Poster TZH JOBIM - Zucman labzucmanlab.com/wp-content/uploads/2019/06/Poster_TZH_JOBIM.pdf ·...
1
Cancer genomes are altered by various mutational processes and, like palimpsests, bear the signatures of these different processes. Mutational signature analysis is a powerful approach to decipher the origin of somatic mutations in cancer [1]. The Palimpsest R package [2] allows to integrate mutational signature and clonality analyses in order to reconstruct the natural history of a tumor. Palimpsest: an R package for studying mutational and structural variant signatures along clonal evolution in cancer 1 Centre de Recherche des Cordeliers, Functional Genomics of Solid Tumors laboratory, Sorbonne Université, Inserm, USPC, Université Paris Descartes, Université Paris Diderot, Paris, France 2 Labex OncoImmunology, Equipe labellisée Ligue Contre le Cancer, Centre de Recherche des Cordeliers, Paris, France 3 Laboratory for Bioinformatics, Fondation Jean Dausset – CEPH, Paris F-75010, France Cancer genomes bear the signatures of different mutational processes Palimpsest package: extracting signatures of known and new mutational processes Blind source separation with NMF (non-negative matrix factorization) SBS5 SBS4 SBS12 ? ID7 Defective DNA mismatch repair DBS7 SV1 SBS7 Replicative stress Single base substitution Double base substitution Small insertions and deletions Structural variants SV ID DBS SBS ... ... ... s samples c categories = k signatures c categories ... ... ... x k signatures s samples ... ... ... SBS1 SBS5 SBS16 SBS23 Mutation catalog Processes Exposures De novo extraction of novel signatures Already known signatures (Cosmic database) Can be applied to any type of somatic alteration (SBS, DBS, ID or SV) De novo extracted signatures can be compared to Cosmic signatures with cosine similarities Clonal architecture of tumors - single sample Assigning the causal process of each alteration https://github.com/FunGeST/Palimpsest Get the poster Our GitHub Theo Z HIRSCH 1, 2 , Jayendra SHINDE 1, 2 , Benedict MONTEIRO 1, 2 , Quentin BAYARD 1, 2 , Sandrine IMBEAUD 1, 2 , Feng LIU 3 , Victor RENAULT 3 , Jessica ZUCMAN-ROSSI 1, 2 and Eric LETOUZÉ 1, 2 Corresponding Authors: TZ. Hirsch ([email protected]); E. Letouzé ([email protected]) Somatic alterations are classified into categories [3,4] according to: - Type (T>A, Inversion, Deletion ...) - Context (trinucleotide, repeats ...) - Size (for ID and RS) - ... Identification of recurrent patterns reveals the signatures of each mutational processes Zygote Gestation Childhood Adulthood Benign tumor Malignant tumor Cumulative probabilities time Proportion of Structural Variants Deletion Tandem dup Inversion Trans Deletion Tandem dup Inversion Trans 0-1kb 1-10kb 10-100kb 100kb-1Mb 1-10Mb >10Mb 0-1kb 1-10kb 10-100kb 100kb-1Mb 1-10Mb >10Mb 0-1kb 1-10kb 10-100kb 100kb-1Mb 1-10Mb >10Mb 0-1kb 1-10kb 10-100kb 100kb-1Mb 1-10Mb >10Mb 0-1kb 1-10kb 10-100kb 100kb-1Mb 1-10Mb >10Mb 0-1kb 1-10kb 10-100kb 100kb-1Mb 1-10Mb >10Mb Clustered rearrangements Non-clustered rearrangements For each variant in each sample: - probability that it was due to each identified signatures - assign the most likely signature Drivers Etiologies Clonal architecture of tumors - multiple samples Cancer Cell Fraction (CCF) + Tumor purity [1] Alexandrov et al, Nature 2013. PMID: 23945592 [2] Shinde et al, Bioinformatics 2018. PMID: 29771315 [3] Bayard et al, Nature Communications 2018. PMID: 30531861 [4] Alexandrov et al, Preprint 2018. bioRxiv 322859 [5] Letouzé*, Shinde* et al, Nature Communications 2017. PMID: 29101368 + Copy-number Clonal (early) mutations Subclonal (late) mutations VAF Comparison of mutational processes between clonal and subclonal mutations: Variant-allele frequency (VAF) [5] Sample 1 Sample 2 Sample 3 Trunk C13 For each mutation, compute the CCF in each sample Bayesian Dirichlet process to identify clusters of CCF Reconstruct the evolutionary tree of the tumor Analyze mutational signatures in each branch Clock-like These can occur from the zygote through adulthood, and also throughout the lifespan of the tumor NMF: finding the solution to AVAILABLE SOON Contribution of mutational signatures to driver mutations in liver tumors [5]
Transcript of Poster TZH JOBIM - Zucman labzucmanlab.com/wp-content/uploads/2019/06/Poster_TZH_JOBIM.pdf ·...
