Thumbnails:
List:
Year:
Category:
Session:
Poster:
Getting poster data...
Evan Molinelli, Martin Miller, Arman Aksoy (Computational Biology Center, Memorial Sloan Kettering Cancer Center, New York, NY USA)
Motivation: De novo inference of cellular signaling network models from rich drug-perturbation datasets is an underdetermined problem and produces thousands of equally fitting network models. Here, we introduce a visualization tool (NetTrapper) to interactively explore user-defined network features on all models and simultaneously visualize a user-selected subset of these models. Technical details: The input to NetTrapper consists of (1) a large set of network models with parameterized interaction strengths, and (2) a set of pre-calculated graph-theoretical features stored in an organized file architecture. A Processing applet allows users to query these features and display them on simple, but popular parallel coordinate plots. Users can add, remove, and rearrange features as well as select a subset of network models to visualize in a separate window. The interactive network viewer, itself, is a unique way of visually identifying patterns on a collection of network models in a single view. Results: The ability to visually explore features can assist in developing better network inference algorithms and extract meaningful biological results from models. For example, we see i