Cancer-specific Splicing and Splicing QTLs

Revealed By Pan-Cancer Genome Analysis

Kjong-Van Lehmann

Ratsch Lab, New York City, USA

TCGA Symposium, May 13, 2014

Splicing and Drug sensitivity

S. Bonnal et al. (2012); Nature Reviews Drug Discovery

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 1

Memorial Sloan-Kettering Cancer Center

Analysis Across Multiple Cancer Types

Goals1 Identify cancer-specific splicing patterns

2 Identify variants regulating splicing in same gene (cis)

3 Identify variants regulating splicing in other cancer genes (trans)

TCGA provides RNA-seq and matching exome data

RNA-seq Find & quantify splicing events

Exome Identify variants in exons & flanking intronic regions

Problem: Non-uniform processing (alignments & variant calling)

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 2

Memorial Sloan-Kettering Cancer Center

Analysis Across Multiple Cancer Types

Goals1 Identify cancer-specific splicing patterns

2 Identify variants regulating splicing in same gene (cis)

3 Identify variants regulating splicing in other cancer genes (trans)

TCGA provides RNA-seq and matching exome data

RNA-seq Find & quantify splicing events

Exome Identify variants in exons & flanking intronic regions

Problem: Non-uniform processing (alignments & variant calling)

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 2

Memorial Sloan-Kettering Cancer Center

Analysis Across Multiple Cancer Types

Goals1 Identify cancer-specific splicing patterns

2 Identify variants regulating splicing in same gene (cis)

3 Identify variants regulating splicing in other cancer genes (trans)

TCGA provides RNA-seq and matching exome data

RNA-seq Find & quantify splicing events

Exome Identify variants in exons & flanking intronic regions

Problem: Non-uniform processing (alignments & variant calling)

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 2

Memorial Sloan-Kettering Cancer Center

Re-analysis of Raw Sequencing Data

Computing at cluster colocated with CGHub

Scale: 9,000 exome & 4,500 RNA-seq libraries 400 TB data

⇒ Re-mapping (STAR): ≈ 6 CPU years

⇒ Variant Calling (GATK U.G. & MuTect): ≈ 12 CPU years

⇒ Splice variant quantification (SplAdder): ≈ 0.5 CPU years

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 3

Memorial Sloan-Kettering Cancer Center

Detecting Alternative Splice Events with SplAdder

Estimating Splice Index

Build and extend splice-graph using re-aligned reads

Spliced reads support either Isoform 1 or Isoform 2

Count reads supporting alternate event

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 4

Memorial Sloan-Kettering Cancer Center

Detecting Alternative Splice Events with SplAdder

Estimating Splice Index

Build and extend splice-graph using re-aligned reads

Spliced reads support either Isoform 1 or Isoform 2

Count reads supporting alternate event

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 4

Memorial Sloan-Kettering Cancer Center

Detecting Alternative Splice Events with SplAdder

Estimating Splice Index

Build and extend splice-graph using re-aligned reads

Spliced reads support either Isoform 1 or Isoform 2

Count reads supporting alternate event

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 4

Memorial Sloan-Kettering Cancer Center

Detecting Alternative Splice Events with SplAdder

Estimating Splice Index

Build and extend splice-graph using re-aligned reads

Spliced reads support either Isoform 1 or Isoform 2

Count reads supporting alternate event

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 4

Memorial Sloan-Kettering Cancer Center

Splicing Variation Across Cancer Types

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 5

Memorial Sloan-Kettering Cancer Center

Detection of Cancer-Specific Splicing

≈500ENCODE RNA-Seq

TCGA/SplAdder Splicing Events

≈400 TCGA Normal

RNA-Seq

≈4000 TCGA Tumor

RNA-Seq

≈4000 TCGA Tumor RNA-

SeqId

entif

y Sp

lice

varia

nts

Confi

rm

Splic

e va

riant

sSc

ore

Splic

e va

riant

s

Detected new splicing events that occur frequently in specificcancer typesNeeds independent confirmationPotential targets for treatment

