Research Advisor |
Project Title |
|||
|
|
Function Guided Clustering of Protein-Protein Interaction Networks | ||
| Brian Burg | Dr. Jan Vitek | Evolutionary languages and programs in Th |
Recent technological advancements have enabled biologists to collect very large
datasets of protein-protein interactions. These results are typically represented by
undirected graphs. In these graphs, proteins are represented as nodes while
interactions between proteins are represented as edges between these nodes. Since
proteins often function in clusters, identifying protein clusters is often the goal of
constructing these graphs. Unfortunately, identifying protein clusters is difficult to
do when only considering the topology of an interaction graph. The techniques used
to observe protein-protein interactions often result in a high percentage of false
positives. For example, it is impossible for proteins which are never found in the
same region of the cell to interact. In many experimental techniques, however, cell
structure is compromised and these proteins are allowed to make contact, possibly
resulting in an observed interaction. To combat these false positives, additional
information about the proteins can be applied to the graph. In this study, we explore
a number of topology based clustering algorithms. We then examine these results in
the context of biological information known about the proteins from the Gene
Ontology. Finally, we propose a technique for identifying a stopping point for the
clustering using the gap statistic of our ontology enriched graph and a label
randomized version.
Evolutionary languages and programs in Th
Brian Burg, Computer Science
Research Mentor : Jan Vitek, Computer Science
(back to top)
Recently domain-specific languages have become an important topic in
programming languages research. Domain-specific languages are tailored to specific
use cases, and simplify program development within a certain domain. Such
customized languages are often expensive to create from scratch, so easily extensible
languages and compilers for these languages provide a quicker development model.
We introduce Th, a core scripting language and extensible compiler system designed
to enhance support for language customization, gradual script-program refinement,
and rapid development via object-oriented and functional language features.
Modifications to the core language are implemented via an extensible compiler
plugin framework. The Thorn language is a superset of Th, and includes additional
features, libraries, and default plugins. Finally, implementations of example
extensions and applications are presented to illustrate the utility of the extensible
framework.