NAR Genomics and Bioinformatics最新文献

Small regulatory RNAs (sRNAs) are widespread in bacteria. However, characterizing the targets of sRNA regulation in a way that scales with the increasing number of identified sRNAs has proven challenging. Computational methods offer one means for efficient characterization of sRNA targets, but the sensitivity and precision of such computational methods is limited. Here, we investigate whether publicly available expression data from RNA-seq experiments can improve the accuracy of computational prediction of sRNA regulatory targets. Using compendia of 2143 Escherichia coli RNA-seq samples and 177 Salmonella RNA-seq samples, we identify groups of co-expressed genes in each organism and incorporate this expression information into computational prediction of sRNA targets based on machine learning methods. We find that integrating expression information significantly improves the accuracy of computational results. Further, we observe that computational methods perform better when trained on smaller, higher quality sets of targets rather than on larger, noisier sets of targets identified by high-throughput methods.

Ribonucleic acid (RNA) structural motif identification is a crucial step for understanding RNA structure and functionality. Due to the complexity and variations of RNA 3D structures, identifying RNA structural motifs is challenging and time-consuming. Particularly, discovering new RNA structural motif families is a hard problem and still largely depends on manual analysis. In this paper, we proposed an RNA structural motif clustering tool, named GINClus, which uses a semi-supervised deep learning model to cluster RNA motif candidates (RNA loop regions) based on both base interaction and 3D structure similarities. GINClus converts base interactions and 3D structures of RNA motif candidates into graph representations and using graph isomorphism network (GIN) model in combination with K-means and hierarchical agglomerative clustering, GINClus clusters the RNA motif candidates based on their structural similarities. GINClus has a clustering accuracy of 87.88% for known internal loop motifs and 97.69% for known hairpin loop motifs. Using GINClus, we successfully clustered the motifs of the same families together and were able to find 927 new instances of Sarcin-ricin, Kink-turn, Tandem-shear, Hook-turn, E-loop, C-loop, T-loop, and GNRA loop motif families. We also identified 12 new RNA structural motif families with unique structure and base-pair interactions.

The recent expansion of single-cell technologies has enabled simultaneous genome-wide measurements of multiple modalities in the same single cell. The potential to jointly profile such modalities as gene expression, chromatin accessibility, protein epitopes, or multiple histone modifications at single-cell resolution represents a compelling opportunity to study developmental processes at multiple layers of gene regulation. Here, we present Ocelli, a lightweight Python package implemented in Ray for scalable visualization and analysis of developmental multimodal single-cell data. The core functionality of Ocelli focuses on diffusion-based modeling of biological processes involving cell state transitions. Ocelli addresses common tasks in single-cell data analysis, such as visualization of cells on a low-dimensional embedding that preserves the continuity of the developmental progression of cells, identification of rare and transient cell states, integration with trajectory inference algorithms, and imputation of undetected feature counts. Extensive benchmarking shows that Ocelli outperforms existing methods regarding computational time and quality of the reconstructed low-dimensional representation of developmental data.

Defective interfering particles (DIPs) are viral deletion mutants that hamper virus replication and are, thus, potent novel antiviral agents. To evaluate possible antiviral treatments, we first need to get a deeper understanding of DIP characteristics. Thus, we performed a meta-analysis of 20 already published sequencing datasets of influenza A and B viruses (IAV and IBV) from in vivo and in vitro experiments. We analyzed each dataset for characteristics, such as deletion-containing viral genome (DelVG) length distributions, direct repeats, and nucleotide enrichment at the deletion site. Our analysis suggests differences in the length of the 3'- and 5'-end retained in IAV and IBV viral sequences upon deletion. Moreover, in vitro DelVGs tend to be shorter than those in vivo, which is a novel finding with potential implications for future DIP treatment design. Additionally, our analysis demonstrates the presence of DelVGs with longer than expected sequences, possibly related to an alternative mechanism of DelVG formation. Finally, a joint ranking of DelVGs originating from 7 A/Puerto Rico/8/1934 datasets revealed 11 highly abundant, yet unnoticed, candidates. Together, our study highlights the importance of meta-analyses to uncover yet unknown DelVG characteristics and to pre-select candidates for antiviral treatment design.

The digenic inheritance hypothesis holds the potential to enhance diagnostic yield in rare diseases. Computational approaches capable of accurately interpreting and prioritizing digenic combinations of variants based on the proband's phenotypes and family information can provide valuable assistance during the diagnostic process. We developed diVas, a hypothesis-driven machine learning approach that interprets genomic variants across different gene pairs. DiVas demonstrates strong performance in both classifying and prioritizing causative digenic combinations of rare variants within the top positions across 11 cases with the complete list of variants available (73% sensitivity and a median ranking of 3). Furthermore, it achieves a sensitivity of 0.81 when applied to 645 published causative digenic combinations. Additionally, diVas leverages explainable artificial intelligence to elucidate the digenic disease mechanism for predicted positive pairs.

