Joaquín Dopazo

Computational Genomics Department,

Centro de Investigación Príncipe Felipe (CIPF),

Functional Genomics Node, (INB),

Bioinformatics in Rare Diseases (BiER-CIBERER),

Valencia, Spain.

Digging into thousands of variants to

find disease genes in Mendelian and

complex diseases

http://bioinfo.cipf.es http://www.babelomics.org

@xdopazo

IV DCEX symposium, Barcelona, November 17th, 2015

Precision medicine (P4*) is based on a better knowledge of phenotype-

genotype relationships

Requires of a better way of defining diseases by introducing genomic

technologies in the diagnostic procedures and treatment decisions

*P4: Predictive, Preventive, Personalized, Participative

Setting the problem in context: The transition to precision medicine

Intuitive Based on trial

and error

Identification of probabilistic

patterns

Decisions and actions based on knowledge

Intuitive Medicine Empirical Medicine Precision Medicine (P4)

Today Tomorrow

Degree of personalization

Empirical Medicine First Phase I: generation of the knowledge database

-------------------------------------------------------

sequencing

Patient Variants

Database. Query

Therapy outcome

System feedback

Genomic variants (biomarkers) can be quickly associated to precise diagnostics and therapy outcomes

Initially the system will need much feedback: Knowledge generation phase.

Empirical medicine

Knowledge database

Genome projects enable knowledge generation

Rare and familiar diseases

exome sequencing initiative

• Metabolic (86 samples)

• Optiz

• Atypical fracture

• coQ10 deficiency

• Congenital disorder of glycosylation types I and II

• Maple syrup urine disease

• Pelizaeus-like

• 4 unknown syndroms

• Genetic (24 samples)

• Charcot-Marie-Tooth

• Rett Syndrome

• Neurosensorial (35 samples)

• Usher

• AD non-syndromic hearing loss

• AR non-syndromic hearing loss

• RP

• Mitochondrial (28 samples)

• Progressive External Oftalmoplegy

• Multi-enzymatic deficiency in mitochondrial

respiratory complexes

• CoQ disease

• Other

• APL (10 samples)

Autism (37 samples)

Mental retardation (autosomal recessive) (24)

Immunodeficiency (18)

Leber's congenital amaurosis (9)

Cataract (2)

RP(AR) (60)

RP(AD) (46)

Deafness (24)

CLAPO (4)

Skeletal Dysplasia (3)

Cantú syndrome (1)

Dubowitz syndrome (2)

Gorham-Stout syndrome (1)

Malpuech syndrome (4)

Hirschprung’s disease (81)

Hereditary macrothrombocytopenia (3)

MTC (41)

Controls (301)

1044 samples = 183 samples + 200 controls + 360 samples + 301 controls

Data management, analysis

and storage

http://www.gbpa.es/

GCGTATAG

CACGGGTA

TCTGTATTA

TGGTGGAT

ATCAGCGG

ATTGCGATT

GGCAGAGC

GGCAAAGT

GCGTATAG

CACGGGTA

TCTGTATTA

TGGTGGAT

ATCAGCGG

ATTGCGATT

GGCAGAGC

GGCAAAGT

GCGTATAG

CACGGGTA

TCTGTATTA

TGGTGGAT

ATCAGCGG

ATTGCGATT

GGCAGAGC

GGCAAAGT

GCGTATAG

CACGGGTA

TCTGTATTA

TGGTGGAT

ATCAGCGG

ATTGCGATT

GGCAGAGC

GGCAAAGT

Raw files

(FastQ)

DB

Analysis

Pipeline

Storage

K-DB

Gene 1 ksdhkahcka

Gene 2 jckacsksda

Gene 3 lkkxkccj<jdc

Gene 4 ksfdjvjvlsdkvjd

Gene 5 kckcksñdksd

Gene 6 ldkdkcksdcldl

Gene x kcdlkclkldsklk

Gene Y jcdksdkcdks

Prioritization

report Dialog with experts in the

disease + validations

Samples

GCGTATAG

CACGGGTA

TCTGTATTA

TGGTGGAT

ATCAGCGG

GCGTATAG

CACGGGTA

TCTGTATTA

TGGTGGAT

ATCAGCGG

VCF BAM Processed files

Casablanca: Round up all suspicious characters and search them for stolen documents.

