysis of ASD-associated CpGs. Grey defines 
intermediate genes. 



33 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Supplemental Table S7: Bonferroni-corrected p-values of association between modules 
and known covariates 



MaHi i lo 
IVIUU U Its 


MO LJ 


UCl lUCl 




YRI % 
i r\i /o 


PFII % 

1 LJ /0 


IVI vcl IUcI UlUol \jL 












IVIC|JI U 1 1 IO 


— 


— 


— 


— 


— 


IVIcUUI d \L- 


— 


— 


— 


— 


— 


IVICll MolltfO 


— 


— 


— 


— 


— 


IVIfcM 1 ICUIUI 1 1UI \j\ II U 


— 


— 


— 


— 


— 


ivicuccuun ii\ 


— 


— 


— 


— 


— 


IVIfcMiyi 1 lold IcUl U t; 


— 


— 


— 


— 


— 


Monaloti irm inico 

IVItJjJaltJlUI LjUUIoc 


— 


— 


— 


— 


— 


IVIoUaloVIUIdloUO 


— 


— 


1 ^ftF-D? 


— 


— 




— 


— 


^ 1ftF-D4 


— 


— 


IVIfcM I ICU I U I I I |JU I |JIC I 


— 


— 


1 

I 


— 


— 


Mona\/aimA/hito 1 
IVIfcM IdvdJUWI IILC I 


— 


— 




— 


— 


ivlCUdl l\ot;dyi ccl IO 


— 


— 




— 


— 


IVlC|JII IIV+ 


— 


— 


1 RfiF-fl? 


— 


— 


ivici iui icyucvv 


— 


— 


c; nQF-D4 


— 


— 


IVICSdUUIcUl UWI I 


L/.L/iJ 


— 


^ fi7F-n^ 

O.KJ t ^ L/O 


— 


— 


ivit;uaioviuit;Lit;u^. 


1 
1 


— 




— 


— 


IVICll llollc 1 




— 




— 


— 


Mom anon ta A 

IVI CI 1 Idy CI 1 La4 




— 


^ DQF-17 


— 


— 


Mo\/olln\A/ 
ivicy diuvv 


ft ^df-da 


— 


ft 4DF-1R 


L.iJO^ L/Z. 


I .c/O^ L/Z. 


l\/lof irohrink'-t 

IVICl II CUI ILsl\0 


ft P^F-O? 


— 


0 -31 F-flft 
£-.0 I ^ L/O 




n Rft 

u.oo 


Monalov/inlotroH 1 

IVlCL/dlC VIUICll CU I 


0 ftciF-D? 


— 


I . / 1 u 


9 DfiF-D? 

Z..L/U^ L/Z. 


1 ^ftF-fl? 


IVIC|JI U I I \L- 


A 
1 


— 


H.UU^. L/C/ 






l\/lolinhtctoolhli io1 
ivitMiy 1 1 Loicciuiuc i 


1 
1 


— 


O.U 1 ^ 1 ij 


L/.iJC/ 


D 74 

L/. / *T 


Monav/ainvi/hito'? 
Ivlcl IdvdJUWI 


U.L 1 


— 


c acc.-i -i 

J.OU^ 1 1 


L/.OO 


L/.iJ 1 


IVIcUUI d IO 


1 


— 


A lorjc 

t. 1 1 ij 






l\/lohrn\A/n 

IVIcUl UWI 1 


Q Q^F-D^ 


— 


CZ.U 1 ^ 




I .H-L/^ L/Z. 


MockA/hli io1 
ivicor\yuiuc i 


1 
1 


— 








IVIgUUI al*+ 


0 -5-1 F-09 


— 




— 


— 


Monx/an 
ivicuy di i 


1 
1 


— 


I 


— 


— 


MotanA 
ivicidi i*+ 


Q9F-09 


— 


I .«J«J^ \JH 


— 


— 


MohnnowHowl 
ivici iui icyucvv i 


1 
1 


— 


^ 9ftF-D9 

JZOL UL 


— 


— 


Molinhtr , nral 
ivici ly 1 1 loui ai 


1 
1 


— 


1 19F-DA 

I . I \JH 


L/.Z.O 


— 


Meorange 


1 





1.64E-08 


9.54E-02 


0.280779828 


Meantiquewhite2 


1 




1 






Megreen 


1 




9.28E-03 






Meyellow4 


1 




4.17E-06 






Meblue 


1 




3.95E-06 






Meturquoise 


1 




8.25E-06 






Medarkseagreen4 


1 




1 






Medarkgreen 


1 




1 







34 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Mefloralwhite 


1 


0.92 


1 


6.57E-03 


2.44E-04 


Meorangered4 


1 


T~ 


1 


1 


I" 


