RNA 3D structure prediction remains challenging though after years of efforts. Inspired by the recent breakthrough in protein structure prediction, we developed trRosettaRNA, a novel deep learning-based approach to de novo prediction of RNA 3D structure. Like trRosetta, the trRosettaRNA pipeline comprises two major steps: 1D and 2D geometries prediction by a transformer network; and full-atom 3D structure folding by energy minimization with constraints from the predicted geometries. We benchmarked trRosettaRNA on two independent datasets. The results show that trRosettaRNA outperforms other conventional methods by a large margin. For example, on 25 targets from the RNA-Puzzles experiments, the mean RMSD of the models predicted by trRosettaRNA is 5.5 Å, compared with 10.5 Å from the state-of-the-art human group (i.e., Das). Further comparisons with two recently released deep learning-based methods (i.e., DeepFoldRNA and RoseTTAFoldNA) show that all three methods have similar accuracy. However, trRosettaRNA yields more accurate and physically more realistic side-chain atoms than DeepFoldRNA and RoseTTAFoldNA. Finally, we apply trRosettaRNA to predict the structures for the Rfam families that do not have known structures. Analysis shows that for 263 families, the predicted structure models are estimated to be accurate with RMSD < 4 Å. The trRosettaRNA server and the package are available at: https://yanglab.nankai.edu.cn/trRosettaRNA/.

No abstract

RNA-seq technology is widely employed in various research areas related to transcriptome analyses, and the identification of all the expressed transcripts from short sequencing reads presents a considerable computational challenge. In this study, we introduce TransRef, a new computational algorithm for accurate transcriptome assembly by redefining a novel graph model, the neo-splicing graph, and then iteratively applying a constrained dynamic programming to reconstruct all the expressed transcripts for each graph. When TransRef is utilized to analyze both real and simulated datasets, its performance is notably better than those of several state-of-the-art assemblers, including StringTie2, Cufflinks and Scallop. In particular, the performance of TransRef is notably strong in identifying novel transcripts and transcripts with low-expression levels, while the other assemblers are less effective.

Identifying significant biclusters of genes with specific expression patterns is an effective approach to reveal functionally correlated genes in gene expression data. However, none of existing algorithms can simultaneously identify both broader and narrower biclusters due to their failure of balancing between effectiveness and efficiency. We introduced ARBic, an algorithm which is capable of accurately identifying any significant biclusters of any shape, including broader, narrower and square, in any large scale gene expression dataset. ARBic was designed by integrating column-based and row-based strategies into a single biclustering procedure. The column-based strategy borrowed from RecBic, a recently published biclustering tool, extracts narrower biclusters, while the row-based strategy that iteratively finds the longest path in a specific directed graph, extracts broader ones. Being tested and compared to other seven salient biclustering algorithms on simulated datasets, ARBic achieves at least an average of 29% higher recovery, relevance and$\ {F}_1$ scores than the best existing tool. In addition, ARBic substantially outperforms all tools on real datasets and is more robust to noises, bicluster shapes and dataset types.

With the release of AlphaFold2, protein model databases are growing at an unprecedented rate. Efficient structure retrieval schemes are becoming more and more important to quickly analyze structure models. The core problem in structural retrieval is how to measure the similarity between structures. Some structure alignment algorithms can solve this problem but at a substantial time cost. At present, the state-of-the-art method is to convert protein structures into 3D Zernike descriptors and evaluate the similarity between structures by Euclidean distance. However, methods for computing 3D Zernike descriptors of protein structures are almost always based on structural surfaces and most are web servers, which is not conducive for users to analyze customized datasets. To overcome this limitation, we propose PGAR-Zernike, a convenient toolkit for computing different types of Zernike descriptors of structures: the user simply needs to enter one line of command to calculate the Zernike descriptors of all structures in a customized datasets. Compared with the state-of-the-art method based on 3D Zernike descriptors and an efficient structural comparison tool, PGAR-Zernike achieves higher retrieval accuracy and binary classification accuracy on benchmark datasets with different attributes. In addition, we show how \red{PGAR-Zernike} completes the construction of the descriptor database and the protocol used for the PDB dataset so as to facilitate the local deployment of this tool for interested readers. We construct a demonstration containing 590685 structures; at this scale, our retrieval system takes only 4 to 9 seconds to complete a retrieval, and experiments show that it reaches the state-of-the-art level in terms of accuracy. PGAR-Zernike is an open-source toolkit, whose source code and related data are accessible at \url{https://github.com/junhaiqi/PGAR-Zernike/