The prioritization process Typically, an exome renders between 40 and 60K variants (and a

genome about 1 million). Only one or a few among all of them are

expected to be the causative factors of the disease.

The prioritization process is like a police investigation in which

suspected are discarded by their alibies

Thus, through sequential heuristic filtering steps, unlikely candidates

are discarded and a final, reduced list with one or a few candidate

genes is (hopefully) produced

Agatha Christie

40-60K suspects?

We need…

Pipeline of data analysis Primary processing

Initial QC

FASTQ file

Mapping

BAM file

Variant calling

VCF File

Knowledge-based prioritization

Proximity to other known disease genes

Functional proximity

Network proximity

Burden tests

Other prioritization methods

Secondary analysis

(Heuristic filtering)

Variant annotation

Filtering by effect

Filtering by MAF

Filtering by family segregation

Primary analysis

Gene prioritization

Pipeline of data analysis Primary processing

Initial QC

FASTQ file

Mapping

BAM file

Variant calling

VCF File

Knowledge-based prioritization

Proximity to other known disease genes

Functional proximity

Network proximity

Burden tests

Other prioritization methods

Secondary analysis

(Heuristic filtering)

Variant annotation

Filtering by effect

Filtering by MAF

Filtering by family segregation

Primary analysis

Gene prioritization

CellBase

Knowledge DB

(Now at EBI)

The knowledge database for filtering by effect

CellBase (Bleda, 2012, NAR), a comprehensive integrative database and RESTful Web Services API:

● Core features: genes, transcripts, exons, cytobands, proteins (UniProt),...

● Variation: dbSNP and Ensembl SNPs, HapMap, 1000Genomes, Cosmic, ClinVar, ...

● Functional: 40 OBO ontologies (Gene Ontology, HPO, etc.), Interpro, etc.

● Regulatory: TFBS, miRNA targets, conserved regions, tissue expression etc.

● System biology: Interactome (IntAct), Reactome database, co-expressed genes.

NoSQL (Mongo) and scales to TB

Wiki: http://docs.bioinfo.cipf.es/projects/cellbase/wiki

Project: http://bioinfo.cipf.es/compbio/cellbase

Now available at the EBI: http://www.ebi.ac.uk/cellbase/webservices/rest/v3/ and the

University of Cambridge: http://bioinfo.hpc.cam.ac.uk/cellbase/webservices/rest

Pipeline of data analysis Primary processing

Initial QC

FASTQ file

Mapping

BAM file

Variant calling

VCF File

Knowledge-based prioritization

Proximity to other known disease genes

Functional proximity

Network proximity

Burden tests

Other prioritization methods

Secondary analysis

(Heuristic filtering)

Variant annotation

Filtering by effect

Filtering by MAF

Filtering by family segregation

Primary analysis

Gene prioritization

1000 genomes

EVS

Local variants

3-Methylglutaconic aciduria (3-

MGA-uria) is a heterogeneous

group of syndromes

characterized by an increased

excretion of 3-methylglutaconic

and 3-methylglutaric acids.

WES with a consecutive filter

approach is enough to detect

the new mutation in this case.

Heuristic Filtering approach An example with 3-Methylglutaconic aciduria syndrome

Use known variants and their population frequencies to filter out.

• Typically dbSNP, 1000 genomes and

the 6515 exomes from the ESP are

used as sources of population

frequencies.

• We sequenced 300 healthy controls

(rigorously phenotyped) to add and

extra filtering step to the analysis

pipeline

Novembre et al., 2008. Genes mirror

geography within Europe. Nature Comparison of MGP controls to 1000g

How important do you

think local information is

to detect disease genes?