Meblue4 


1 


1 


1 


1 


1 


Megreen4 


1 


1 


1 


1 


1 


Melightsteelblue 


1 


1 


1 


1 


1 


Meskyblue3 


1 


1 


1 


1 


1 


Mesalmon2 


1 


1 


1 


1 


1 


Meskyblue2 


1 


1 


1 


1 


1 


Mesienna3 


1 


1 


1 


1 


1 


Meorangered3 


1 


1 


1 


1 


1 


Mepluml 


1 


1 


1 


1 


1 


Meyellowgreen 


1 


1 


1 


1 


1 


Mesienna4 


1 


1 


1 


1 


1 


Mesteelblue 


1 


1 


1 


1 


1 


Meindianred4 


1 


1 


1 


1 


1 


Melavenderblushl 


1 


1 


1 


1 


1 


Medarkslateblue 


1 


1 


1 


1 


1 


Meskyblue4 


1 


1 


1 


1 


1 


Mesalmonl 


1 


1 


1 


1 


1 


Mewhite 


1 


1 


1 


1 


1 


Memidnightblue 


1 


1 


1 


1 


1 


Methistle2 


1 


1 


1 


1 


1 


Medarkviolet 












Melightgreen 


5.26E-02 


1 


1 


1 


1 


Meskyblue 




I" 


1 


I" 


I" 


Meyellow3 


1 


1 


1 


1 


1 


Medarkseagreen2 


1 


1 


1 


1 


1 


Meorangeredl 


1 


1 


1 


1 


1 


Meivory 


1 


1 


1 


1 


1 


Medarkolivegreen 


1 


1 


1 


1 


1 


Memagenta3 


1 


1 


1 


1 


1 


Melightcyan 


1 


1 


1 


4.30E-02 


0.59 


Melightpink4 


1 


1 


1 


1 


~ 


Meroyalblue 


1 


1 


1 


1 


0.10 


Mecoral 


1 


1 


1 


1 


1 


Medarkgrey 


1 


1 


1 


1 


1 


Memagenta 


1 


1 


1 


0.19 


0.64 


Melavenderblush3 


1 


1 


1 


I" 


~ 


Mefirebrick4 


1 


1 


1 


1 


1 


Methistle4 






0.28 






Medarkmagenta 




1 


1 


0.67 


0.24 


Medarkolivegreen2 


1.01E-02 




0.18 


0.42 


0.54 


Memediumpurple2 








1 


1 


Meviolet 








1 


1 



35 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Meblack 


1 


1 


1 


5.83E-02 


0.47 


Megreenyellow 


1 


1 


1 


3.28E-03 


3.61 E-02 


Memaroon 


1 


0.64 


1 


0.59 


0.82 


Mepurple 


0.51 


1.08E-42 


1 


T~ 


T~ 


Meantiquewhite4 


1 


T~ 


1 


1 


1 


Meblueviolet 


1 


1 


1 


1 


1 


Meplum4 


1 


1 


1 


1 


1 


Memediumpurple3 


1 


1 


1 


1 


1 


Menavajowhite 


1 


1 


1 


1 


1 


Mesalmon4 


1 


1 


1 


1 


1 


Meindianred3 


1 


1 


1 


1 


1 


Meblue2 


1 


1 


1 


1 


1 


MecoraM 


1 


1 


1 


1 


1 


Meplum 


1 


1 


1 


1 


1 


Mered 


1 


1 


1 


5.97E-03 


3.75E-02 


Mesalmon 


1 


1 


1 


6.59E-02 


0.39877902 


Melightblue4 


1 


1 


1 


T~ 


T~ 


Mebrown2 


0.91 


1 


1 


1 


1 


Medarkturquoise 


4.07E-02 


1 


8.32E-29 


1 


1 


Melightyellow 


3.68E-02 


1 


2.75E-20 


1 


1 


Medarkorange2 


1 


1 


6.18E-04 


1 


1 


Metan 


1 


1 


2.51 E-09 


1 


1 


Melightcyanl 


1 


1 


6.23E-02 


1 


1 


Mebisque4 


1 


1 


T~ 


1 


1 


Memediumpurple4 


1 


1 


1 


1 


1 


Meantiquewhitel 


1 


1 


1.42E-03 


1 


1 


Medarkolivegreen4 


0.38 


1 


1.13E-05 


0.99 


0.47 


Medarkorange 


2.48E-02 


1 


1.26E-07 


T~ 


~ 


Melightpink3 


1 


1 


T~ 


1 


1 


Medarkred 


1 


1 


1 


1 


1 


Megrey60 


1 


1 


1 


0.19 


1 


Mebrown4 


1 










Mepink 


1 






2.08E-02 


0.15 


Melightpink2 


1 










Methistle 


1 




1.75E-02 






Megrey 


6.01 E-05 




2.81 E-30 


4.72E-03 


3.28E-03 



36 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Supplemental Table S8: Genes in light green module associated with ASD 



Gene 


Chromosome 


Position 


rlODc 


Module Membership (MM) 