Cancer genomes are altered by various mutational processes and, like palimpsests, bear the signatures of these different processes. Mutational signature analysis is a powerful approach to decipher the origin of somatic mutations in cancer [1].The Palimpsest R package [2] allows to integrate mutational signature and clonality analyses in order to reconstruct the natural history of a tumor.
Palimpsest: an R package for studying mutational and structural variant signatures along clonal evolution in cancer
1 Centre de Recherche des Cordeliers, Functional Genomics of Solid Tumors laboratory, Sorbonne Université, Inserm, USPC, Université Paris Descartes, Université Paris Diderot, Paris, France2 Labex OncoImmunology, Equipe labellisée Ligue Contre le Cancer, Centre de Recherche des Cordeliers, Paris, France 3 Laboratory for Bioinformatics, Fondation Jean Dausset – CEPH, Paris F-75010, France
Cancer genomes bear the signatures of different mutational processes
Palimpsest package: extracting signatures of known and new mutational processes
Blind source separation with NMF(non-negative matrix factorization)
SBS5
SBS4
SBS12 ? ID7 Defective DNAmismatch repair
DBS7
SV1
SBS7
Replicative stress
Single basesubstitution
Double basesubstitution
Small insertionsand deletions
Structural variantsSV
ID
DBS
SBS
...
...
...
s samples
c ca
tego
ries
=
k signatures
c ca
tego
ries
...
...
...
x
k si
gnat
ures
s samples...
...
...
SBS1 SBS5 SBS16 SBS23
Mutation catalog Processes Exposures
De novo extraction of novel signatures
Already known signatures(Cosmic database)
Can be applied to any type of somatic alteration (SBS, DBS, ID or SV)
De novo extracted signaturescan be compared to Cosmicsignatures with cosine similarities
Clonal architecture of tumors - single sample
Assigning the causal process of each alteration
https://github.com/FunGeST/Palimpsest
Get the posterOur GitHub
Theo Z HIRSCH1, 2, Jayendra SHINDE1, 2, Benedict MONTEIRO1, 2, Quentin BAYARD1, 2, Sandrine IMBEAUD1, 2, Feng LIU3, Victor RENAULT3, Jessica ZUCMAN-ROSSI1, 2 and Eric LETOUZÉ1, 2
Corresponding Authors: TZ. Hirsch ([email protected]); E. Letouzé ([email protected])
Somatic alterations are classifiedinto categories [3,4] according to:- Type (T>A, Inversion, Deletion ...)- Context (trinucleotide, repeats ...)- Size (for ID and RS)- ...
Identification of recurrent patternsreveals the signatures of each mutational processes
Zygote
Gestation
Childhood
Adulthood
Benign tumor
Malignanttumor
Cum
ulat
ive
prob
abili
ties
time
Pro
porti
on o
f Stru
ctur
al V
aria
nts Deletion Tandem dup Inversion Trans Deletion Tandem dup Inversion Trans
0-1k
b
1-10
kb
10-1
00kb
100k
b-1M
b
1-10
Mb
>10M
b
0-1k
b
1-10
kb
10-1
00kb
100k
b-1M
b
1-10
Mb
>10M
b
0-1k
b
1-10
kb
10-1
00kb
100k
b-1M
b
1-10
Mb
>10M
b
0-1k
b
1-10
kb
10-1
00kb
100k
b-1M
b
1-10
Mb
>10M
b
0-1k
b
1-10
kb
10-1
00kb
100k
b-1M
b
1-10
Mb
>10M
b
0-1k
b
1-10
kb
10-1
00kb
100k
b-1M
b
1-10
Mb
>10M
b
Clustered rearrangements Non-clustered rearrangements
For each variant in each sample:
- probability that it was due to each identified signatures- assign the most likely signature
Drivers Etiologies
Clonal architecture of tumors - multiple samples
Cancer Cell Fraction (CCF)+ Tumor purity
[1] Alexandrov et al, Nature 2013. PMID: 23945592[2] Shinde et al, Bioinformatics 2018. PMID: 29771315[3] Bayard et al, Nature Communications 2018. PMID: 30531861[4] Alexandrov et al, Preprint 2018. bioRxiv 322859[5] Letouzé*, Shinde* et al, Nature Communications 2017. PMID: 29101368
+ Copy-number Clonal (early)mutations
Subclonal (late)mutations
VAF
Comparison of mutational processes between clonal and subclonal mutations:
Variant-allele frequency (VAF)
[5]
Sample 1
Sample
2
Sam
ple
3
Trunk
C13For each mutation, compute the CCF in each sample
Bayesian Dirichlet process to identify clusters of CCF
Reconstruct the evolutionarytree of the tumor
Analyze mutationalsignatures in each branch
Clock-like
These can occur from the zygotethrough adulthood, and also throughout the lifespan of thetumor
NMF: finding the solution to
AVAILABLESOON
Contribution of mutational signaturesto driver mutations in liver tumors [5]
