Ye Lab, A&S - Mathematics

Protein contact map prediction

Description: Understanding of protein structure and its activities plays a major role in drug design. Protein contact map contains key information that helps to understand the structure and functioning of a protein. This research project develops an RNN and CNN based neural network to infer contact map from the protein sequence (amino acid sequence of the protein) together with one dimensional features (predicted secondary structure and solvent accessibility) and two dimensional features (i.e., co-evolution information, pairwise contact and distance potential).

Computational methods:

Our proposed method is to train a neural network based on a combination of RNNs and CNNs. Our implementation is in Keras with TensorFlow as back-end to predict the contact map given the sequence and feature vectors.

Students

K.D.Gayan Maduranga
Kyle Helfrich

Vasily I Zadorozhnyy

Cole M Pospisil, Added 02/18/2021

Edison, Mucllari, Added 08/13/2021

Software:

Python
Keras
ensorFlow
pdbxer
biopython
numPy
sciPy
scikit learn
matplotlib libraries

Grants

Qiang Ye, National Science Foundation DMS-1821144, CDSE: Ecient and Robust Recurrent Neural Networks, 2018-2021

Publications

(resulting from DLX usage):
none

Collaborators

Grants

Ye, Qiang DMS-1317424 Collaborative Research: Robust Algorithms for Interpolation and Extrapolation in Manifold Learning National Science Foundation 9/1/2013 - 8/31/2016 $139,907
Ye, Qiang DMS-1318633 Accurate and Efficient Algorithms for Computing Exponentials of Large Matrices with Applications National Science Foundation 7/15/2013 - 6/30/2016 $189,971