Bioinformatics advances最新文献

Motivation: motifbreakR scans genetic variants against position weight matrices of transcription factors (TFs) to determine the potential for the disruption of binding at the site of the variant. It leverages the Bioconductor suite of software packages and annotations to query a diverse array of genomes and motif databases. Initially developed to interrogate the effect of single-nucleotide variants on TF binding sites, in motifbreakR v2, we have updated the functionality.

Results: New features include the ability to query other types of complex genetic variants, such as short insertions and deletions. This capability allows modeling a more extensive array of variants that may have significant effects on TF binding. Additionally, predictions based on sequence preference alone can indicate many more potential binding events than observed. Adding information from DNA-binding sequencing datasets lends confidence to motif disruption prediction by demonstrating TF binding in cell lines and tissue types. Therefore, motifbreakR can directly query the ReMap2022 database for evidence that a TF matching the disrupted motif binds over the disrupting variant. Finally, in motifbreakR, in addition to the existing interface, we implemented an R/Shiny graphical user interface to simplify and enhance access to researchers with different skill sets.

Availability and implementation: motifbreakR is implemented in R. Source code, documentation, and tutorials are available on Bioconductor at https://bioconductor.org/packages/release/bioc/html/motifbreakR.html and GitHub at https://github.com/Simon-Coetzee/motifBreakR.

Summary: The annotation of deeply sequenced, de novo assembled transcriptomes continues to be a challenge as some of the state-of-the-art tools are slow, difficult to install, and hard to use. We have tackled these issues with TransAnnot, a fast, automated transcriptome annotation pipeline that is easy to install and use. Leveraging the fast sequence searches provided by the MMseqs2 suite, TransAnnot offers one-step annotation of homologs from Swiss-Prot, gene ontology terms and orthogroups from eggNOG, and functional domains from Pfam. Users also have the option to annotate against custom databases. TransAnnot accepts sequencing reads (short and long), nucleotide sequences, or amino acid sequences as input for annotation. When benchmarked with test data sets of amino acid sequences, TransAnnot was 333, 284, and 18 times faster than comparable tools such as EnTAP, Trinotate, and eggNOG-mapper respectively.

Availability and implementation: TransAnnot is free to use, open sourced under GPLv3, and is implemented in C++ and Bash. Source code, documentation, and pre-compiled binaries are available at https://github.com/soedinglab/transannot. TransAnnot is also available via bioconda (https://anaconda.org/bioconda/transannot).

Motivation: Hydrophobic patches on protein surfaces play important functional roles in protein-protein and protein-ligand interactions. Large hydrophobic surfaces are also involved in the progression of aggregation diseases. Predicting exposed hydrophobic patches from a protein sequence has shown to be a difficult task. Fine-tuning foundation models allows for adapting a model to the specific nuances of a new task using a much smaller dataset. Additionally, multitask deep learning offers a promising solution for addressing data gaps, simultaneously outperforming single-task methods.

Results: In this study, we harnessed a recently released leading large language model Evolutionary Scale Models (ESM-2). Efficient fine-tuning of ESM-2 was achieved by leveraging a recently developed parameter-efficient fine-tuning method. This approach enabled comprehensive training of model layers without excessive parameters and without the need to include a computationally expensive multiple sequence analysis. We explored several related tasks, at local (residue) and global (protein) levels, to improve the representation of the model. As a result, our model, PatchProt, cannot only predict hydrophobic patch areas but also outperforms existing methods at predicting primary tasks, including secondary structure and surface accessibility predictions. Importantly, our analysis shows that including related local tasks can improve predictions on more difficult global tasks. This research sets a new standard for sequence-based protein property prediction and highlights the remarkable potential of fine-tuning foundation models enriching the model representation by training over related tasks.

Availability and implementation: https://github.com/Deagogishvili/chapter-multi-task.

Motivation: Discovering new protein-protein interactions (PPIs) across entire proteomes offers vast potential for understanding novel protein functions and elucidate system properties within or between an organism. While recent advances in computational structural biology, particularly AlphaFold-Multimer, have facilitated this task, scaling for large-scale screenings remains a challenge, requiring significant computational resources.

Results: We evaluated the impact of reducing the number of models generated by AlphaFold-Multimer from five to one on the method's ability to distinguish true PPIs from false ones. Our evaluation was conducted on a dataset containing both intra- and inter-species PPIs, which included proteins from bacterial and eukaryotic sources. We demonstrate that reducing the sampling does not compromise the accuracy of the method, offering a faster, efficient, and environmentally friendly solution for PPI predictions.

