NAR Genomics and Bioinformatics最新文献

Biomedical research takes advantage of omic data, such as transcriptomics, to unravel the complexity of diseases. A conventional strategy identifies transcriptomic biomarkers characterized by expression patterns associated with a phenotype by relying on feature selection approaches. Hybrid ensemble feature selection (HEFS) has become increasingly popular as it ensures robustness of the selected features by performing data and functional perturbations. However, it remains difficult to make the best suited choices at each step when designing such approaches. We conducted an extensive analysis of four possible HEFS scenarios for the identification of Stage IV colorectal, Stage I kidney and lung and Stage III endometrial cancer biomarkers from transcriptomic data. These scenarios investigate the use of two types of feature reduction by filters (differentially expressed genes and variance) conjointly with two types of resampling strategies (repeated holdout by distribution-balanced stratified and random stratified) for downstream feature selection through an aggregation of thousands of wrapped machine learning models. Based on our results, we emphasize the advantages of using HEFS approaches to identify complex disease biomarkers, given their ability to produce generalizable and stable results to both data and functional perturbations. Finally, we highlight critical issues that need to be considered in the design of such strategies.

Data from the single-cell assay for transposase-accessible chromatin using sequencing (scATAC-seq) are now widely available. One major computational challenge is dealing with high dimensionality and inherent sparsity, which is typically addressed by producing lower dimensional representations of single cells for downstream clustering tasks. Current approaches produce such individual cell embeddings directly through a one-step learning process. Here, we propose an alternative approach by building embedding models pre-trained on reference data. We argue that this provides a more flexible analysis workflow that also has computational performance advantages through transfer learning. We implemented our approach in scEmbed, an unsupervised machine-learning framework that learns low-dimensional embeddings of genomic regulatory regions to represent and analyze scATAC-seq data. scEmbed performs well in terms of clustering ability and has the key advantage of learning patterns of region co-occurrence that can be transferred to other, unseen datasets. Moreover, models pre-trained on reference data can be exploited to build fast and accurate cell-type annotation systems without the need for other data modalities. scEmbed is implemented in Python and it is available to download from GitHub. We also make our pre-trained models available on huggingface for public use. scEmbed is open source and available at https://github.com/databio/geniml. Pre-trained models from this work can be obtained on huggingface: https://huggingface.co/databio.

Microbial specialized metabolite biosynthetic gene clusters (SMBGCs) are a formidable source of natural products of pharmaceutical interest. With the multiplication of genomic data available, very efficient bioinformatic tools for automatic SMBGC detection have been developed. Nevertheless, most of these tools identify SMBGCs based on sequence similarity with enzymes typically involved in specialised metabolism and thus may miss SMBGCs coding for undercharacterised enzymes. Here we present Synteruptor (https://bioi2.i2bc.paris-saclay.fr/synteruptor), a program that identifies genomic islands, known to be enriched in SMBGCs, in the genomes of closely related species. With this tool, we identified a SMBGC in the genome of Streptomyces ambofaciens ATCC23877, undetected by antiSMASH versions prior to antiSMASH 5, and experimentally demonstrated that it directs the biosynthesis of two metabolites, one of which was identified as sphydrofuran. Synteruptor is also a valuable resource for the delineation of individual SMBGCs within antiSMASH regions that may encompass multiple clusters, and for refining the boundaries of these SMBGCs.

As the sister group to all other animals, ctenophores (comb jellies) are important for understanding the emergence and diversification of numerous animal traits. Efforts to explore the evolutionary processes that promoted diversification within Ctenophora are hindered by undersampling genomic diversity within this clade. To address this gap, we present the sequence, assembly and initial annotation of the genome of Beroe ovata. Beroe possess unique morphology, behavior, ecology and development. Unlike their generalist carnivorous kin, beroid ctenophores feed exclusively on other ctenophores. Accordingly, our analyses revealed a loss of chitinase, an enzyme critical for the digestion of most non-ctenophore prey, but superfluous for ctenophorivores. Broadly, our genomic analysis revealed that extensive gene loss and changes in gene regulation have shaped the unique biology of B. ovata. Despite the gene losses in B. ovata, our phylogenetic analyses on photosensitive opsins and several early developmental regulatory genes show that these genes are conserved in B. ovata. This additional sampling contributes to a more complete reconstruction of the ctenophore ancestor and points to the need for extensive comparisons within this ancient and diverse clade of animals. To promote further exploration of these data, we present BovaDB (http://ryanlab.whitney.ufl.edu/bovadb/), a portal for the B. ovata genome.

