About this Event
Add to calendar
Ananthan Nambiar '19 (University of Illinois, Urbana-Champaign)
"Transforming the Language of Life: Transformer Neural Networks for Protein Prediction Tasks."
Abstract: The scientific community is rapidly generating protein sequence information, but only a fraction of these proteins can be experimentally characterized. While promising deep learning approaches for protein prediction tasks have emerged, they have computational limitations or are designed to solve a specific task. In this talk, I will introduce two protein prediction tasks: protein family classification and protein interaction prediction. I will then present a natural language processing inspired Transformer neural network that pre-trains task-agnostic sequence representations. This model is fine-tuned to solve protein family classification and protein interaction prediction with state-of-the-art results. These results offer a promising framework for fine-tuning the pre-trained sequence representations for other protein prediction tasks.
Tobias Rubel Janssen '19 (Reed College)
"A Higher Order Approach to Pathway Reconstruction"
Abstract: Signaling pathways drive cellular response and understanding such pathways is fundamental to molecular systems biology. A mounting volume of experimental protein interaction data has motivated the development of algorithms to computationally reconstruct signaling pathways. However, existing methods suffer from low recall in recovering protein interactions in ground truth pathways, limiting our confidence in any new predictions for experimental validation. I'll introduce the Pathway Reconstruction AUGmentor (PRAUG), a higher order function which generates pathway reconstruction methods able to outperform their unaugmented counterparts.