Filtering with or without local variants

Number of genes as a function of individuals in the study of a dominant disease Retinitis Pigmentosa autosomal dominant

The use of local

variants makes

an enormous

difference

The CIBERER Spanish Variant Server (CSVS): the first repository of variability of the Spanish population

Only another similar initiative

exists: the GoNL

http://www.nlgenome.nl/ http://ciberer.es/bier/exome-server/

And more recently

the Finnish

and the Icelandic

populations

Information provided

Genotype frequencies

in the different

reference populations

Genomic coordinates, variation, gene.

SNPid

if any

Information provided

Pathogenicity indexes

Phenotype,

if available

Accelerating the loop of knowledge generation BiERapp: interactive web-based tool for easy candidate prioritization by successive filtering

SEQUENCING CENTER

Data preprocessing

VCF FASTQ

Genome Maps

BAM

BiERapp filters

No-SQL (Mongo) VCF indexing

Population frequencies Consequence types

Experimental design

BAM viewer and Genomic context ?

Easy

sc

ale

up

Quick contribution to empirical medicine: new variants

and disease genes found with WES and heuristic filtering

WES

IRDs

arRP (EYS)

BBS

arRP arRP (USH2)

3-MGA-uria

(SERAC1)

NBD (BCKDK )

In less than two years, the CIBERER initiative rendered 36 new mutations in

known disease genes and 27 new mutations in13 new disease genes.

Limitations of the single-gene

approach One might or might not be aware of it, but the single-gene

approach implies the following assumptions:

• Only one gene (or at least one main gene) is the

causative agent of the disease.

• The disease has complete (or very high) penetrance.

• Deleterious variation with no phenotype is not expected

(or expected to be marginal)

Even Mendelian genes can be elusive.

Intuitive belief: multiple family information should help

Families

1 2 3 4 5 6

Variants 3403 82 4 0 0 0

Genes 2560 331 35 8 1 0

Observation: this is not always true, not even in cases

of Mendelian diseases

Limitations of the single-gene

approach One might or might not be aware of it, but the single-gene

approach implies the following assumptions:

• Only one gene (or at least one main gene) is the

causative agent of the disease.

• The disease has complete (or very high) penetrance.

• Deleterious variation with no phenotype is not expected

(or expected to be marginal)

However, a high level of deleterious

variability exists in the human genome

Variants predicted to severely affect the function

of human protein coding genes (known as loss-of-

function -LOF- variants) were thought:

To have a potential deleterious effect

To be associated to severe Mendelian

disease

An unexpectedly large number of LOF variants

have been found in the genomes of apparently

healthy individuals: 281-515 missense

substitutions per individual, 40-85 of them in

homozygous state and predicted to be highly

damaging.

A similar proportion was observed in miRNAs

and possibly affect to any functional element in

the genome

Such apparently deleterious mutation must be first detected and

then distinguished from real pathological mutations

So far so good… but is the single-gene

approach realistic? Can we detect such

disease-related variants so easily?

There are several problems:

a) There is a non-negligible amount of apparently deleterious variants

that seem to have no pathologic effect (effect filter fails)

b) Sometimes we are not targeting rare but common variants (which

occur in normal population) (MAF filter fails)

c) Interrogating 60Mb sites produces too many variants. Many of these

segregating with our experimental design (segregation filter fails)

d) In many cases only one variant does not explain the disease but rather

a combination of them (single-gene concept fails)

That results in a low ratio of discovery of variants associated to the

disease, that usually account for a small portion of the trait heritability

Clear individual gene associations are difficult to find for most diseases

Affected cases in complex diseases actually conform a heterogeneous population with different mutations (or combinations).

Many cases and controls are needed to obtain a few significant associations (more prevalent, often population-specific variants, non-reproducible).

The only common element is the (know or unknown) functional module affected.

Disease understood as the failure of a functional module

Cases Controls

Modular nature of human

genetic diseases • With the development of systems biology, studies have shown that

phenotypically similar diseases are often caused by functionally related

genes, being referred to as the modular nature of human genetic

diseases (Oti and Brunner, 2007; Oti et al, 2008).