Ribosomal DNA (rDNA) repeat units are organized into tandem clusters in eukaryotic cells. In mice, these clusters are located on at least eight chromosomes and show extensive variation in the number of repeats between mouse genomes. To analyze intra- and inter-genomic variation of mouse rDNA repeats, we selectively isolated 25 individual rDNA units using Transformation-Associated Recombination (TAR) cloning. Long-read sequencing and subsequent comparative sequence analysis revealed that each full-length unit comprises an intergenic spacer (IGS) and a ∼13.4 kb long transcribed region encoding the three rRNAs, but with substantial variability in rDNA unit size, ranging from ∼35 to ∼46 kb. Within the transcribed regions of rDNA units, we found 209 variants, 70 of which are in external transcribed spacers (ETSs); but the rDNA size differences are driven primarily by IGS size heterogeneity, due to indels containing repetitive elements and some functional signals such as enhancers. Further evolutionary analysis categorized rDNA units into distinct clusters with characteristic IGS lengths; numbers of enhancers; and presence/absence of two common SNPs in promoter regions, one of which is located within promoter (p)RNA and may influence pRNA folding stability. These characteristic features of IGSs also correlated significantly with 5'ETS variant patterns described previously and associated with differential expression of rDNA units. Our results suggest that variant rDNA units are differentially regulated and open a route to investigate the role of rDNA variation on nucleolar formation and possible associations with pathology.

Mass spectrometry is a powerful and widely used tool for generating proteomics, lipidomics and metabolomics profiles, which is pivotal for elucidating biological processes and identifying biomarkers. However, missing values in mass spectrometry-based omics data may pose a critical challenge for the comprehensive identification of biomarkers and elucidation of the biological processes underlying human complex disorders. To alleviate this issue, various imputation methods for mass spectrometry-based omics data have been developed. However, a comprehensive comparison of these imputation methods is still lacking, and researchers are frequently confronted with a multitude of options without a clear rationale for method selection. To address this pressing need, we developed omicsMIC (mass spectrometry-based omics with Missing values Imputation methods Comparison platform), an interactive platform that provides researchers with a versatile framework to evaluate the performance of 28 diverse imputation methods. omicsMIC offers a nuanced perspective, acknowledging the inherent heterogeneity in biological data and the unique attributes of each dataset. Our platform empowers researchers to make data-driven decisions in imputation method selection based on real-time visualizations of the outcomes associated with different imputation strategies. The comprehensive benchmarking and versatility of omicsMIC make it a valuable tool for the scientific community engaged in mass spectrometry-based omics research. omicsMIC is freely available at https://github.com/WQLin8/omicsMIC.

We present GTDrift, a comprehensive data resource that enables explorations of genomic and transcriptomic characteristics alongside proxies of the intensity of genetic drift in individual species. This resource encompasses data for 1506 eukaryotic species, including 1413 animals and 93 green plants, and is organized in three components. The first two components contain approximations of the effective population size, which serve as indicators of the extent of random genetic drift within each species. In the first component, we meticulously investigated public databases to assemble data on life history traits such as longevity, adult body length and body mass for a set of 979 species. The second component includes estimations of the ratio between the rate of non-synonymous substitutions and the rate of synonymous substitutions (dN/dS) in protein-coding sequences for 1324 species. This ratio provides an estimate of the efficiency of natural selection in purging deleterious substitutions. Additionally, we present polymorphism-derived N _e estimates for 66 species. The third component encompasses various genomic and transcriptomic characteristics. With this component, we aim to facilitate comparative transcriptomics analyses across species, by providing easy-to-use processed data for more than 16 000 RNA-seq samples across 491 species. These data include intron-centered alternative splicing frequencies, gene expression levels and sequencing depth statistics for each species, obtained with a homogeneous analysis protocol. To enable cross-species comparisons, we provide orthology predictions for conserved single-copy genes based on BUSCO gene sets. To illustrate the possible uses of this database, we identify the most frequently used introns for each gene and we assess how the sequencing depth available for each species affects our power to identify major and minor splice variants.

