Identification of pathological mutations in Fabry disease Casandra Riera Translational Bioinformatics in Neuroscience Prof. Xavier de la Cruz
Translational bioinformatics at VHIR: Understanding molecular damage in Fabry disease
Jul 08, 2015
Casandra Riera's predoctoral presentation at the 6th VHIR Scientific Session
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Identification of
pathological mutations
in Fabry disease
Casandra Riera
Translational Bioinformatics in Neuroscience
Prof. Xavier de la Cruz
Towards personalized medicine
Exome
Sequencing Sample
Variant
identification
and Quality control
Introduction
Goals
Methods
Results
Future
INTERPRETATION
?
AUTOMATIZATION
Enormous amounts
of raw data…
Identification
Prioritization
Fast & reliable
Exome-ready mutation annotation tools
Introduction
Goals
Methods
Results
Future
Primary tools
PolyPhen-2 (Adzhubei et al.,2010), SIFT (Kumar et al.,2009)
Average over many mutations and genes
Consensus methods CONDEL (González-Pérez & López-Bigas, 2011),
CAROL (Lopes et al., 2012)
Depend on the existance of primary tools
Fabry disease • Systemic disorder characterized by: progressive renal failure, cardiovascular or cerebrovascular disease, etc. • Caused by mutations in lysosomal enzyme α-galactosidase A.
Structure of a monomer of alpha-galactosidase A. In red mutation at residue 279.
Introduction
Goals
Methods
Results
Future
Gene-specific predictors to identify pathological/neutral mutations
Fabry disease
HCM
AT-III deficiency
…
General applications
Disease specific
applications
Example: M42L Example: M42L
DATA MINING
CHARACTERIZE
PROTEIN
DAMAGE
(2, 0.63, -1.45, 4.27, 0.04)
Pathological/Neutral
mutations
Neural Network training and testing
Example: M42L
Sequence properties Evolutionary properties properties
Neutral
Introduction
Goals
Methods
Results
Future
Pathological
BUILD
COMPUTATIONAL
MODEL APPLICATION: SCORE
PRIORITIZE
General & Specific* db VHIR collections Literature Close homologs
*fabry-database.org
Recover family members with PsiBlast (E-value:0.001; seq.id.>40%) UniRef100
Align with MUSCLE (R. Edgar, 2004)
Introduction
Goals
Methods
Results
Future Evolutionary-based properties: the MSA
Our model 7 properties: sequence-based (DV, Df, Blos62), structure (relative accessibility), MSA-based (entropy, pssm(wt), pssm(mt)) Neural networks (Weka package)
Multilayer percetron (1 hidden layer-4 units) No hidden layer
Training: 2 fold cross validation scheme 25 replicas to assess performance
Introduction
Goals
Methods
Results
Future What we’ve got…
Mutation dataset: 313 pathological & 59 neutral mutations Discriminant power of parameters
Introduction
Goals
Methods
Results
Future What we’ve got…
Performance of the method – ROC Curves
Introduction
Goals
Methods
Results
Future What we’ve got…
Performance of the method – Success Rate and MCC
Introduction
Goals
Methods
Results
Future What we’ve got…
Performance of the method
Qtot Sensitivity Specificity MCC
GLA-specific 0.91 0.85 0.92 0.69
Polyphen-2 0.88 0.87 0.89 0.65
MYH7-specific 0.87 0.94 0.84 0.73
Polyphen-2 0.81 0.82 0.81 0.62
Introduction
Goals
Methods
Results
Future Future directions
Extend to more genes
Can we predict other disease
phenotypes? First tests suggest a similar
approach could work for
severity
TRANSLATIONAL BIOINFORMATICS
IN NEUROSCIENCES GROUP
Xavier de la Cruz
Sergio Lois
Montserrat Barbany
NEUROVASCULAR DISEASE,
NEUROSCIENCES
Joan Montaner
Israel Fernández-Cadenas
NANOMED. LYSOS. STORAGE DIS.,
CIBBIM, NANOMEDICINE
M.Carmen Domínguez
Related Documents