• This modularity suggests that causative genes for the same or

phenotypically similar diseases may generally reside in the same

biological module, either a protein complex (Lage et al, 2007), a sub-

network of protein interactions (Lim et al, 2006) , or a pathway (Wood et

al, 2007)

Disease genes are close in the interactome

Goh 2007 PNAS

Same disease

in different

populations is

caused by

different genes

affecting the

same functions Fernandez 2013

Orphanet J Rare

Dis. 2013

Pipeline of data analysis Primary processing

Initial QC

FASTQ file

Mapping

BAM file

Variant calling

VCF File

Knowledge-based prioritization

Proximity to other known disease genes

Functional proximity

Network proximity

Burden tests

Other prioritization methods

Secondary analysis

(Heuristic filtering)

Variant annotation

Filtering by effect

Filtering by MAF

Filtering by family segregation

Primary analysis

Gene prioritization

Defining functional modules with the interactome

SNPs

WES/WGS

Gene

expression

Using protein interaction

networks as an scaffold to

interpret the genomic data in

a functionally-derived context

AND/OR

What part of the

interactome is active

and/or is damaged

CHRNA7 (rs2175886 p = 0.000607)

IQGAP2 (rs950643 p = 0.0003585)

DLC1 (rs1454947 p = 0.007526)

SNPs validated in independent cohorts

Network analysis helps to find disease genes

Network analysis and the 3D interactome in cancer

Positively selected mutations in cancer tend to affect interaction

domains, and preferentially in proteins central in the interactome

Porta-Pardo et al., 2015 PLoS Comp Biol

Fig 5. Mutations in TP53 interface predict patient's outcomes.

Porta-Pardo E, Garcia-Alonso L, Hrabe T, Dopazo J, Godzik A (2015) A Pan-Cancer Catalogue of Cancer Driver Protein Interaction

Interfaces. PLoS Comput Biol 11(10): e1004518. doi:10.1371/journal.pcbi.1004518

http://journals.plos.org/ploscompbiol/article?id=info:doi/10.1371/journal.pcbi.1004518

Fig 3. Network properties of interface-driver genes.

Porta-Pardo E, Garcia-Alonso L, Hrabe T, Dopazo J, Godzik A (2015) A Pan-Cancer Catalogue of Cancer Driver Protein Interaction

Interfaces. PLoS Comput Biol 11(10): e1004518. doi:10.1371/journal.pcbi.1004518

http://journals.plos.org/ploscompbiol/article?id=info:doi/10.1371/journal.pcbi.1004518

A total of 103

genes with

potentially

driver

interfaces were

found (73 of

them yet

undescribed)

From gene-based to

mechanism-based perspective Transforming gene expression values into another value that accounts for a function. Easiest example of modeling function: signaling pathways. Function: transmission of a signal from a receptor to an effector

Receptors Effectors Important assumption:

collective changes in

gene expression within

the context of a

signaling circuit are

proxies of changes in

protein activation

What would you

predict about the

consequences of

gene activity changes

in the apoptosis

pathway in a case

control experiment of

colorectal cancer?

The figure shows the

gene up-regulations

(red) and down-

regulations (blue)

The effects of changes in gene

activity are not obvious

Apoptosis

inhibition is

not obvious

from gene

expression

Two of the three possible sub-

pathways leading to apoptosis

are inhibited in colorectal

cancer. Upper panel shows the

inhibited sub-pathways in blue.

Lower panel shows the actual

gene up-regulations (red) and

down-regulations (blue) that

justify this change in the activity

of the sub-pathways

Pathway analysis helps, in addition, to

understand disease mechanisms

Fanconi Anemia is a rare chromosome instability syndrome characterized by aplastic

anemia and cancer and leukemia susceptibility. It has been proposed that disruption of the

apoptotic control, a hallmark of FA, accounts for part of the phenotype of the disease.

