196539

56902

278382

7482

2831

2951

12718

6253

33899

2284423 2709

15905

43617

1656

CircExplorer2 CIRI2

Find_circ CircRNA_finder

A

B

Rs = 0.08, p = 0.7

BLCA

BRCA

CESC

COAD

DLBC

ESCA

HNSC

KIRC

LCLL

LGG

LIHC

LUSC

MM

OVPAAD

SARC

SKCM

STAD

THCA

150

160

170

50 60 70 80 90 100Average lineage specific circRNAs per sample

Aver

age

tota

l map

pabl

e re

ads

per s

ampl

e (M

illion

s)

C

CIRI2 CircExplorer2

CircRNA_finder Find_circ

LUSC

LCLL

LGG

DLB

CBR

CA

CO

ADST

ADSK

CM OV

SAR

CPA

ADLI

HC

ESC

AH

NSC

KIR

CBL

CA

MM

CES

CTH

CA

UC

ECPR

ADM

ESO

LUSC

LCLL

LGG

DLB

CBR

CA

CO

ADST

ADSK

CM OV

SAR

CPA

ADLI

HC

ESC

AH

NSC

KIR

CBL

CA

MM

CES

CTH

CA

UC

ECPR

ADM

ESO

0

25000

50000

75000

100000

125000

0

25000

50000

75000

100000

125000

# ci

rcR

NAs

Figure S1

Figure S1. Expression landscape of circRNAs across cancer cell lines. (A) Numbers of circRNAs detected by four circRNA identification algorithms across cancer cell lines (Cir-cRNA_finder, Find_circ, CIRI2, CircExplorer2). (B) Venn diagram depicting the overlap among the four algo-rithms. (C) No significant correlation (Rs = 0.08, p = 0.7) between average cancer lineage specific circRNA and average mappable reads across cancer lineages.

Rs = 0.37 , p < 2.2 x 10-16

12

14

16

−2 −1 0 1 2

EMT Score

Tota

l circ

RN

As re

ads

p = 5.5 x 10-16

12

14

16

18

low highEMT Score

log(

No.

of c

ircR

NA

read

s)

A B

Figure S2

0

10000

20000

30000

40000

50000

control TGF−beta treated

Adju

sted

tota

l BS

read

s

HMLEP = 0.042

C

Figure S2. Potential regulation of EMT on biogenesis of circRNAs. (A) Significant correlation between

EMT score and total circRNA backsplicing reads across cancer cell lines. (B) Increased number of

circRNA backsplicing reads detected from high EMT Score cell lines compared to low EMT Score cell

lines. (C) Increased backsplicing reads upon TGF-β treatment in human HMLE cells.

1

10

100

1000

10000

num

ber o

f RBP

bin

ding

pea

ks

19.4%

43.7%

0

10

20

30

40

50

Backg

round

gene

s

Clinica

lly Acti

onab

le ge

nes

Perc

enta

ge o

f gen

es g

ener

ate

circ

RN

AsA B

p < 2.2 ×10-16 p < 2.2 ×10-16

Figure S3

Backg

round

gene

s

Clinica

lly Acti

onab

le ge

nes

Figure S3. Enrichment of circRNAs in clinically actionable genes.(A) Enrichment of circRNAs from clinically actionable genes compared with that of background genes (Pearson’s Chi-squared test, p < 2.2×10-16). (B) Significantly higher number of RBP binding peaks of clinically actionable genes compared to that of back-ground genes (Wilcoxon rank-sum test, p < 2.2 × 10-16).

Figure S4

D

circMYCN

circMYCL

circMYC

cancer cell lines

22.2%

2.4%

0.2%

******

RNase R

500bp

400bp

300bp

200bp

100bp

500bp400bp

300bp

200bp

100bp

circMYC- + - +- - + +

No reversetranscriptase

linea

r MYC

circ

MYC

A

B C

circMYC-

9

10

11

12

13

14

15

●

●●●●

●

●●●●●

●●

●●●●

●●●●

●●●●●●●●●●●●●●●●

●●

●●

●

●

●

●●●●●●●●●●●●●●●●

●●●●●●●●●●●●● ●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●● ●●●●● ●●●●●●●●●● ●●●●●●●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●● ●●●●●●●●●● ●●● ●●●●●●● ●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●● ●●●●●●●●●●●●● ●●●●●●●●●● ●●●●●●●●●●●●● ●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●● ●●●●● ●●●●●●●●● ●●●●●●●●●● ●●●● ●●●●●●●●●●●●●●●●●● ●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●● ●●●●● ●●●●●● ●●●●● ●●●●●●●●●●●●●●●●●●●●●●●● ●● ●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●

●●

● ●

●●●●

●

circMYC+

FDR = 0.00048

Vorinostat (CTRP)

0.4

0.5

0.6

0.7

0.8

0.9

1.0

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●●

●

●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●●

●●

●

●

●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●●

●

●

●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●●

●

●

●

●●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●●

●

●

●●

●●

●

●

●

●

circMYC-circMYC+

Dru

g R

espo

nse

Vorinostat (GDSC)

