Identifying complex motifs in massive omics data with a variable-convolutional layer in deep neural network

Abstract

Motif identification is among the most common and essential computational tasks for bioinformatics and genomics. Here we proposed a novel convolutional layer for deep neural network, named variable convolutional (vConv) layer, for effective motif identification in high-throughput omics data by learning kernel length from data adaptively. Empirical evaluations on DNA-protein binding and DNase footprinting cases well demonstrated that vConv-based networks have superior performance to their convolutional counterparts regardless of model complexity. Meanwhile, vConv could be readily integrated into multi-layer neural networks as an ‘in-place replacement’ of canonical convolutional layer. All source codes are freely available on GitHub for academic usage.

Prediction and Planning of Sports Competition Based on Deep Neural Network

“…The indexes of the optimized objective function are data rate, signal-to-interference-noise ratio (SINR), power consumption, and energy efficiency, and then the optimized objective function is allocated to a neural network for resource allocation [ 2 ]. It expounds a new deep neural network convolution layer-variable convolution (vConv) layer, which learns the kernel length of data adaptively by its own cycle to realize the motif recognition of data sets with high throughput [ 3 ]. It proves the effectiveness of pretraining neural networks on different data sets and shows that in many practical cases, the convolution layer can be replaced by a smaller fully connected layer, and the accuracy degradation is relatively small [ 4 ].…”