Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

Interaction Propagation Matrix Completion

Talk at RECOMB'14 Conference, Pittsburgh, PA, USA.
by

Marinka Zitnik

on 24 June 2014

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Interaction Propagation Matrix Completion

Imputation Performance
Missing value abundance and distribution
Interaction propagation
matrix completion
Experimental
Setup
Predicting Genetic Interactions
in Epistatic MAPs by Interaction Propagation Matrix Completion

Low-rank probabilistic matrix completion
Marinka Zitnik and Blaz Zupan
University of Ljubljana, Slovenia
Baylor College of Medicine, Houston, TX, USA

1
1,2
1
2
Interaction propagation
matrix completion

Probabilistic model to gene interactions with any gene network
interaction data imputation method
Can also predict gene interaction for
in the original assays
Significant
over existing approaches
Conditional probability
of observed interactions
Zero-mean Gaussian prior
for gene latent features
Gene latent feature
propagation
Inference of maximum
a posteriori estimate of
the posterior
Global
Local
Matrix
completion
Knowledge
assisted
g
1
g
2
g
3
g
4
Shared Gene
Ontology terms
Physical
interaction
Missing data
in epistatic MAPs
up to ~40% missing data rate
pairs of genes for which the interaction strength could not be measured due to noise
Objective
Additional knowledge presented through gene networks
improve the correspondence
between genetic and functional similarity
Estimate the missing entries given the incomplete E-MAP score data matrix to ...
discover causality
Epistatic Miniarray Profile (E-MAP)
Beltrao
et al.
, 2010
Hoppins
et al.
, 2011
Large-scale genetic
interaction discovery
Discovering:
Schuldiner
et al.
, 2005
Collins
et al.
, 2006
Roguev
et al.
, 2008
structural complexes
biological pathways
Surma
et al.
, 2013
Schuldiner
et al.
, 2005
Collins
et al.
, 2006
Double mutants
Phenotype = colony size
Comparison with
single mutants
quantification of
genetic interactions
negative
positive
gene functions
Collins
et al.
, 2007
Aguilar
et al.
, 2010
genes
genes
Propagation of latent gene interaction profiles through gene network
gene
network
input
data
latent
factors
g
5
Our
approach
instead of correlation
New
Causality
combine
performance improvement
genes not
included
Random
Submatrix
Cross
Prediction
Easier
Harder
essential genes
join of E-MAP data sets
genes not included in
the original assay
I.I.D.
Hidden values
Correlation coefficient
MC
IP-MC Gene Ontology
IP-MC physical interaction
3.
1.
2.
Acknowledgements
Dr. Uros Petrovic
Dr. Gad Shaulsky
Funding
Dr. Blaz Zupan
Imputation accuracy
Pearson correlation and NRMSE
between predicted and true values
Full transcript