MM P-value 


1 II — C A 


1 


nocc a o 
93bb10 


cg1 5825501 


O.oO 


c- occ oo 

6.35E-22 


A 1 /O L/yf A A 

ALDH4A1 


1 


h oooozroo 


cg01 972394 


f\ 7C 

0.7b 


O OH C HO 

9.95E-19 


(jRIK3 


1 


37498270 


cg04364463 


0.72 


8.1bE-1b 


DPH2 


1 


44435457 


cg06790019 


0.70 


■1 /i 7r -i/i 

1 .47E-14 


EIF2B3 


1 


454521 bo 


cg05654404 


0.73 


O j|OC -( "7 

9.48E-17 


CD A Hi C A 1 

HAD54L 


1 


46712932 


eg 161 63324 


0.79 


H -IOC oo 

1 .1 8E-20 


/"» A ^,,-fA OO 


1 


H ^010<l 0~7~7 

1 12281877 


cg2 13042 11 


O.oo 


O A A d DA 

2Aa E-31 


&T7L 


1 


1 131b2073 


cg1 7861 791 


("l "7"7 

0.77 


O A "7CT HO 

2.17E-19 


LAP /.A 1 


1 


1 131b2073 


cg1 7861 791 


f\ ~7~7 

0.77 


O A 7C HO 

2.17E-19 


DENND2C 


1 


-i Hzronoezro 

1 1 5212b59 


eg 1350 1090 


O.oo 


o fine o~7 

8.00E-27 


L(jHo 


1 


2021b2209 


cg04351903 


O.oo 


5.85E-27 


n~rnh I A a 

PTPN14 




OH /1~70ZTO"7yl 

214725274 


eg 17024643 


f\ 70 

0.7o 


O t~\ A C OO 

8.04E-20 


MORN2 


o 


39103277 


cg07479988 


0.79 


H OOC OO 

1 .38E-20 


DHX57 


o 


oo h noo77 

39103277 


cg07479988 


0.79 


H OOC OO 

1 .38E-20 


A A/~* A TC 


o 
2 


1 34949571 


eg 16240751 


A 7C 

0.7o 


H C7T H o 

1 .b7E-18 


di ca 

RIF1 


o 


1 522bb33b 


eg 18320648 


O.oO 


O O "7 [~ OO 

9.07E-22 


HPb 


o 


oh noe7rti:n 

21 08o7059 


cg10181911 


0.7z 


o nrr He 

3.05E-1b 


IGFBP5 


o 


OH VITZTOOOO 

21 7559020 


cg03222971 


0.74 


O flflC H "7 

2.99E-17 


bCAP 


3 


4751o975 


cg01 374398 


r"i "7"7 

0.77 


4.27E-19 


TNNC1 


3 


HO A OOO A C 

5248934b 


cg01910272 


0.71 


O CCC H H 

3.55E-15 


NlbCH 


3 


HO A OOO A C 

5248934b 


cg01910272 


0.71 


O IZIZC A IZ 

3.55E-15 


A A A /"» IA 


3 


bb024b91 


eg 15764058 


0.7z 


b.bbE-1b 


hi A A t A /""W O 

NAALADL2 


3 


A ~T A finiTO A A 

1 74095241 


cg17014718 


-0.71 


H OCT H CT 

1 .8bE-15 


1 ii — o 

HE&1 


3 


1 93852754 


cg26348180 


A 7C 

0.7o 


o o H rr HO 

2.91 E-18 


li A A A /OD O 

MAN2B2 


4 


b577027 


cg09477292 


A 7C 

0.7o 


O O O C AO 

2.38E-18 




4 


1 1 81 0bb7 


ch.4.11419765R 


f\ 7C 

0.7o 


O O H C HO 

2.21 E-18 


GDjo 


4 


A HVOOZTOO 

1 5780522 


cg27473538 


A 70 

0.7o 


H O H AC 

1 .31 E-1b 


RBFJ 


4 


oeoooo/ie 

2b32324b 


cg06812693 


O.oz 


/I 71C OO 

4.72E-23 


CD A O A 

FRAbl 


4 


78977b90 


cg12573119 


f\ 7C 

0.7o 


H ~70 HO 

1 .78E-18 


CD A O A 

FRA&l 


4 


700704 oo 

789781 33 


cgO 1402409 


0.74 


i) njir H "7 

3.84E-17 


COQ2 


4 


842059b 1 


cg20161984 


f\ ~7~7 

0.77 


O f\flC H O 

2.09E-19 


LARP7 


4 


H H OCOCOO H 

1 1 3b2b831 


cg03148140 


0.71 


H O/IC A IZ 

1 .34E-15 


Ok A A /OH 


4 


14b29b778 


cg06295548 


-0.7z 


C 7Hr HC 

b.74E-1b 


L/A *f /"* D O 

HMGB2 


A 

4 


1 74254825 


eg 14705778 


r"i 70 

0.7o 


H nrtC H e 

1 .90E-1b 


Rbb/BF 


5 


b38021 84 


cg08709073 


A 70 

0.7o 


/i ocrr oo 

4.8bE-20 


FAM1 R1R 


o 


7Q7QO,OOQ 
1 c/ 1 OJOOo 


Cg0b50b598 


0 79 


q nop ifi 


PRDM6 


5 


122435202 


cg01 196322 


0.75 


5.20E-18 


NEUROG1 


5 


134871686 


eg 17772342 


0.72 


3.06E-16 


SPARC 


5 


151031796 


cg21530174 


0.72 


3.59E-16 


TSPAN17 


5 


176131088 


eg 1304 1389 


-0.79 


7.70E-21 


ZFP62 


5 


180287685 


cg24877792 


0.80 


2.45E-21 


LY86 


6 


6648823 


cg16101278 


-0.70 


6.61E-15 



37 



SUPPLEMENTAL MATERIAL 



Berko et al. 