No

proliferation

No

degradation

Survival

No

degradation

No

apoptosis

Activation

apoptosis

pathway

Cases and controls: Different signaling

As tumor grade / stage

progresses…

Up

• Cell survival

• Proliferation, differentiation

• Cell cycle progression

• Antiapoptosis

Down

• Glucose homeostasis

• Metabolism

• Degradation

Tumor grade / stage

Signaling changes and functionalities affected in

543 Kidney Renal Clear Cell Carcinoma samples

from ICGC/TCGA database

Using circuit activation probabilities

as features for prediction

Circuit activation

probabilities are

mechanism-based

biomarkers

1. Generation of features: signaling

circuit activities 2. Training set

3. Prediction

Prediction of IC50 values from the

activity of signaling circuits

Is gene expression a good proxy

for protein activity?

However, collective changes

in gene expression within the

context of a signalling circuit

can be considered good

proxies of changes in protein

activation

The scarce phosphoproteomic

data available are consistent

with circuit activations

predicted from gene

expression changes

The observation of the expression of a gene does not guarantee that

the corresponding protein is present and active.

Circuit TNF-NPY from the Adipocytokine signaling pathway (hsa04920).

Nodes in red contain proteins found to be hyperphosphorylated when

comparing treated versus untreated lung cancer cells. Nodes in blue

contain proteins found to be dephosphorylated in the same comparison.

Bio

logic

al variabili

ty

Technic

al variabili

ty

Technical variability is enormously

reduced at the pathway level Human neuroblastoma cell lines.

RNA-seq vs microarrays

Errors across genes seem

to be cancelled within the

signaling circuits

Gene1

Gene2

Gene3

Gene4

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

Gene22000

Raw measurement

Transformed

measurement:

Receptor to effector

signaling activities

Circuit1

Circuit2

Circuit3

Circuit4

:

:

:

:

:

:

:

:

Circuit800

:

:

:

Abstraction levels when converting

genomic data into functional information

Second

transformation:

All receptors to

final effector

signaling

activities

Third

transformation:

Final effector

activities to

functions triggered

Some

functions

can be

cancer

hallmarks

Gaining insight into disease

mechanisms and MoA From gene-based to function-based analysis perspective

SNPs, gene

expression,

etc.

GO

Protein

interaction

networks

Models of

cellular

functions

Detection

power

Low (only

very

prevalent

genes)

High High Very high

Information

coverage

Almost all Almost all Low (~9000

genes in

human)

Low (~6700

genes in

human)*

Use Biomarker Illustrative,

give hints

Biomarker Biomarker

that explain

disease

mechanism

*Only ~800 genes in human signaling pathways

The use of new algorithms that enable the transformation of genomic

measurements into cell functionality measurements that account for

disease mechanisms and for drug mechanisms of action will ultimately

allow the real transition from today’s empirical medicine to precision

medicine and provide increasingly personalized medicine

The real transition to precision medicine

Intuitive Based on trial

and error

Identification of probabilistic

patterns

Decisions and actions based on knowledge

Intuitive Medicine Empirical Medicine Precision Medicine

Today Tomorrow

Degree of personalization

Software available

See interactive map of for the last 24h use http://bioinfo.cipf.es/toolsusage Babelomics is the third most cited tool for functional analysis. Includes more than 30 tools for advanced, systems-biology based data analysis

More than 150.000 experiments were analyzed in our tools during the last year

HPC on CPU, SSE4, GPUs on NGS data processing Speedups up to 40X

Genome maps is now part

of the ICGC data portal

Ultrafast genome viewer with google technology

Mapping

Visualization

Functional analysis

Variant annotation

CellBase

Knowledge

database

Variant

prioritization

NGS

panels

Signaling network Regulatory

network Interaction

network

Diagnostic

The Computational Genomics Department at the

Centro de Investigación Príncipe Felipe (CIPF),

Valencia, Spain, and… ...the INB, National

Institute of Bioinformatics

(Functional Genomics Node)

and the BiER (CIBERER Network of

Centers for Rare Diseases)

@xdopazo

@bioinfocipf

Follow us on twitter