Reconstruction and Analysis of Transcriptional Regulatory ... · (TReNA) Sequence Motifs JASPAR...
9
Seth A. Ament Institute for Systems Biology Seattle,Washington Reconstruction and Analysis of Transcriptional Regulatory Networks with TReNA
Transcript of Reconstruction and Analysis of Transcriptional Regulatory ... · (TReNA) Sequence Motifs JASPAR...
Seth A. AmentInstitute for Systems BiologySeattle, Washington
Reconstruction and Analysis of Transcriptional Regulatory Networks with TReNA
Genes influence phenotypes through a network of networks
DNA
NeuronalNetwork
BrainConnectivity
Network
SocialNetwork
Individual
MolecularNetwork
Transcriptional Regulatory Network Analysis (TReNA)
Sequence MotifsJASPAR
DNase footprintsENCODE
Epigenomic StatesROADMAP/FANTOM
Evolutionary ConservationphastCons
Tissue-SpecificTF Binding Sites
FootprintFinder
Software Availability:https://github.com/PriceLab/TReNA
Transcriptional Regulatory Network Analysis (TReNA)
Sequence MotifsJASPAR
DNase footprintsENCODE
Epigenomic StatesROADMAP/FANTOM
Evolutionary ConservationphastCons
Tissue-SpecificTF Binding Sites
FootprintFinderTissue-Specific
Transcriptome ProfilesGTEx/GEO
Tissue-SpecificTranscriptional Regulatory Network
(TF-Target Gene Interactions)
fitTRN
Software Availability:https://github.com/PriceLab/TReNA
Combining diverse annotations improves prediction of TF binding sites
Specificity
Sens
itivi
ty
0.0
0.2
0.4
0.6
0.8
1.0
1.0 0.8 0.6 0.4 0.2 0.0
Specificity
Sens
itivi
ty
0.0
0.2
0.4
0.6
0.8
1.0
1.0 0.8 0.6 0.4 0.2 0.0
Specificity
Sens
itivi
ty
0.0
0.2
0.4
0.6
0.8
1.0
1.0 0.8 0.6 0.4 0.2 0.0
FIMO + Wellington + ChromHMM + phastConsFIMO p−valueWellington p−value
TRUE/FALSE classes:USF1 DNase footprints with/without USF1 ChIP-seqpeaks
All USF1 footprints: 79% sensitivity31% specificity
USF1 footprints with modeled probability > 50%:55% sensitivity70% specificity
Pea
rson C
orr
elation
LASS
O
Ran
dom Fore
st
Mut
ual Inform
ation
ARACNe
TFB
S co
unt (0
-1 kb)
TFB
S co
unt (1
-10 kb)
TFB
S co
unt (1
0-1
00 kb)
TFB
S co
unt (1
00kb
-1Mb)
TFB
S z-sc
ore
(0-1
kb)
TFB
S z-sc
ore
(1-1
0 kb)
TFB
S z-sc
ore
(10-1
00 kb)
TFB
S z-sc
ore
(100kb
-1Mb)
Combin
ed C
o−Ex
pres
sion
Combined
TFB
S
Ense
mble
0.0
0.2
0.4
0.6
0.8
1.0
AU
RO
C
******
Co-Expression TF Binding Sites Ensemble
shRNA-microarray profiling of 25 TFs in lymohoblasts
Combining TF binding sites and gene co-expression improves prediction of TFs’ functional target genes
Expression patterns of TFs accurately predict the expression patterns of thousands of genes in each tissue
Training SetTest Set (5-fold CV)
Vari
ance
Exp
lain
ed (
R2 )
Genes Ranked by R2
Prediction of brain gene expression with fitTRN
Genome-scale TRN model for the human brainInput data• 4.6M predicted human brain TFBSs
• 2,756 gene expression profiles
from the Allen Brain AtlasSummary Statistics
• 745 TFs• 11,093 target genes
• 201,218 interactions
(Ament et al, unpublished)(Ament et al., in prep.)
BD SCZ MDD
SOX3SMAD1OTX1
FOXN3FOXO4PPARA
POU3F2SREBF1
IRF9NPAS3RUNX1POU3F4TEAD1
HMBOX1FOXN2PRRX1FOXO1FOXJ1SOX2SOX9
p-value1e−21e−61e−20
PGC_BIP32b_mds7a.0.chr2
Chromosome 2 (kb)57900 58100 58300
0
2
4
6
8
Obs
erve
d (−
logP
)
0
20
40
Rec
ombi
natio
n ra
te (
cM/M
b)
p = 5.0e−08
●
●
●●
●
●
● ●
●
●●
●
●
●
●
●
●
●
●
●●●
●
●●●
●
●
●
●
●
●●●
●
●
●
●
●● ●●
●
●
●
●
●
●
●
●
●
●
●
●
●●
● ● ●●●
●
●
●●
●
●
●● ●
●
●
●●
●●
●
●
●
●
●
●●● ●
●
●
●
● ●●●
●● ●●
●●●
●●
●
●
●●●●● ● ●
●
●
●
●
●
●
●●
●●
●
●
●
●
●
●
●
● ●
●
●
●
●●
●
●
●
●
● ●
●
●
●
●
●
●
●●
●●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
● ●
●
●
●
●
●
● ●
●
●●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●●
●
●●
●
●
●