In addition to their sequence, multiple functions of RNAs are encoded within their structure, which is often difficult to solve using physico-chemical methods. Incorporating low-resolution experimental data such as chemical probing into computational prediction significantly enhances RNA structure modeling accuracy. While medium- and high-throughput RNA structure probing techniques are widely accessible, the subsequent analysis process can be cumbersome, involving multiple software and manual data manipulation. In addition, the relevant interpretation of the data requires proper parameterization of the software and a strict consistency in the analysis pipeline. To streamline such workflows, we introduce IPANEMAP Suite, a comprehensive platform that guides users from chemically probing raw data to visually informative secondary structure models. IPANEMAP Suite seamlessly integrates various experimental datasets and facilitates comparative analysis of RNA structures under different conditions (footprinting), aiding in the study of protein or small molecule interactions with RNA. Here, we show that the unique ability of IPANEMAP Suite to perform integrative modeling using several chemical probing datasets with phylogenetic data can be instrumental in obtaining accurate secondary structure models. The platform's project-based approach ensures full traceability and generates publication-quality outputs, simplifying the entire RNA structure analysis process. IPANEMAP Suite is freely available at https://github.com/Sargueil-CiTCoM/ipasuite under a GPL-3.0 license.

Biodiversity genomics projects are underway with the aim of sequencing the genomes of all eukaryotic species on Earth. Here we describe the BioGenome Portal, a web-based application to facilitate organization and access to the data produced by biodiversity genomics projects. The portal integrates user-generated data with data deposited in public repositories. The portal generates sequence status reports that can be eventually ingested by designated metadata tracking systems, facilitating the coordination task of these systems. The portal is open-source and fully customizable. It can be deployed at any site with minimum effort, contributing to the democratization of biodiversity genomics projects. We illustrate the features of the BioGenome Portal through a number of specific instances. One such instance is being used as the reference portal for the Catalan Initiative for the Earth Biogenome Project, a regional project aiming to sequencing the genomes of the species of the Catalan linguistic area.

The spatial conformation of chromosomes and genomes of single cells is relevant to cellular function and useful for elucidating the mechanism underlying gene expression and genome methylation. The chromosomal contacts (i.e. chromosomal regions in spatial proximity) entailing the three-dimensional (3D) structure of the genome of a single cell can be obtained by single-cell chromosome conformation capture techniques, such as single-cell Hi-C (ScHi-C). However, due to the sparsity of chromosomal contacts in ScHi-C data, it is still challenging for traditional 3D conformation optimization methods to reconstruct the 3D chromosome structures from ScHi-C data. Here, we present a machine learning-based method based on a novel SO(3)-equivariant graph neural network (HiCEGNN) to reconstruct 3D structures of chromosomes of single cells from ScHi-C data. HiCEGNN consistently outperforms both the traditional optimization methods and the only other deep learning method across diverse cells, different structural resolutions, and different noise levels of the data. Moreover, HiCEGNN is robust against the noise in the ScHi-C data.

Non-canonical small open reading frames (sORFs) are among the main regulators of gene expression. The most studied of these are upstream ORFs (upORFs) located in the 5'-untranslated region (UTR) of coding genes. Internal ORFs (intORFs) in the coding sequence and downstream ORFs (dORFs) in the 3'UTR have received less attention. Different bioinformatics tools permit the prediction of single nucleotide variants (SNVs) altering upORFs, mainly those creating AUGs or deleting stop codons, but no tool predicts variants altering non-canonical translation initiation sites and those altering intORFs or dORFs. We propose an upgrade of our MORFEE bioinformatics tool to identify SNVs that may alter all types of sORFs in coding transcripts from a VCF file. Moreover, we generate an exhaustive catalog, named MORFEEdb, reporting all possible SNVs altering existing upORFs or creating new ones in human transcripts, and provide an R script for visualizing the results. MORFEEdb has been implemented in the public platform Mobidetails. Finally, the annotation of ClinVar variants with MORFEE reveals that > 45% of UTR-SNVs can alter upORFs or dORFs. In conclusion, MORFEE and MORFEEdb have the potential to improve the molecular diagnosis of rare human diseases and to facilitate the identification of functional variants from genome-wide association studies of complex traits.

While numerous methods have been developed for analyzing scRNA-seq data, benchmarking various methods remains challenging. There is a lack of ground truth datasets for evaluating novel gene selection and/or clustering methods. We propose the use of crafted experiments, a new approach based upon perturbing signals in a real dataset for comparing analysis methods. We demonstrate the effectiveness of crafted experiments for evaluating new univariate distribution-oriented suite of feature selection methods, called GOF. We show GOF selects features that robustly identify crafted features and perform well on real non-crafted data sets. Using varying ways of crafting, we also show the context in which each GOF method performs the best. GOF is implemented as an open-source R package and freely available under GPL-2 license at https://github.com/siyao-liu/GOF. Source code, including all functions for constructing crafted experiments and benchmarking feature selection methods, are publicly available at https://github.com/siyao-liu/CraftedExperiment.