Publications of Oliver Ratmann :chronological combined listing:
%% Papers Published
@article{Ratmann2010,
Author = {Ratmann O and Andrieu C and Wiuf C and Richardson
S},
Title = {Reply to {R}obert et al.: {M}odel criticism informs model
choice and model comparison},
Journal = {Proc Natl Acad Sci USA},
Volume = {107},
Number = {3},
Pages = {E6-E7},
Year = {2010},
Month = {January},
url = {http://dx.doi.org/10.1073/pnas.0912887107},
Abstract = {Robert, Mengersen, and Chan (RMC) represent our approach to
model criticism in situations when the likelihood cannot be
computed (1) as a way to “contrast several models with
each other” (2, 3). Moreover, RMC argue that model
assessment with Approximate Bayesian Computation under model
uncertainty (ABCμ) is unduly challenging and question its
Bayesian foundations. We disagree, and clarify that ABCμ is
a probabilistically sound and powerful tool for criticizing
a model against aspects of the observed data.},
Doi = {10.1073/pnas.0912887107},
Key = {Ratmann2010}
}
@article{Ratmann2009c,
Author = {Ratmann O and Wiuf C and Pinney JW},
Title = {From evidence to inference: probing the evolution of protein
interaction networks},
Journal = {HFSP Journal},
Volume = {3},
Number = {5},
Pages = {290-306},
Year = {2009},
Month = {December},
url = {http://dx.doi.org/10.2976/1.3167215},
Abstract = {The evolutionary mechanisms by which protein interaction
networks grow and change are beginning to be appreciated as
a major factor shaping their present-day structures and
properties. Starting with a consideration of the biases and
errors inherent in our current views of these networks, we
discuss the dangers of constructing evolutionary arguments
from naïve analyses of network topology. We argue that
progress in understanding the processes of network evolution
is only possible when hypotheses are formulated as plausible
evolutionary models and compared against the observed data
within the framework of probabilistic modeling. The value of
such models is expected to be greatly enhanced as they
incorporate more of the details of the biophysical
properties of interacting proteins, gene phylogeny, and
measurement error and as more advanced methodologies emerge
for model comparison and the inference of ancestral network
states.},
Doi = {10.2976/1.3167215},
Key = {Ratmann2009c}
}
@article{Ratmann2009,
Author = {Ratmann O and Andrieu C and Wiuf C and Richardson
S},
Title = {Model criticism based on likelihood-free inference, with an
application to protein network evolution},
Journal = {Proc Natl Acad Sci USA},
Volume = {106},
Number = {26},
Pages = {10576-10581},
Year = {2009},
Month = {June},
url = {http://dx.doi.org/10.1073/pnas.0807882106},
Abstract = {Mathematical models are an important tool to explain and
comprehend complex phenomena, and unparalleled computational
advances enable us to easily explore them without any or
little understanding of their global properties. In fact,
the likelihood of the data under complex stochastic models
is often analytically or numerically intractable in many
areas of sciences. This makes it even more important to
simultaneously investigate the adequacy of these models—in
absolute terms, against the data, rather than relative to
the performance of other models—but no such procedure has
been formally discussed when the likelihood is intractable.
We provide a statistical interpretation to current
developments in likelihood-free Bayesian inference that
explicitly accounts for discrepancies between the model and
the data, termed Approximate Bayesian Computation under
model uncertainty (ABCμ). We augment the likelihood of the
data with unknown error terms that correspond to freely
chosen checking functions, and provide Monte Carlo
strategies for sampling from the associated joint posterior
distribution without the need of evaluating the likelihood.
