BIOINFORMATICS FOR HEALTH SCIENCES
Introduction to Cancer GenomicsNuria Lopez-Bigas
Moving towards personalized cancer medicine
Marc Rosenthal
Cancer Genomics
Cancer Genomics
ACTCAGCCCCAGCGGAGGTGAAGGACGTCCTTCCCCAGGAGCCGGTGAGAAGCGCAGTCGGGGGCACGGGGATGAGCTCAGGGGCCTCTAGAAAGATGTAGCTGGGACCTCGGGAAGCCCTGGCCTCCAGGTAGTCTCAGGAGAGCTACTCAGGGTCGGGCTTGGGGAGAGGAGGAGCGGGGGTGAGGCCAGCAGCAGGGGACTGGACCTGGGAAGGGCTGGGCAGCAGAGACGACCCGACCCGCTAGAAGGTGGGGTGGGGAGAGCATGTGGACTAGGAGCTAAGCCACAGCAGGACCCCCACGAGTTGTCACTGTCATTTATCGAGCACCTACTGGGTGTCCCCAGTGTCCTCAGATCTCCATAACTGGGAAGCCAGGGGCAGCGACACGGTAGCTAGCCGTCGATTGGAGAACTTTAAAATGAGGACTGAATTAGCTCATAAATGGAAAACGGCGCTTAAATGTGAGGTTAGAGCTTAGAATGTGAAGGGAGAATGAGGAATGCGAGACTGGGACTGAGATGGAACCGGCGGTGGGGAGGGGGAGGGGGTGTGGAATTTGAACCCCGGGAGAGAAAGATGGAATTTTGGCTATGGAGGCCGACCTGGGGATGGGGAAATAAGAGAAGACCAGGAGGGAGTTAAATAGGGAATGGGTTGGGGGCGGCTTGGTAACTGTTTGTGCTGGGATTAGGCTGTTGCAGATAATGGAGCAAGGCTTGGAAGGCTAACCTGGGGTGGGGCCGGGTTGGGGTCGGGCTGGGGGCGGGAGGAGTCCTCACTGGCGGTTGATTGACAGTTTCTCCTTCCCCAGACTGGCCAATCACAGGCAGGAAGATGAAGGTTCTGTGGGCTGCCCCGACCCGCTAGAAGGTGGGGTGGGGAGAGCATGTGGACTAGGAGCTAAGCCACAGCAGGACCCCCACGAGTTGTCACTGTCATTTATCGAGCACCTACTGGGTGTCCCCAGTGTCCTCAGATCTCCATAACTGGGAAGCCAGGGGCAGCGAC
Arrays Parallel Sequencing
Cancer Genomics Projects
Expression patterns
Somatic mutations
Epigenomic profiles
Structural aberrations
Copy number alterations
Patient cohortPrimary tumors
Cancer Genomics Projects
Expression patterns
Somatic mutations
Epigenomic profiles
Structural aberrations
Copy number alterations
Patient cohortPrimary tumors
Cancer Genomics Projects
Cancer Genomic Projects
OBJECTIVE:Obtain full catalog of genetic alterations in
500 tumors from 50 tumor types
• Somatic mutations• Copy Number Alterations • Abnormal expression of genes• Translocations• Epigenetic modifications• etc.
Cancer Genomics: What for?
•Finding new cancer genes (cancer drivers)•Finding new therapeutic targets• Identify molecular signatures to stratify tumors•Move towards personalized cancer treatment
Cancer Genomics: What for?
•Finding new cancer genes (cancer drivers)•Finding new therapeutic targets• Identify molecular signatures to stratify tumors•Move towards personalized cancer treatment
Cancer Genomics: What for?
•Finding new cancer genes (cancer drivers)•Finding new therapeutic targets• Identify molecular signatures to stratify tumors•Move towards personalized cancer treatment
1985
BCR-ABL fusion cause Chronic Myelogenous Leukemia (CML)
Weisberg et al., Nature Reviews Cancer 2007
BCR-ABL: constitutive active ABL kinase activity
Imatinib
Imatinib inhibits tyrosine-kinase activity of ABL
Kantarjian et al., Blood 2012
Dramatically improved long term survival rates (95.2%) since the introduction of Gleevec in 2001
BRAF is frequently mutated in melanoma
VemurafenibPLX432
BRAF is frequently mutated in melanoma
2 setmanesVemurafenib
2 setmanesVemurafenib
2 setmanesVemurafenib
Personalized medicine / Precision medicine
Vemurafenib
Targeted Cancer Therapy
Cancer-causing mutations with drug treatment available
Mutation with no drug available
Radiation and chemotherapyCancer drug
Schema adapted from NY times
Cancer Genomics: What for?
•Finding new cancer genes (cancer drivers)•Finding new therapeutic targets• Identify molecular signatures to stratify tumors•Move towards personalized cancer treatment
Identify molecular signatures to stratify tumors
Good prognosisFavorable response
Bad prognosisUnfavorable response Increased toxicity
Cancer Genomics: What for?
