Inference of biomolecular interactions from sequence data

This thesis describes our work on the inference of biomolecular interactions from
sequence data. In particular, the first part of the thesis focuses on proteins and
describes computational methods that we have developed for the inference of both
intra- and inter-protein interactions from genomic data. The second part of the thesis
centers around protein-RNA interactions and describes a method for the inference of
binding motifs of RNA-binding proteins from high-throughput sequencing data.
The thesis is organized as follows. In the first part, we start by introducing a
novel mathematical model for the characterization of protein sequences (chapter 1).
We then show how, using genomic data, this model can be successfully applied to
two different problems, namely to the inference of interacting amino acid residues
in the tertiary structure of protein domains (chapter 2) and to the prediction of
protein-protein interactions in large paralogous protein families (chapters 3 and 4).
We conclude the first part by a discussion of potential extensions and generalizations
of the methods presented (chapter 5).
In the second part of this thesis, we first give a general introduction about RNA-
binding proteins (chapter 6). We then describe a novel experimental method for the
genome-wide identification of target RNAs of RNA-binding proteins and show how
this method can be used to infer the binding motifs of RNA-binding proteins (chapter
7). Finally, we discuss a potential mechanism by which KH domain-containing RNA-
binding proteins could achieve the specificity of interaction with their target RNAs
and conclude the second part of the thesis by proposing a novel type of motif finding
algorithm tailored for the inference of their recognition elements (chapter 8).