FDR = 0.0032

MYC

GAPDH

Figure S4. Characterization of functional effects and drug response of circMYC in cancer. (A) CircRNA expression of MYC, MYC human paralogous genes (MYCN and MYCL) across cancer cell lines. (B) PCR of linear MYC and circMYC with or without treatment of RNase R. (C) Overexpression of MYC protein level in MDA-MB-231 cells transfected with circMYC. (D) Comparison of Vorinostat response between circMYC positive and negative cells from GDSC and CTRP dataset (FDR calculated from Wilcox-on rank-sum test with multiple adjustment). *** : p < 0.001

0

20

40

60

80

100

50

40

30

20

10

Vorin

osta

t

Pano

bino

stat

ISO

X

Apic

idin

Mer

ck60

Entin

osta

t

BRD

-K61

1665

97

Tace

dina

line

BRD

-K80

1833

49

Rep

ligen

136

Belin

osta

t

BRD

-K66

5322

83

BRD

-K94

3779

14

BRD

-K51

4902

54

BRD

-K88

7421

10

BRD

-K29

3133

08

BRD

-K24

6903

02

BRD

-K11

5332

27

BRD

-K85

1332

07

Pand

acos

tat

Tuba

stat

in A

Ison

icot

inoh

ydro

xam

ic A

cid

Num

ber o

f gen

es

Positive associationNegative association

HDAC inhibtors in CTRP

Figure S5

Figure S5. Therapeutic liability of circRNAs in CTRP. The numbers of HDAC inhibitor drugs significantly associated with circRNAs in Cancer Therapeutic Response Portal (CTRP) dataset. Blue bar denotes negative association and red bar denotes positive association.

AcircNotch2:chr1_120508075_120508189

B

10

12

14

16

0 2 4 6

EIF4G3 expression

log2

(tot

al b

acks

plic

ing

read

s)

ASPH

THBS1

CRIM1 PTMS

EFEMP1ANXA2

KRT7MUC16

PEA15

DCBLD2

0.0

2.5

5.0

7.5

10.0

−0.05 0.00 0.05 0.10

Vorinostat Response Difference

−Log10(FDR)

FDR >= 0.05FDR < 0.05

C

D

0

1

2

3

circRNA(+) circRNA(−)

log2

(TPM

)

−1

0

1

2

RPP

A_pr

otei

n_ab

unda

nces

circRNA(+) circRNA(−)

ERBB3 mRNA, FDR<2.2×10-16 ERBB3 protein,FDR=7.91×10-11

circKRT19 circKRT19

MMLUSC LGGUCEC STADKIRCCOAD SKCM ESCA LCLL OV

SARC PAAD BLCA MESO LIHCCESC DLBC HNSC THCA BRCAPRAD

circKRT19

ERBB3_mutation

FDR = 6.66×10-12

Figure S6

CESCLIHC

BRCASARC

LGGESCA

THCAHNSC

BLCAPAAD

LUSCDLBC

LCLLKIRC

OVMM

STADUCEC

COADPRAD

SKCMMESO

Rs = 0.37, FDR < 2.2 x 10-16

Figure S6. Examples of modules in CircRiC. (A) Expression landscape of circMYC across different cancer cell lines. (B) Correlation of EIF4G3 gene expression with total backsplicing reads. (C) Association between circRNAs and sensitivity to vorinostat using GDSC drug–response dataset. (D) Associ-ations between circKRT19 and ERBB3 mRNA, protein and mutation profile.

Table S1. Summary of circRNAs identified across different cancer cell lines

Cancer Lineages No. of cell lines No. of circRNA Lung squamous cell carcinoma [LUSC] 186 77284 Chronic lymphocytic leukemia [LCLL] 81 26172

Brain lower grade glioma [LGG] 65 26038 Colon adenocarcinoma [COAD] 58 21450

Diffuse large B-cell lymphoma [DLBC] 57 20654 Breast invasive carcinoma [BRCA] 56 20525 Skin cutaneous melanoma [SKCM] 52 19366

Ovarian serous cystadenocarcinoma [OV] 45 18296 Pancreatic adenocarcinoma [PAAD] 41 17639

Sarcoma [SARC] 40 15715 Stomach adenocarcinoma [STAD] 39 13381

Head and neck squamous cell carcinoma [HNSC] 33 11476 Liver hepatocellular carcinoma [LIHC] 32 10157 Bladder urothelial carcinoma [BLCA] 26 9132

Esophageal carcinoma [ESCA] 26 8899 Kidney renal clear cell carcinoma [KIRC] 25 8177

Multiple myeloma plasma cell leukemia [MM] 25 7786 Cervical squamous cell carcinoma and endocervical adenocarcinoma [CESC]

24 7427

Thyroid carcinoma [THCA] 12 4186 Prostate adenocarcinoma [PRAD] 7 2704

Uterine corpus endometrial carcinoma [UCEC] 4 2684 Mesothelioma [MESO] 1 181