BAT 4 


6 


31634141 


cg271 37280 


0.70 


5.78E-15 


CSNK2B 


6 


31634141 


cg271 37280 


0.70 


5.78E-15 


PPIL1 


6 


36842651 


cg04860157 


0.81 


7.24E-23 


CRIP3 


6 


43276478 


eg 18857062 


0.74 


2.91E-17 


LCA5 


6 


80246572 


eg 19879479 


0.84 


5.79E-26 


NDUFAF4 


6 


97345972 


cg04571327 


0.70 


9.24E-15 


C6orf174 


6 


127837548 


cg21986718 


0.75 


6.74E-18 


TNRC18 


7 


5463409 


cg23209537 


0.71 


1.93E-15 


IGFBP3 


7 


45961943 


eg 15208757 


0.76 


3.38E-18 


TYW1B 


7 


72298667 


cg21423973 


0.72 


4.62E-16 


SBDSP 


7 


72298667 


cg21423973 


0.72 


4.62E-16 


GATSL1 


7 


74379144 


eg 182 10722 


-0.69 


1.63E-14 


LHFPL3 


7 


103969483 


eg 17373058 


0.71 


2.38E-15 


LHFPL3 


7 


103970195 


eg 11826638 


0.70 


5.50E-15 


TSPAN12 


7 


120497479 


cg26340461 


0.72 


3.42E-16 


RHEB 


7 


151215566 


cg20495206 


0.80 


7.33E-22 


SORBS3 


8 


22423994 


cg21291431 


0.78 


1.07E-19 


TNFRSF10B 


8 


22926800 


cg26918957 


0.71 


2.51E-15 


DPYSL2 


8 


26434689 


eg 14967899 


0.72 


5.60E-16 


RRM2B 


8 


103251909 


eg 12374732 


0.70 


1.30E-14 


MYC 


8 


128748155 


cg25080152 


0.73 


1.62E-16 


PCSK5 


9 


78506874 


eg 135 12204 


0.79 


3.30E-21 


ZEB1 


10 


31608136 


cg25231972 


0.75 


1.02E-17 


FZD8 


10 


35930499 


cg00645593 


0.72 


3.60E-16 


ZNF503-AS1 


10 


77054788 


eg 15394763 


0.70 


5.06E-15 


FGF8 


10 


103535362 


eg 11 706469 


0.78 


1.15E-19 


ZNF215 


11 


6948101 


eg 10765857 


0.80 


9.54E-22 


GTF2H1 


11 


18343657 


cg1 1347316 


0.74 


4.99E-17 


HPS5 


11 


18343657 


cg1 1347316 


0.74 


4.99E-17 


PAX6 


11 


31831591 


eg 16822387 


0.72 


7.24E-16 


KBTBD4 


11 


47600851 


cg07996345 


0.73 


1.87E-16 


NDUFS3 


11 


47600851 


cg07996345 


0.73 


1.87E-16 


FADS1 


11 


61584442 


cg25837350 


0.76 


2.55E-18 


VEGFB 


11 


64002754 


eg 18872604 


0.70 


1.51E-14 


BAD 


11 


64052221 


cg25163015 


0.79 


6.58E-21 


GPR137 


11 


64052221 


cg25163015 


0.79 


6.58E-21 


KRTAP5-11 


11 


71340352 


cg22870994 


-0.81 


2.92E-22 


GAB2 


11 


78129288 


cg05492810 


0.77 


4.88E-19 


TMEM126B 


11 


85339628 


eg 12830327 


0.78 


2.11E-20 


DLG2 




85339628 


eg 12830327 


0.78 


2.11E-20 


MPZL3 




118123074 


cg27161463 


0.71 


2.29E-15 


CLEC4C 


12 


7904267 


eg 18348303 


-0.73 


7.50E-17 


DDX11 


12 


31226536 


eg 16864700 


0.75 


1.16E-17 


NELL2 


12 


45270304 


cg2 1846305 


0.74 


2.41E-17 



38 



SUPPLEMENTAL MATERIAL 



Berko et al. 