•Finding new cancer genes (cancer drivers)•Finding new therapeutic targets• Identify molecular signatures to stratify tumors•Move towards personalized cancer treatment
Move towards personalized cancer treatment
YESTERDAY TODAY TOMORROW
Find the right treatment for the right patient
at the right time
Cancer Genomics: What for?
•Finding new cancer genes (cancer drivers)•Finding new therapeutic targets• Identify molecular signatures to stratify tumors•Move towards personalized cancer treatment
Finding Cancer Drivers
Cancer genome sequencing
Which mutations are cancer drivers?
Normal cell Cancer cell
Sequencing machines
Somatic mutations
Finding Cancer Drivers
1. Predict consequences of mutations2. Assess the functional impact of nsSNVs3. Identify cancer drivers based on recurrence4. Identify cancer drivers based on FMbias
1. Predict consequences of mutationsACTGCCTACGTCTCACCGTCGACTTCAAATCGCTTAACCCGTACTCCCATGCTACTGCATCTCGGGTTAACTCGACGTTTTTCATGCATGTGTGCACCCCAATATATATGCAACTTTTGTGCACCTCTGTCACGCGCGAGTTGGCACTGTCGCCCCTGTGTGCATGTGCACTGTCTCTCGCTGCACTGCCTACGTCTCACCGTCGACTTCAAATCGCTTAACCCGTACTCCCATGCTACTGCATCTCGGGTTAACTCGACGTTTTGCATGCATGTGTGCACCCCAATATATATGCAACTTTTGTGCACCTCTGTCACGCGCGAGTTGGCACTGTCGCCCCTGTGTGCATGTGCACTGTCTCTCGA
Map mutations into genome annotations to predict its possible effect
Tools to annotate consequences of mutations
ANNOVAR
snpEff
VAGrENT
annTools
ASOoVIREnsembl VEP
2. Assess the functional impact of nsSNVs
ATC GAA GCA CGTMet Glu Ala Gly
nsSNVs = non-synonymos Single Nucleotide Variant (missense)
ATC GAC GCA CGTMet Asp Ala Gly
Computational methods to assess the functional impact of nsSNVs
MutationTaster
SIFTPolyPhen2
CondelCHASM
PMut
SNPs&GO
SNPeffect
MutPred
MutationAssessor
CanPredict
LogRe
transFIC
Which mutations are cancer drivers?
Normal cell Cancer cell
Sequencing machines
Somatic mutations
Patient cohort
3. Identify cancer drivers from somatic mutations
Find signals of selection across tumors
Cancer is an evolutionary process
Yates and Campbell et al, Nat Rev Genet 2012
How to differentiate drivers from passengers?
ACTGCCTACGTCTCACCGTCGACTTCAAATCGCTTAACCCGTACTCCCATGCTACTGCATCTCGGGTTAACTCGACGTTTTTCATGCATGTGTGCACCCCAATATATATGCAACTTTTGTGCACCTCTGTCACGCGCGAGTTGGCACTGTCGCCCCTGTGTGCATGTGCACTGTCTCTCGCTGCACTGCCTACGTCTCACCGTCGACTTCAAATCGCTTAACCCGTACTCCCATGCTACTGCATCTCGGGTTAACTCGACGTTTTGCATGCATGTGTGCACCCCAATATATATGCAACTTTTGTGCACCTCTGTCACGCGCGAGTTGGCACTGTCGCCCCTGTGTGCATGTGCACTGTCTCTCGAGTTTTGCATGCATGTGTGCACTGTGCACCTCTGTTACGTCT
How to differentiate drivers from passengers?
ACTGCCTACGTCTCACCGTCGACTTCAAATCGCTTAACCCGTACTCCCATGCTACTGCATCTCGGGTTAACTCGACGTTTTTCATGCATGTGTGCACCCCAATATATATGCAACTTTTGTGCACCTCTGTCACGCGCGAGTTGGCACTGTCGCCCCTGTGTGCATGTGCACTGTCTCTCGCTGCACTGCCTACGTCTCACCGTCGACTTCAAATCGCTTAACCCGTACTCCCATGCTACTGCATCTCGGGTTAACTCGACGTTTTGCATGCATGTGTGCACCCCAATATATATGCAACTTTTGTGCACCTCTGTCACGCGCGAGTTGGCACTGTCGCCCCTGTGTGCATGTGCACTGTCTCTCGAGTTTTGCATGCATGTGTGCACTGTGCACCTCTGTTACGTCT
Find signals of positive selection across tumour re-sequenced genomes
Recurrence
Identify genes mutated more frequently than background mutation rate
MuSiC-SMG / MutSigCV
Mutation
Signals of positive selection
Recurrence
Identify genes mutated more frequently than background mutation rate
MuSiC-SMG / MutSigCV
Mutation
Signals of positive selection
Challenge: Background mutation rate varies across patients and genomic regions
Replication time
Stamatoyannoppoulos et al., Nature Genetics 2009 Schuster-Böckler and Lehner, Nature 2011
Chromatin organization
Signals of positive selection
Functional impact bias (FMbias)
Mutation
OncodriveFM
Gonzalez-Perez and Lopez-Bigas. NAR 2012
Functional Impact
Signals of positive selection
• Based on consequences of mutations (eg. synonymous is
lowest and STOPgain, frameshift indel highest)
• And SIFT, PPH2 and MA for missense
How to measure functional impact of mutations?
