Protein Secondary Structure Prediction Using Cascaded Convolutional and Recurrent Neural Networks

Authors: Zhen Li, Yizhou Yu

IJCAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on the CB6133 dataset, the public CB513 benchmark, and the recent CASP10 and CASP11 datasets demonstrate that our proposed deep network outperforms existing methods and achieves state-of-the-art performance.
Researcher Affiliation Academia Zhen Li, Yizhou Yu Department of Computer Science, The University of Hong Kong zli@cs.hku.hk, yizhouy@acm.org
Pseudocode No The paper describes the network architecture using figures and equations but does not include explicit pseudocode or algorithm blocks.
Open Source Code Yes Our model and results are publicly available1. 1https://github.com/icemansina/IJCAI2016
Open Datasets Yes We use four publicly available datasets, CB6133 produced with PISCES Cull PDB [Wang and Dunbrack, 2003], CB513 [Cuff and Barton, 1999] 3, CASP10 [Kryshtafovych et al., 2014] and CASP11 [Moult et al., 2014], to evaluate the performance of our proposed deep neural network. 3http://www.princeton.edu/~jzthree/datasets/ICML2014/
Dataset Splits Yes CB6133 is a large non-homologous protein sequence and structure dataset, that has 6128 proteins, which include 5600 proteins (index 0 to 5599) for training, 256 proteins (index 5877 to 6132) for validation and 272 proteins (index 5605 to 5876) for testing.
Hardware Specification Yes The entire deep network is trained on a single NVIDIA Ge Force GTX TITAN X GPU with 12GB memory.
Software Dependencies No Our code is implemented in Theano [Bastien et al., 2012; Bergstra et al., 2010], a publicly available deep learning software4, on the basis of the Keras [Chollet, 2015] library5. While the software and frameworks are mentioned, specific version numbers for Theano and Keras are not provided.
Experiment Setup Yes In our experiments, multiscale CNN layers with kernel size 3, 7, and 11 are used... Each of the three stacked BGRU layers has 600 hidden units... The output from the BGRU layers is regularized with dropout (= 0.5)... We set λ1 = 1, λ2 = 0.001... The batch size is set to 128.