TRHDE 


12 


72667326 


cg09972192 


0.71 


2.35E-15 


LOC283392 


12 


72667326 


cg09972192 


0.71 


2.35E-15 


N4BP2L1 


13 


33002431 


cg06513149 


0.78 


2.70E-20 


ABCC4 


13 


95953574 


eg 15396799 


0.75 


1.21E-17 


ARHGEF7 


13 


111768023 


cg03925425 


0.77 


5.42E-19 


TUBGCP3 


13 


113263221 


cg24531141 


-0.76 


2.00E-18 


EFS 


14 


23834995 


cg022 13260 


0.75 


4.72E-18 


PSME1 


14 


24604912 


cg24054649 


0.71 


2.82E-15 


SNX6 


14 


35099518 


eg 13093793 


0.74 


6.03E-17 


SRP54 


14 


35451984 


cg04980793 


0.70 


5.47E-15 


FOXN3 


14 


90084672 


cg26386436 


0.74 


4.68E-17 


SLC25A29 


14 


100751514 


eg 10963 192 


0.75 


4.31E-18 


PACS2 


14 


105827276 


eg 15936935 


-0.85 


5.52E-27 


RTF1 


15 


41708917 


cg15581235 


0.78 


2.32E-20 


COR02B 


15 


68870836 


eg 12043722 


0.74 


1.96E-17 


CSPG4 


15 


75986363 


eg 14576802 


0.71 


2.67E-15 


ZNF200 


16 


3285262 


cg03530756 


0.74 


5.77E-17 


KLHDC4 


16 


87811505 


cg09562174 


0.76 


2.23E-18 


RNMTL1 


17 


685915 


cg27220681 


0.71 


1.85E-15 


GLOD4 


17 


685915 


cg27220681 


0.71 


1.85E-15 


NF1 


17 


29421732 


cg02726883 


0.75 


4.54E-18 


SLFN11 


17 


33700513 


cg05504685 


0.72 


5.32E-16 


COPZ2 


17 


46114574 


cg21 384971 


0.79 


3.32E-21 


MIR152 


17 


46114574 


cg21 384971 


0.79 


3.32E-21 


LOC146880 


17 


62777690 


eg 15869463 


0.76 


1.34E-18 


KPNA2 


17 


66031814 


eg 13777502 


0.72 


8.83E-16 


RECQL5 


17 


73629082 


cg042 19446 


0.74 


3.28E-17 


LOC643008 


17 


73629082 


cg042 19446 


0.74 


3.28E-17 


TIMP2 


17 


76921528 


eg 10466987 


0.80 


2.29E-21 


CCDC165 


18 


8707237 


eg 14095692 


0.77 


1.28E-19 


RAB27B 


18 


52495848 


cg05095774 


0.70 


6.23E-15 


CCDC102B 


18 


66382471 


eg 15552529 


0.78 


3.46E-20 


TMX3 


18 


66382471 


eg 15552529 


0.78 


3.46E-20 


SGTA 


19 


2761892 


cg04171554 


0.70 


6.43E-15 


ZNF77 


19 


2945000 


cg04335562 


0.80 


5.98E-22 


TNP02 


19 


12833533 


cg1 1788103 


0.71 


4.20E-15 


LYL1 


19 


13213716 


cg020 11446 


0.71 


1.22E-15 


UQCRFS1 


19 


29704262 


cg02905964 


0.74 


6.92E-17 


RTN2 


19 


45996498 


eg 198696 10 


0.74 


6.33E-17 


SMOX 


20 


4129314 


cg20604317 


0.73 


1.77E-16 


HMGN1 


21 


40720919 


cg01 338834 


0.73 


9.04E-17 


DGCR6 


22 


18893614 


cg07004357 


0.71 


1.42E-15 


LZTR1 


22 


21337040 


cg07047601 


0.81 


2.81 E-22 


PNPLA5 


22 


44287772 


cg24258125 


0.77 


1.49E-19 



39 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Supplemental Table S9: Genes in dark olive green module associated with ASD 



Gene 


Chromosome 


Position 


Drnhci 
rlODc 


Module Membership (MM) 