We discuss the benefit of incorporating model diagnostics
within an ABC framework, and demonstrate how this method
diagnoses model mismatch and guides model refinement by
contrasting three qualitative models of protein network
evolution to the protein interaction datasets of
Helicobacter pylori and Treponema pallidum. Our results make
a number of model deficiencies explicit, and suggest that
the T. pallidum network topology is inconsistent with
evolution dominated by link turnover or lateral gene
transfer alone.},
Doi = {10.1073/pnas.0807882106},
Key = {Ratmann2009}
}
@article{Surmeli2008,
Author = {Surmeli D and Ratmann O and Mewes HW and Tetko
IV},
Title = {FunCat functional inference with belief propagation and
feature integration},
Journal = {Comput Biol Chem},
Volume = {32},
Number = {5},
Pages = {375-7},
Year = {2008},
Month = {August},
url = {http://dx.doi.org/10.1016/j.compbiolchem.2008.06.004},
Abstract = {Pairwise comparison of sequence data is intensively used for
automated functional protein annotation, while graphical
models emerge as promising candidates for an integration of
various heterogeneous features. We designed a model, termed
hRMN that integrates different genomic features and
implemented a variant of belief propagation for functional
annotation transfer. hRMN allows the assignment of multiple
functional categories while avoiding common problems in
annotation transfer from heterogeneous datasets, such as an
independency of the investigated datasets. We benchmarked
this system with large-scale annotation transfer (based on
the MIPS FunCat ontology) to proteins of the prokaryotes
Bacillus subtilis, Helicobacter pylori, Listeria
monocytogenes, and Listeria innocua. hRMN consistently
outperformed two competitors in annotation of four bacterial
genomes. The developed code is available for download at
http://mips.gsf.de/proj/bfab/hRMN.html.},
Doi = {10.1016/j.compbiolchem.2008.06.004},
Key = {Surmeli2008}
}
@article{Ratmann2007,
Author = {Ratmann O and Jorgensen O and Hinkley T and Stumpf MPH and Richardson S and Wiuf C},
Title = {Using Likelihood-Free Inference to Compare Evolutionary
Dynamics of the Protein Networks of H. pylori and P.
falciparum},
Journal = {PLoS Comp Biol},
Volume = {3},
Number = {2007},
Pages = {e230},
Year = {2007},
Month = {November},
url = {http://dx.doi.org/10.1371/journal.pcbi.0030230},
Abstract = {Gene duplication with subsequent interaction divergence is
one of the primary driving forces in the evolution of
genetic systems. Yet, little is known about the precise
mechanisms and the role of duplication-divergence in the
evolution of protein networks from the prokaryote and
eukaryote domains. We developed a novel, model-based
approach for Bayesian inference on biological network data
that centres on Approximate Bayesian Computation, or
Likelihood-Free Inference. Instead of computing the
intractable likelihood of the protein network topology, our
method summarizes key features of the network and, based on
these, uses a MCMC algorithm to approximate the posterior
distribution of the model parameters. This allowed us to
reliably fit a flexible mixture model that captures
hallmarks of evolution by gene duplication and
subfunctionalization to PIN data sets of Helicobacter pylori
and Plasmodium falciparum. The 80% credible intervals for
the duplication-divergence component are [0.6,0.98] for H.
pylori and [0.89,0.99] for P. falciparum. The remaining
parameter estimates are not inconsistent with sequence data.
An extensive sensitivity analysis showed that incompleteness
of PIN data does not largely affect the analysis of models
of protein network evolution, and that the degree sequence
alone barely captures the evolutionary footprints of protein
networks relative to other statistics. Our likelihood-free
inference approach enables a fully Bayesian analysis of a
complex and highly stochastic system that is otherwise
intractable at present. Modelling the evolutionary history
of PIN data, it transpires that only the simultaneous
analysis of several global aspects of protein networks
enables credible and consistent inference to be made from
available data sets. Our results indicate that gene
duplication has played a larger part in the network
evolution of the eukaryote than in the prokaryote, and
suggests that single gene duplications with immediate
divergence alone may explain more than 60% of biological
network data in both domains.},
Doi = {10.1371/journal.pcbi.0030230},
Key = {Ratmann2007}
}
%% Book Chapters
@inbook{Wiuf2009,
Author = {Wiuf C and Ratmann O},
Title = {Statistical and Evolutionary Analysis of Biological
Networks},
Pages = {17-49},
Publisher = {Imperial Press London},
Editor = {Stumpf MPH and Wiuf C},
Chapter = {Evolutionary Analysis of Protein Interac},
Year = {2009},
Abstract = {Systems approaches to understanding the structure,
organization and functioning of organisms and cells are now
becoming commonplace. In this chapter we focus on protein
interaction networks and their potential use for inference
on the evolutionary processes that have shaped the
interactome, the collection of all proteins in a cell
together with their physical interactions. We demonstrate
that simple mathematical models may capture essential
aspects of the processes and use these to develop a Bayesian
Likelihood-free scheme for inference on three small
organisms T. pallidum, H. pylori, and P.