Functional impact bias (FMbias)
Mutation
OncodriveFM
Gonzalez-Perez and Lopez-Bigas. NAR 2012
Functional Impact
Signals of positive selection
Functional impact bias (FMbias)
Mutation
• It does not depend on background mutation rates
• Only needs list of somatic mutations
• It is computationally cheap
Main Advantages of FM bias approach
Gonzalez-Perez and Lopez-Bigas. NAR 2012
Functional Impact
OncodriveFM
Signals of positive selection
Functional impact bias (FMbias)
Mutation
One example: TCGA Glioblastoma FMbiasqvalue
TP53PTENEGRFNF1RB1FKBP9ERBB2PIK3R1PIK3CAPIK3C2GIDH1ZNF708FGFR3CDKN2AALDH1A3PDGFRAFGFR1MAPK9DCNPIK3C2ACHEK2PSMD13GSTM5
8.5E-118.5E-118.5E-118.5E-112.5E-98.5E-111.2E-81.2E-82.3E-40.0028.5E-117.4E-103.2E-92.5E-85.2E-51.5E-62.0E-62.2E-51.5E-66.2E-5111
not mutatedMA score
5-2 0 0.05 10
FM bias qvalue
OncodriveFM
Functional Impact
PIK3CA is recurrently mutated in the same residue in breast tumours
H1047L
PIK3CA
Protein position0 1047
Prot
ein
affe
ctin
g m
utat
ions
80
0
Signals of positive selection
Mutation clustering
Mutation
OncodriveCLUST
Tamborero et al., Bioinformatics 2013
Th
Gene A Gene B(I)
(II)
(III)
(IV)
(V)
Th
SgeneA
= Sc1 S
geneB = Sc1
+ SC2
(VI)
0
ZA
ZB
mut
atio
ns
Amino acid
C1
C1 C2
Amino acid
mut
atio
ns
mut
atio
ns
mut
atio
ns
SgeneA
SgeneB
Background model obtained by calculating the clustering score per gene of the coding-silent mutations
Signals of positive selection: OncodriveCLUST
Tamborero et al., Bioinformatics 2013
List of tumor somatic
mutations
Input data
IntOGen mutations pipeline To interpret catalogs of cancer somatic mutations
✓ Identify consequences of mutations (Ensembl VEP)✓ Assess functional impact of nsSNVs (SIFT, PPH2, MA and TransFIC)✓ Compute frequency of mutations per gene and pathway✓ Identify candidate driver genes (OncodriveFM and OncodriveCLUST)✓ Identify pathways with FM bias (OncodriveFM)
Gonzalez-Perez et al, Nature Methods 2013
Analysis Pipeline Browser
✓ Identify consequences of mutations (Ensembl VEP)✓ Assess functional impact of nsSNVs (SIFT, PPH2, MA and TransFIC)✓ Compute frequency of mutations per gene and pathway✓ Identify candidate driver genes (OncodriveFM and OncodriveCLUST)✓ Identify pathways with FM bias (OncodriveFM)
Input data
Working version:49 Projects 28 Cancer types6792 tumours
.org
http://www.intogen.org/mutations
IntOGen mutations pipeline To interpret catalogs of cancer somatic mutations
Gonzalez-Perez et al, Nature Methods 2013
List of tumor somatic
mutations
Current version:31 Projects 13 Cancer sites4623 tumours
Analysis Pipeline Browser
http://www.intogen.org/mutations
Gonzalez-Perez et al, Nature Methods 2013
http://www.intogen.org/mutations/analysis
Gonzalez-Perez et al, Nature Methods 2013
IntOGen-mutations pipelineTo interpret catalogs of cancer somatic mutations
Cancer Genomics: What for?
•Finding new cancer genes (cancer drivers)•Finding new therapeutic targets• Identify molecular signatures to stratify tumors•Move towards personalized cancer treatment
Stratify tumors based on molecular patterns
Good prognosisFavorable response
Bad prognosisUnfavorable response Increased toxicity
Stratify tumors based on molecular patterns
Stratify tumors based on molecular patterns
One example: Breast Cancer Intrinsic Subtypes
TCGA Pan-Cancer project