MM P-value 


FEXW 


1 


000700 <1 


cg16523185 


0.81 0337844 


1 .S2E-22 


r~* nio rtA i A 


1 


a c a neofifi 

46106399 


eg 17827803 


0.773502529 


A C A IT A O 

1 .61 E-19 


D^ / DO 

BCAR3 


1 


9424522b 


cg24937735 


0.740245636 


O A A A~7 

3.41 E-17 


A A 


1 


203753483 


eg 16020436 


f\ 7C O A EZ ~7Cf\~7 

0.768457607 


3.85E-19 


A A 




ooo~7eo/oo 

203763498 


cg02337583 


f\ CC\f\A OOOA IZ 

0.6991 3321 5 


9.15E-15 


/"» A A AtSli AT 

CAMKMT 


o 


4454551 0 


eg 1052 1567 


0.830944239 


A o -1 OA 

\.2a E-24 


FRKCE 


o 


45121488 


eg 16884847 


f\ 70 O EZ OO A r» A 

0.728533494 


h oer AC 

1 .86E-16 


RBM&l 


o 


-1C-1irOOO"7H 

151593271 


eg 13392885 


0.776332535 


o one oo 

9.80E-20 


ARFC2 


o 
2 


oa fin7finoo 

21 9079038 


eg 12884009 


0.797400033 


A OflC OA 

1 .89E-21 


DOM 

BhN 


o 
3 


A flCOOCOl 

49538532 


cg22881573 


f\ OAEZf\AA~7AA 

0.845041 71 1 


o oiznz oc 

3.35E-26 


r^ni \sa 

CFLX1 


4 


795538 


cg021 33849 


f\ ~7f\'~iEZf\EZAEZA 

0.702505454 


IZ OOe A IZ 

5.99E-15 


NFXL1 


4 


47854334 


cg27531236 


0.802345605 


"7 flflC OO 

7.00E-22 


C5ort27 


5 


on a ooo /i n 

951 92949 


cg26898099 


f\ ~7f\ A ^ A A HO 

0.70421452 


4.83E-15 




5 


1722531 12 


cg24581650 


0.683902398 


r- —Jizn A A 

5.75E-14 


t a no 

TAF2 


o 
6 


32803058 


cg00386460 


f\ 7707CTO H C\~7 

0.778753197 


e 07c oo 

6.37E-20 


/—» r~\ I a A A O 

CUL1 1A2 


o 
6 


ooaiza ooo 

33151008 


cg07457375 


f*\ 70fl/l 70'l IT7 

0.7294781 57 


a eoc -i c 

1 .63E-16 


WDR4b 


o 
6 


33254880 


cg23652681 


0.801 745369 


"7 O -1 OO 

7.91 E-22 


i a /r~\ a Ad 

WDR4b 


o 
6 


oooc / o ioo 

33254892 


eg 174 17645 


f\ ~7C A A OZ A~7l~\ 

0.761 165479 


-1 OflC -1 o 

1 .30E-18 


Rb>FH3 


o 
6 


1 59423743 


eg 15999887 


0.769393089 


O OO C -1 o 

3.28E-19 


A /"* /"") A T" A 

AGFAT4 


o 
6 


-i c -i n /i on 4 o 

151549519 


cg26780581 


0.790267496 


"7 C7r o-l 

7.57E-21 


A A A /"» /"» A 

MACC1 


7 


ooo >i o-i a iz 

20240145 


cg21710826 


0.698674927 


o c o rr -in 

9.68E-15 


CUX1 


7 


A CiA A 70 tZO IZ 

101478535 


cg08582182 


0.802229538 


"7 -1 7C OO 

7.1 7E-22 


/!_//"» A A A r~\ 

J HUMID 


7 


a ooono/ieyi 

1 39859454 


cg26800802 


0.71 341 8296 


1 .47E-15 


IS 1 A A r\A A C 


o 
8 


/I O EZ EZ~7 A Ol~\ 

48557420 


cg04247508 


0.792246082 


cr <or o-l 

5.1 8E-21 


IS 1 A A f\ A AC 

KIAA014o 


o 
8 


A OEZO~7 A A f\ 

48587440 


cg1 840451 3 


r\ ~7 EZO A O A tO A r\ 

0.75348461 9 


4.48E-18 


PABPC1 


o 
8 


a o a or\o AAA 

101802144 


cg08995449 


0.774123593 


A A IZTT A O 

1 .45E-19 


a o i — /no 

ARFIF2 


1 1 


tO A r\f\EZ~7 EZ 

5499575 


eg 17403702 


0.736923523 


5.57E-17 


A A 1 O 

MICAL2 


■1 A 
1 1 


A oooonvo 

12222570 


cg09371112 


0.722898685 


yl fioc -1 C 

4.08E-16 


CNIH2 


1 1 


ccf\A o~7no 

55048759 


cg061 55341 


or\izr\or\or\o 

0.805030303 


A flOC OO 

4.03E-22 


A A A A At O 

MAML2 


1 1 


C\EZOOC\A EZ A 

95889454 


eg 1552 1790 


f\ 70-1 7(*lC770 

0.781 796778 


O 0~7 1 — OO 

3.67E-20 


KRT79 


4 O 


53228551 


cg1 31 19928 


r\ ~7C A A f\IZ A ~ro 

0.7641 051 73 


O flflC A O 

8.00E-19 


NCOR2 


-1 o 


A OnOOO~7/l A 

125030744 


cg07243762 


0.722336993 


a A A rr -1 c 

4.41 E-16 


a tioa a a 

ATP1 1 A 


1 3 


A A OO A OOf\A 

1 1 3348391 


cg09507215 


0.7301 82655 


-1 ^ "7 [ — AC 

1 .47E-16 


L//~\A AC7 

FtUMEZ 


A A 

14 


0074 A Of\ A 

23744304 


cg04420752 


0.747121013 


-1 HC A "7 

1 .21 E-17 


A Dl—I/^ a niz 

ARHGAF5 


14 


32597733 


cg07564690 


f\ OOAAIZ~7ACA 

0.831457161 


-i n7C oz 

1 .07E-24 


r\Divi s.u 


14 

I H 


1 00/.H/.00 


eg 04422024 


D 7^71 W\A7 


i.tOC I o 


c15orf50 


15 


70147094 


eg 16497945 


0.763106126 


9.44E-19 


MLYCD 


16 


83945978 


cg01 984843 


0.776011546 


1.04E-19 


ZNF426 


19 


9645566 


cg2 1474288 


0.749726963 


8.08E-18 


PSMD8 


19 


38869646 


eg 11607742 


0.801710907 


7.97E-22 


HM13 


20 


30126382 


cg04306926 


0.728729016 


1.81E-16 


PPM1F 


22 


22290866 


cg01 800253 


0.710229687 


2.23E-15 



40 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Code 

Using R version 2.15.0 

#LOADING IN RAW IDAT FILES 
setwd(" INSERT USER DIRECTORY") 

man<-read.csv("HumanMethylation450_15017482_v.1.1.csv", skip=7, header=TRUE, 
stringsAsFactors=FALSE) 

require(minfi) 

library(llluminaHumanMethylation450kmanifest) 

targets <- read.450k.sheet("USER DIRECTORY", pattern="pheno96.csv") 
Basename<-apply(targets, 1 , function(x){ 

paste("USER DIRECTORY", x[13], 7", x[13], x[12], sep="") 

}) 

targets<-data.frame(targets[,1 :13], Basename) 

RGset <- read.450k.exp(base = "USER DIRECTORY", targets = targets) 

##Extract bisulphite control values from getProbelnfo from RGset 

#CALCULATE DETECTION PVALS TO REMOVE POORLY PERFORMING PROBES 
sampleNames(RGset) = targets$Sample_Name 
pVals = detection P(RGset) 