Transcription factor (TF) binding is a key component of genomic regulation. There are numerous high-throughput experimental methods to characterize TF-DNA binding specificities. Their application, however, is both laborious and expensive, which makes profiling all TFs challenging. For instance, the binding preferences of ∼25% human TFs remain unknown; they neither have been determined experimentally nor inferred computationally. We introduce a structure-based learning approach to predict the binding preferences of TFs and the automated modelling of TF regulatory complexes. We show the advantage of using our approach over the classical nearest-neighbor prediction in the limits of remote homology. Starting from a TF sequence or structure, we predict binding preferences in the form of motifs that are then used to scan a DNA sequence for occurrences. The best matches are either profiled with a binding score or collected for their subsequent modeling into a higher-order regulatory complex with DNA. Co-operativity is modelled by: (i) the co-localization of TFs and (ii) the structural modeling of protein-protein interactions between TFs and with co-factors. We have applied our approach to automatically model the interferon-β enhanceosome and the pioneering complexes of OCT4, SOX2 (or SOX11) and KLF4 with a nucleosome, which are compared with the experimentally known structures.

In eukaryotes, translation initiation is a highly regulated process, which combines cis-regulatory sequences located on the messenger RNA along with trans-acting factors like eukaryotic initiation factors (eIF). One critical step of translation initiation is the start codon recognition by the scanning 43S particle, which leads to ribosome assembly and protein synthesis. In this study, we investigated the involvement of secondary structures downstream the initiation codon in the so-called START (STructure-Assisted RNA translation) mechanism on AUG and non-AUG translation initiation. The results demonstrate that downstream secondary structures can efficiently promote non-AUG translation initiation if they are sufficiently stable to stall a scanning 43S particle and if they are located at an optimal distance from non-AUG codons to stabilize the codon-anticodon base pairing in the P site. The required stability of the downstream structure for efficient translation initiation varies in distinct cell types. We extended this study to genome-wide analysis of functionally characterized alternative translation initiation sites in Homo sapiens. This analysis revealed that about 25% of these sites have an optimally located downstream secondary structure of adequate stability which could elicit START, regardless of the start codon. We validated the impact of these structures on translation initiation for several selected uORFs.

The 'canonical' protein sets distributed by UniProt are widely used for similarity searching, and functional and structural annotation. For many investigators, canonical sequences are the only version of a protein examined. However, higher eukaryotes often encode multiple isoforms of a protein from a single gene. For unreviewed (UniProtKB/TrEMBL) protein sequences, the longest sequence in a Gene-Centric group is chosen as canonical. This choice can create inconsistencies, selecting >95% identical orthologs with dramatically different lengths, which is biologically unlikely. We describe the ortho2tree pipeline, which examines Reference Proteome canonical and isoform sequences from sets of orthologous proteins, builds multiple alignments, constructs gap-distance trees, and identifies low-cost clades of isoforms with similar lengths. After examining 140 000 proteins from eight mammals in UniProtKB release 2022_05, ortho2tree proposed 7804 canonical changes for release 2023_01, while confirming 53 434 canonicals. Gap distributions for isoforms selected by ortho2tree are similar to those in bacterial and yeast alignments, organisms unaffected by isoform selection, suggesting ortho2tree canonicals more accurately reflect genuine biological variation. 82% of ortho2tree proposed-changes agreed with MANE; for confirmed canonicals, 92% agreed with MANE. Ortho2tree can improve canonical assignment among orthologous sequences that are >60% identical, a group that includes vertebrates and higher plants.