● ●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●●
● ●
●●
●
●
●
●
●
●●
● ●
●●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●●
●
●
●
● ●
●
●
●
●
●●
●
● ●
●
● ●●
●
●
● ●●
●
●
●
●
●
●
●●●
●
●
●
●
●● ●
●
●
●●
●●
● ●
● ●
●●●
●
●
●●
●
●●
●●
●
● ●
●
●
●
●
●
●
●
●
●
●●
● ●
●
●
●
●
●
●
●● ●
●
●
●
●
●
●●
●●
● ●●
●
●●
●
●
●
●
●
●
●
●●
●●
●●
●
●
●●
●
●
●
●
●
●
●
●●
●●
●
●● ●●●
●
●
●
●
●●●
●
●
●
●
●●
●
●
●
●
●
●●
●
●
●●
●
● ●● ●●●●
●
●
●
●
●
●
●●
●
●
●
●●
●
● ●●●
●
●
●
●●●
●
●
●● ●● ● ●● ●
●
●
●
●
●
●
● ●●
●
●
●
●
● ●
●
●
●●
●
●●●
●
●
●
● ●
●●
●
●
●
●
●
●
●
●
●
●●
●
●
● ●● ●
●
●
●
●
●
●
●
●
●●
● ●
●
●
●
●
●
●
●
●
●●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●●
●
●
●
●
●
●●
●
●
●
● ●●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●●
● ●
●
●
●●●
●
●
●●●
●
●
● ●●
●
●
●●
●●●
●
●● ●
●
●
●
●
●
●
●
●
●●
●
●
●●●
●
●
●
●●
● ●●
●
●
●
●
●
●●
●
●
●
●●
●
● ●
●●
●
●
●
●●
●
●
●
●
●
●
●● ●
●
●
●
●●
●
●
●●
●
● ●
●
●
●●
●
●
●
●●
● ●●
●
●
●
●
● ●
●
●
●●
●●
●
●
●
●
●
●
●●
●
●
●●
●
●●
●●
●●
●
●●●
●
●●
●
●
●
●●
●
●
●
●
●
●
●
●●
● ●
●●
●●
●
● ●
●
●
●
●
●●
●
●●
●
●
●●
●●
●
●
●
●
●
●
●
●
●●
●●
●●
●
●
●●
●
●
●
●
●
●
●
●
●
● ●
●
●
●●
●
●
● ●
●
●
●● ●●
●
● ●
●●
●
●
●
●●
●
●
● ●
●
●
●
●
●●●
●
●
●
●
●
●
●● ●
●
●●
●
●
●
●
●
●
●
●
● ● ●●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
● ●
●●●
●
●
●
●
●
●●
●
●
●
● ●
●●
●●
●
● ●●
●
●
●
●●●●
●●
●
●
●
●
●
●
●
●
●●
●
●
●●
●
● ●● ●●
●
●
●●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●● ●
●
●●
●
●
●
●●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
● ●
●●●
●
●●
●
●●
●
●
●
●●● ●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●● ●
●
●
●
●
●●
●
●
●
●
●
●●● ●
●
●
●
●●
●
●
●
● ●
●
●
●
●
●
●
●
●
●●
●
●
●
● ●
●●
●
●
●●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●●
●
●
●
●●
●
●
●
●
●
●
●
●
●●
●● ●
●●●
●
●
●
●
●● ●
●
●
●
●
●
● ●
●
●●●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●●●
●
●
●
● ●●
●
●●●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●● ●
● ●●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●
●
● ●
●
●
●
●
●
●
●●
●● ●● ● ●
●
●
●
●●
●
●
●
●
●
● ● ●
●
●
●●●
● ●●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●●●
●
●
●●●
●
●
●
●
●
●
●
●
●
●●
●
●
●●●
●
●
●
●
●
●
●
●
●
●
●
●● ●
●
●● ●
●
●
●
●
●
●
●● ● ●
●
●
●●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●●
● ●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●●●
●●
●
●
●●
●
●
●●
●
●
●
●● ●
●
●
●
●
●
●
●
●
●●
●
●●
●●●
●
●●
●
●●
●●
●
●
●
●
●
●
●
●
●●
●●
●
●
●
●
●
●
●
●
●●
●●
●●
●
● ●
●
●● ●
●●
●
●
●
●
●
●●
●
● ●●
● ●
●
●
●●●
●
●●
●
●
●
●● ●
●●
●
●
●
●●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●●
●●
●●
●
●
●
●●
●
● ●●●
● ●
●
●
●
●
● ●
●
●
●●
●
●
●
●●
● ●● ●
●
●
●
●
●
●
●
●
●
●
● ●●
●
●●
●
●
●
●
●
●
●
● ●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●
●●
●
●
●
●
●
●●
●
●
●
●
●
●
●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●●
●
●
●
●
●●
●
●
●●
●
●
●
1 0.8 0.6 0.4 0.2 0.1
1 . rs13026414 : Epilepsy(2E−9)
11
2 . rs2312147 : Schizophrenia(3E−7)
22
3 . rs2717068 : Epilepsy(4E−7)
33
snp / p / or / maf / info / directionsa . rs57681866 / 5.00e−08 / 0.85 / 0.06 / 0.969 / 5−27−0
aa
VRK2
VRK2
VRK2
rs13384219
PGC2-BD GWAS (Stahl et al., in prep)
0
2
4
6
8
-log 1
0(p-
valu
e)
BD SCZ MDD
SOX3SMAD1OTX1
FOXN3FOXO4PPARA
POU3F2SREBF1
IRF9NPAS3RUNX1POU3F4TEAD1
HMBOX1FOXN2PRRX1FOXO1FOXJ1SOX2SOX9
p-value1e−21e−61e−20
Risk-associated SNP in a predicted POU3F2 binding site
0.0
0.5
1.0
1.5
Luci
fera
se A
ctiv
ity
A G
0 0.25 1 4 0 0.25 1 4POU3F2 (ng)
rs13384219
TReNA reveals master regulator TFs and regulatory genetic variants in psychiatric disordersMaster Regulator TFs
rs13384219 allele and POU3F2 expression influence the activity of the VRK2 promoter