M.set<-preprocessRaw(RGset) #this is what the table of signal intensities is extracted from 

#REMOVE SEX CHROMOSOMES BASED ON ILLUMINA MANIFEST ANNOTATION 

cpgs<-getManifestlnfo(M.set, type = cflocusNames")) 

Y<-which(man[,12]=="Y") 

X<-which(man[,12]=="X") 

cpgsremove<-c(man[X,1 ], man[Y,1 ]) 

posocpgs %in% cpgsremove 

nomatch<-which(pos=="FALSE") 

M . set2<-M .set[nomatch ,] 

#NORMALIZE AUTOSOME DATA WITH SWAN 
M.setswan<-preprocessSWAN(RGset, mSet=M.set2) 

#REMOVE PROBES WITH DETECTION P-VAL > 0.01 
pValsauto<-pVals[nomatch,] 

msetswanfilt<-M.setswan[rowSums(pValsauto)<=0.01 ,] 

#EXTRACT M-VALUES FROM NORMALIZED, FILTERED DATASET 
mvals<-getM(msetswanfilt, type="lllumina") 
write.table(mvals, file="m_values.txt", sep="\t", quote=FALSE) 

#CORRECT FOR BATCH EFFECT 
library(sva) 

pd<-pData(msetswanfilt) 



41 



SUPPLEMENTAL MATERIAL 



Berko et al. 



batch<-pd$Slide 

mod<-model.matrix(~as.factor(Gender) + age. 3. 201 2 + as.factor(Sample_Group), data=pd) 
norm_m<-ComBat(mvals, batch=batch, mod=mod, numCovs=3, 
par.prior=TRUE,prior.plots=FALSE) 

write.table(norm_m, file="norm_m.txt", sep="\t", quote=FALSE) 

##This file contains 96 individuals. Three were then removed from subsequent DMR analysis 
(one had a changed diagnosis, and two did not have genotyping data as their genotyping arrays 
failed quality controls. For simplicity, the phenotype file with the 93 retained individuals is 
provided for the subsequent analysis. 

#BUMPHUNTING ANALYSIS 

setwd("USER DIRECTORY") 

mvak-read.delimCnorm_m.txt", header=TRUE, stringsAsFactors=FALSE) 
cpgnames<-row.names(mval) 

manifest<-read.csv("HumanMethylation450_15017482_v.1.1 .csv", skip=7, sep=",", 

stringsAsFactors=FALSE) 

temp<-match(cpgnames, manifest[,1 ]) 

manifest2<-manifest[temp,] 

pheno<-read.csv("pheno93.csv", stringsAsFactors=FALSE) 

names<-colnames(mval) 
keep<-match(pheno[,1], names) 
mval2<-mval[,keep] 

chr<-manifest2[,12] 
chr<-paste("chr", chr, sep="") 
position<-as.numeric(manifest2[,13]) 
library(charm) 

#this part reorders each chromosome by position and then rbinds them all together, and then 

reorders the matrix of m-values by that order too) 

a<-cbind(chr, position, cpgnames) 

chrs<-levels(as.factor(chr)) 

finak-c() 

for (j in 1 :length(chrs)){ 
b<-subset(a, a[,1]==chrs[j]) 
ord<-order(as.numeric((b[,2]))) 
c<-b[ord,] 

finak-rbind(final, c) 

} 

exclude<-which(final[,1]=="chr") 
final2<-final[-exclude,] 
d<-match(final2[,3], row.names(mval)) 
mval3<-mval2[d,] 



42 



SUPPLEMENTAL MATERIAL 



Berko et al. 



position<-as.numeric(final2[,2]) 
chr<-final2[,1] 

pns<-clusterMaker(chr, position, order. it=TRUE, maxGap=300) 

mod<-model.matrix(~as.factor(pheno[,8]) + (pheno[,9]) + (pheno[,10]) + pheno[,11] + 
as.factor(Sample_Group), data=pheno) 

mod0<-model.matrix(~as.factor(pheno[,8]) + (pheno[,9]) + (pheno[,10]) + pheno[,11], 
data=pheno) 

mval4<-data. matrix(mval3) 

test<-dmrFind(logitp=mval4, mod=mod, modO=modO, coeff=6, pns=pns, chr=chr, pos=position) 

## TYPE test$dmrs for dmrs 

##WGCNA CODE IS FREELY AVAILABLE AT 

http://labs.genetics.ucla.edu/horvath/CoexpressionNetwork/Rpackages/WGCNA/. Parameters 
used in the application of this package are explained in detail in the Supplemental Materials. 



43 



SUPPLEMENTAL MATERIAL 



Berko et al. 



Supplemental References 

Aryee MJ, Wu Z, Ladd-Acosta C, Herb B, Feinberg AP, Yegnasubramanian S, Irizarry RA. 
2011. Accurate genome-scale percentage DNA methylation estimates from microarray 
data. Biostatistics 12(2): 197-210. 

Basu SN, Kollu R, Banerjee-Basu S. 2009. AutDB: a gene reference resource for autism 
research. Nucleic Acids Res 37 (Database issue): D832-836. 