falciparum.},
Key = {Wiuf2009}
}
%% Articles
@article{Ratmann2009d,
Author = {Ratmann O and Andrieu C and Wiuf C and Richardson
S},
Title = {Notes to {R}obert et al.: {M}odel criticism informs model
choice and model comparison},
Journal = {arXiv},
Year = {2009},
Month = {December},
url = {http://arxiv.org/abs/0912.3182},
Abstract = {In their letter to PNAS and a comprehensive set of notes on
arXiv [arXiv:0909.5673v2], Christian Robert, Kerrie
Mengersen and Carla Chen (RMC) represent our approach to
model criticism in situations when the likelihood cannot be
computed as a way to "contrast several models with each
other". In addition, RMC argue that model assessment with
Approximate Bayesian Computation under model uncertainty
(ABCmu) is unduly challenging and question its Bayesian
foundations. We disagree, and clarify that ABCmu is a
probabilistically sound and powerful too for criticizing a
model against aspects of the observed data, and discuss
further the utility of ABCmu.},
Key = {Ratmann2009d}
}
@inproceedings{Ratmann2009b,
Author = {Ratmann O and Andrieu C and Wiuf C and Richardson
S},
Title = {Likelihood-free model criticism applied to stochastic models
of protein network evolution},
Volume = {48},
Pages = {143-146},
Booktitle = {Proceedings of the Sixth International Workshop on
Computational Systems Biology},
Editor = {Manninen T and Wiuf C and Laehdesmaeki CH and Grzegorczyk M and Rahnenfuhrer J and Ahdesmaki M and Linne MJ and Yli-Harja
O},
Year = {2009},
url = {http://www.birc.au.dk/~wiuf/journalWiuf/otherScientific/TICSP2009.pdf},
Abstract = {Various models of network evolution are used in systems
biology with little or no statistical justification. These
may be described as generative as they are framed in terms
of incremental network growth, node-by-node, on an abstract,
discrete timeline, and neglect the actual history of the
proteome. However, despite their simplicity the likelihood
of an observed network is computationally difficult to
calculate; thus goodness-of-fit and model criticism are
difficult to perform and rarely attempted. Here, we provide
a general Bayesian framework for model criticism and apply
it to two real PIN data sets using different models of
network evolution. Using novel statistical tech- niques
appropriate for the analysis of generative models of network
evolution with network data, our results support the view
that (1) even in abstract terms network evolution cannot be
understood as a process of repeated preferential attachment
and that (2) a better appreciation of the vari- ous forms of
measurement error and the reported data are essential to
making inference of existing network data.},
Key = {Ratmann2009b}
}
@inproceedings{Wiuf2008,
Author = {Wiuf C and Ratmann O and Knudsen M},
Title = {Analysis of biological network data using likelihood-free
inference techniques},
Volume = {41},
Pages = {185-188},
Booktitle = {TICSP Series},
Editor = {Ahdesmaki M and Strimmer K and Radde N and Rahnenfuehrer J and Klemm K and Laehdesmaeki CH and Yli-Harja O},
Year = {2008},
url = {http://www.birc.au.dk/~wiuf/journalWiuf/otherScientific/TICSP2008.pdf},
Abstract = {Biological Networks have received much attention in recent
years, but statistical tools for network analysis are still
in their infancy. In this paper we focus on Protein
Interaction Networks (PINs) that typically comprise
thousands of proteins and interactions. PINs are the result
of long evolutionary histories. Here we adopt simple
mathematical models that capture essentials of protein
evolution and develop statistical methods to estimate
evolutionary PIN parameters. Our initial approach is based
on a recursion for the likelihood, but it becomes
computationally intractable for reasonably sized networks.
Our second approach is based on summary statistics and
likelihood-free inference. We discuss problems with
selection of summaries, convergence, and credibility and
apply the methods on Helicobacter pylori and Plasmodium
falciparum data.},
Key = {Wiuf2008}
}
@inproceedings{Wiuf2007,
Author = {Wiuf C and Ratmann O},
Title = {Statistical analysis of biological network
data.},
Pages = {147-156},
Booktitle = {Symposium i Anvendt Statistik},
Editor = {P Linde},
Year = {2007},
url = {http://www.birc.au.dk/~wiuf/journalWiuf/otherScientific/StatSymp2007.pdf},
Key = {Wiuf2007}
}
Duke Biology Box 90338 Durham, NC 27708 Phone: 919-660-7372 Fax: 919-660-7293