Protein secondary structure can be regarded as an information bridge that links the primary sequence and tertiary structure. Accurate 8-state prediction significantly give more precise high resolution on structure-based properties analysis.We present a novel deep learning architecture which exploits integrative synergy of by convolutional neural network, residual bidirectional recurrent network to improve performance protein prediction. A local block comprised filters original input is...

Residue solvent accessibility is closely related to the spatial arrangement and packing of residues. Predicting a protein an important step understand its structure function. In this work, we present deep learning method predict residue accessibility, which based on stacked bidirectional recurrent neural network applied sequence profiles. To capture more long-range information, merging operator was proposed when information from hidden nodes merged for outputs. Three types operators were...

The rapid development of Next-Generation Sequencing technologies enables sequencing genomes with low cost. dramatically increasing amount data raised crucial needs for efficient compression algorithms. Reference-based algorithms have exhibited outstanding performance on compressing single genomes. However, the more challenging and useful problem a large collection n genomes, straightforward application these reference-based suffers series issues such as difficult reference selection...

In order to keep up with the computer science development and national informationization, basic education has been deeply reformed in domestic college. After investigating present situation of its reformation, a new reformatting scheme is proposed. Basic college should be promoted information technology literacy education, which students' practice ability will strengthened. this scheme, teaching process divided into three stages: application specialty education. And modularized courses are...

As a sunrise and national supporting industry, the current specialty of computer science technology is in an awkward situation. Many students cannot get offer, but enterprises want to employ thousands talents. The reason not difficult find out that abilities undergraduates meet industrial demands. So reform education imperative. Industry-linked practice-oriented cultivating mode need be established. Responding Chinese "excellent engineer plan", scheme excellent software proposed. scenario...

The dihedral angle of the protein backbone can describe main structure protein, which is great significance for determining structure. Many computational methods have been proposed to predict this critically important structure, including deep learning. However, these heavyweight require more resources, and training time becomes intolerable. In article, we introduce a novel lightweight method, named dilated convolution multi-head attention (DCMA), that predicts torsion angles

Abstract Protein structural properties are diverse and have the characteristics of spatial hierarchy, such as secondary structures, solvent accessibility backbone angles. tertiary structures formed in close association with these features. Separate prediction has been improved increasing number samples protein advances machine learning techniques, but concurrent tightly related features is more useful to understand overall structure functions. We introduce a multi-task deep method for angles...