Colella S, Yau C, Taylor JM, Mirza G, Butler H, Clouston P, Bassett AS, Seller A, Holmes CC, 
Ragoussis J. 2007. QuantiSNP: an Objective Bayes Hidden-Markov Model to detect and 
accurately map copy number variation using SNP genotyping data. Nucleic Acids Res 
35(6): 2013-2025. 

Conlin LK, Thiel BD, Bonnemann CG, Medne L, Ernst LM, Zackai EH, Deardorff MA, Krantz ID, 
Hakonarson H, Spinner NB. 2010. Mechanisms of mosaicism, chimerism and 
uniparental disomy identified by single nucleotide polymorphism array analysis. Hum Mol 
GeneM9(7): 1263-1275. 

Danecek P, Auton A, Abecasis G, Albers CA, Banks E, DePristo MA, Handsaker RE, Lunter G, 
Marth GT, Sherry ST et al. 2011. The variant call format and VCFtools. Bioinformatics 
27(15): 2156-2158. 

Dedeurwaerder S, Defrance M, Calonne E, Denis H, Sotiriou C, Fuks F. 2011. Evaluation of the 
Infinium Methylation 450K technology. Epigenomics 3(6): 771-784. 

Erten S, Bebek G, Ewing RM, Koyuturk M. 201 1. DADA: Degree-Aware Algorithms for Network- 
Based Disease Gene Prioritization. BioData Min 4: 19. 

Geeleher P, Hartnett L, Egan LJ, Golden A, Raja AN RA, Seoighe C. 2013. Gene-set analysis is 
severely biased when applied to genome-wide methylation data. Bioinformatics 29(15): 
1851-1857. 

Gonzalez JR, Rodriguez-Santiago B, Caceres A, Pique-Regi R, Rothman N, Chanock SJ, 
Armengol L, Perez-Jurado LA. 2011. A fast and accurate method to detect allelic 
genomic imbalances underlying mosaic rearrangements using SNP array data. BMC 
Bioinformatics 12: 166. 

Hansen KD, Aryee M. minfi: Analyze lllumina's 450k methylation arrays., Vol version 1.4.0, p. R 
package. BioConductor. 

Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, Haussler D. 2002. The 
human genome browser at UCSC. Genome Res 12(6): 996-1006. 

Lachmann A, Ma'ayan A. 2010. Lists2Networks: integrated analysis of gene/protein lists. BMC 
Bioinformatics 11: 87. 

Langfelder P, Horvath S. 2008. WGCNA: an R package for weighted correlation network 
analysis. BMC Bioinformatics 9: 559. 

Leek JT, E. JW, Parker HS, E. JA, Storey JD. sva: Surrogate Variable Analysis., Vol version 
3.2.1, p. R package. BioConductor. 

Maksimovic J, Gordon L, Oshlack A. 2012. SWAN: Subset-quantile within array normalization 
for illumina infinium HumanMethylation450 BeadChips. Genome Biol 13(6): R44. 

Neale BM, Kou Y, Liu L, Ma'ayan A, Samocha KE, Sabo A, Lin CF, Stevens C, Wang LS, 
Makarov V et al. 2012. Patterns and rates of exonic de novo mutations in autism 
spectrum disorders. Nature 485(7397): 242-245. 



44 



SUPPLEMENTAL MATERIAL Berko et al. 



Price AL, Tandon A, Patterson N, Barnes KC, Rafaels N, Ruczinski I, Beaty TH, Mathias R, 
Reich D, Myers S. 2009. Sensitive detection of chromosomal segments of distinct 
ancestry in admixed populations. PLoS Genet 5(6): e1000519. 

Rebhan M, Chalifa-Caspi V, Prilusky J, Lancet D. 1997. GeneCards: integrating information 
about genes, proteins and diseases. Trends Genet 13(4): 163. 

Sanders SJ, Ercan-Sencicek AG, Hus V, Luo R, Murtha MT, Moreno-De-Luca D, Chu SH, 
Moreau MP, Gupta AR, Thomson SA et al. 2011. Multiple recurrent de novo CNVs, 
including duplications of the 7q11.23 Williams syndrome region, are strongly associated 
with autism. Neuron 70(5): 863-885. 

Voineagu I, Wang X, Johnston P, Lowe JK, Tian Y, Horvath S, Mill J, Cantor RM, Blencowe BJ, 
Geschwind DH. 2011. Transcriptomic analysis of autistic brain reveals convergent 
molecular pathology. Nature 474(7351): 380-384. 

Warde-Farley D, Donaldson SL, Comes O, Zuberi K, Badrawi R, Chao P, Franz M, Grouios C, 
Kazi F, Lopes CT et al. 2010. The GeneMANIA prediction server: biological network 
integration for gene prioritization and predicting gene function. Nucleic Acids Res 38 
(Web Server issue): W214-220. 

Xi Y, Li W. 2009. BSMAP: whole genome bisulfite sequence MAPping program. BMC 
Bioinformatics 10: 232. 

Xu LM, Li JR, Huang Y, Zhao M, Tang X, Wei L. 2012. AutismKB: an evidence-based 
knowledgebase of autism genetics. Nucleic Acids Res 40 (Database issue): D1016- 
1022. 

Young MD, Wakefield MJ, Smyth GK, Oshlack A. 2010. Gene ontology analysis for RNA-seq: 
accounting for selection bias. Genome Biol 11(2): R14. 



45