Availability and implementation: The code used in this article is available at https://github.com/MIDIfactory/AlphaFastPPi. Note that the same can be achieved using the latest version of AlphaPulldown available at https://github.com/KosinskiLab/AlphaPulldown.

Summary: Target Enrichment Sequencing or Capture-based metagenomics has emerged as an approach of interest for viral metagenomics in complex samples. However, these datasets are usually analyzed with standard downstream Bioinformatics analyses. CAPTVRED (Capture-based metagenomics Analysis Pipeline for tracking ViRal species from Environmental Datasets), has been designed to assess the virome present in complex samples, specially focused on those obtained by Target Enrichment Sequencing approach. This work aims to provide a user-friendly tool that complements this sequencing approach for the total or partial virome description, especially from environmental matrices. It includes a setup module which allows preparation and adjustment of the pipeline to any capture panel directed to a set of species of interest. The tool also aims to reduce time and computational cost, as well as to provide comprehensive, reproducible, and accessible results while being easy to costume, set up, and install.

Availability and implementation: Source code and test datasets are freely available at github repository: https://github.com/CompGenLabUB/CAPTVRED.git.

Summary: Signaling dynamics encode important features and regulatory mechanisms of biological systems, and recent studies have reported the use of simulated signaling dynamics with mechanistic modeling as biomarkers for human diseases. Given the success of deep learning techniques, it is expected that they can extract informative patterns from simulation results more effectively than traditional approaches involving manual feature selection, which can be used for subsequent analyses, such as patient stratification and survival prediction. Here, we propose DynProfiler, which utilizes the entire signaling dynamics, including intermediate variables, as input and leverages deep learning techniques to extract informative features without requiring any labels. Furthermore, DynProfiler incorporates a modern explainable AI solution to provide quantitative time-dependent importance scores for each dynamics. Using simulated dynamics of patients with breast cancer as an example, we demonstrate DynProfiler's ability to extract high-quality features that can predict mortality risk and identify important dynamics, highlighting upregulated phosphorylated GSK3β as a biomarker for poor prognosis. Overall, this tool can be useful for clinical application, as well as for elucidating biological system dynamics.

Availability and implementation: The DynProfiler Python library is available in GitHub at https://github.com/okadalabipr/DynProfiler.

[This corrects the article DOI: 10.1093/bioadv/vbae125.].

Summary: We present tidysbml, an R package able to perform compartments, species, and reactions data extraction from Systems Biology Markup Language (SBML) documents (up to Level 3) in tabular data structures (i.e. R dataframes) to easily access and handle the richness of the biological information. Thanks to its output format, the package facilitates data manipulation, enabling manageable construction, and therefore analysis, of custom networks, as well as data retrieval, by means of R packages such as igraph, RCy3, and biomaRt. Exemplar data (i.e. SBML files) are extracted from Reactome.

Availability and implementation: The tidysbml R package is distributed under CC BY 4.0 License and can be found publicly available in Bioconductor (https://bioconductor.org/packages/tidysbml) and on GitHub (https://github.com/veronicapaparozzi/tidysbml).

Summary: To address the challenges of the storage, sharing, and analysis of multi-omics data, here we introduce the newest version of Panomicon, which includes the improvement of the underlying data model, the introduction of new registration and control access service, together with the seamless integration with other services (like TargetMine for data enrichment analysis), integrated in a completely new, more user friendly web application.

Availability and implementation: Panomicon is available online at https://panomicon.nibiohn.go.jp. Unregistered users can access the publicly available data uploaded to Panomicon using the following account: user: guest, password: anonymous. Source code for the application is also freely available under a GNU license at https://github.com/Toxygates/Panomicon/. A brief user guide for the new features of Panomicon is provided as supplementary material online.

Motivation: Visualization and analysis of biological networks play crucial roles in understanding living systems. Biological networks include diverse types, from gene regulatory networks and protein-protein interactions to metabolic networks. Metabolic networks include substrates, products, and enzymes, which are regulated by allosteric mechanisms and gene expression. However, the analysis of these diverse omics types is challenging due to the diversity of databases and the complexity of network analysis.

Results: We developed iTraNet, a web application that visualizes and analyses trans-omics networks involving four types of networks: gene regulatory networks, protein-protein interactions, metabolic networks, and metabolite exchange networks. Using iTraNet, we found that in wild-type mice, hub molecules within the network tended to respond to glucose administration, whereas in ob/ob mice, this tendency disappeared. With its ability to facilitate network analysis, we anticipate that iTraNet will help researchers gain insights into living systems.

Availability and implementation: iTraNet is available at https://itranet.streamlit.app/.