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 6

Memorial Sloan-Kettering Cancer Center

Detection of Cancer-Specific Splicing

Scor

e Sp

lice

varia

nts

Detected new splicing events that occur frequently in specificcancer types

Needs independent confirmation

Potential targets for treatment

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 6

Memorial Sloan-Kettering Cancer Center

QTL Analyses in Comparison

Challenges in Cancer Genomics

Opportunity: Understand tissue- & cancer specificity of splicing

Opportunity: Large sample size allows to find trans-associations

Problem: Heterogeneity and purity of sample

Problem: Germline vs. Somatic mutations, many rare variants

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 7

Memorial Sloan-Kettering Cancer Center

QTL Analyses in Comparison

Challenges in Cancer Genomics

Opportunity: Understand tissue- & cancer specificity of splicing

Opportunity: Large sample size allows to find trans-associations

Problem: Heterogeneity and purity of sample

Problem: Germline vs. Somatic mutations, many rare variants

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 7

Memorial Sloan-Kettering Cancer Center

Common Variant Association Analysis

Modeling cancer and population structure

Y = Xβ + Pop. Structure + Cancer Structure +ε

Pop. Structure ∼ N(0, σ2pP) with P = XgermX

Tgerm

Cancer Structure ∼ N(0, σ2cC ) with C = XsomaX

Tsoma

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 8

Memorial Sloan-Kettering Cancer Center

Common Variant Association Analysis

Modeling cancer and population structure

Y = Xβ + Pop. Structure + Cancer Structure +ε

Pop. Structure ∼ N(0, σ2pP) with P = XgermX

Tgerm

Cancer Structure ∼ N(0, σ2cC ) with C = XsomaX

Tsoma

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 8

Memorial Sloan-Kettering Cancer Center

Common Variant Association Analysis

Modeling cancer and population structure

Y = Xβ + Pop. Structure + Cancer Structure +ε

Pop. Structure ∼ N(0, σ2pP) with P = XgermX

Tgerm

Cancer Structure ∼ N(0, σ2cC ) with C = XsomaX

Tsoma

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 8

Memorial Sloan-Kettering Cancer Center

Common Variant Association Analysis

Modeling cancer and population structure

Y = Xβ + Pop. Structure + Cancer Structure +ε

Pop. Structure ∼ N(0, σ2pP) with P = XgermX

Tgerm

Cancer Structure ∼ N(0, σ2cC ) with C = XsomaX

Tsoma

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 8

Memorial Sloan-Kettering Cancer Center

Example: cis-Associations in SNRP-C

TRPT1 - tRNA Phosphotransferase I

+ 34,725,183Position on Chr 6

-log1

0 p-

valu

e

Alt (345) Het (1101) Ref (971)

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 9

Memorial Sloan-Kettering Cancer Center

cis-Associations across Multiple Cancer Types

cis-Associations in 45 genes at 5% FDR (in 900 considered genes)

Cancer Census Genes Splicing-related genes

Replicated in multiple cancer types:

KTN1 PRCC

NUP98SSXT COX6C

MMAB

ANM7

LSM1 PRP18

THOCSRU1CU2AF4

TRPT1

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 10

Memorial Sloan-Kettering Cancer Center

Splicing trans-Associations (FDR 5%)

NDRG1

NTRK2

PRP19CELF2

HDAC6

Cancer Census

Transcript. Factor

RNA Binding

other

Cancer spec. Splicing

SNV -> Splicing

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 11

Memorial Sloan-Kettering Cancer Center

Splicing trans-Associations (FDR 5%)

LSM1

LSM2DX398 RU1CU2AF4

TRPT1

ERR1NDS3

FLT3

NDRG1

NTRK2

PRP19

P53

CELF2

FRG1

HDAC6

Cancer Census

Transcript. Factor

RNA Binding

other

Cancer spec. Splicing

SNV -> Splicing

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 11

Memorial Sloan-Kettering Cancer Center

Splicing trans-Associations (FDR 5%)

KTN1 PRCC

NUP98

SSXT

COX6C

MMAB

ANM7

LSM1

LSM2

PRP18

DX398 THOCSRU1CU2AF4

TRPT1

ERR1

RNPS1 SON

HG2A

MERLNDS3

NTRK1

PK3CA

SRSF6

FLT3

NDRG1IGF2

NTRK2

PRP19

P53

ABI1

CBX3

CELF2

FRG1MAGI1

HDAC6GABPA SMC3MAX SMCA4

Cancer Census

Transcript. Factor

RNA Binding

other

Cancer spec. Splicing

SNV -> Splicing

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 11

Memorial Sloan-Kettering Cancer Center

Conclusions and future work

Developed resource of novel & known alternative splice events

Identified cancer-specific isoforms that appear rarely expressed innormal samples

Performed common variant association study to map splicingphenotypes

Sample size in TCGA data enables detection of transassociations

All of these associations still need validation (in particular trans)

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 12

Memorial Sloan-Kettering Cancer Center

Acknowledgements

Andre Kahles

Cyriac Kandoth

William Lee

Cancer Genome Atlas Network

Nikolaus Schultz

Robert Klein

Oliver Stegle

Gunnar Ratsch

Funded by MSKCC.

Kjong-Van Lehmann (MSKCC, New York) TCGA Symposium, 2014 13

Memorial Sloan-Kettering Cancer Center