Search

Data from sequencing instruments are produced at such rates that their analysis is becoming increasingly computationally challenging. Although DNA sequence clustering of very large datasets is an important computational step in various bioinformatics applications, it is a performanceintensive task that often cannot be completed without compromising sensitivity for speed. In order to optimize CPU and RAM usage in DNA clustering, we propose a probabilistic, rigorous, and efficient technique to partition a large DNA sequence dataset into smaller, non-overlapping subsets, which can then be analyzed using more precise clustering algorithms. The approach results in a significant reduction in CPU and RAM requirements, as well as a more intuitive parallelization of the clustering step. We show in our results that our algorithm, implemented in a program called SLYMFAST, can cluster in just a few hours datasets that would otherwise take weeks to cluster without partitioning first.

Infectious pathogens can disrupt the microbiome in addition to directly affecting the host. Impacts of disease may be dependent on the ability of the microbiome to recover from such disturbance, yet remarkably little is known about microbiome recovery after disease, particularly in nonhuman animals. We assessed the resilience of the amphibian skin microbial community after disturbance by the pathogen, Batrachochytrium dendrobatidis (Bd). Skin microbial communities of laboratory-reared mountain yellow-legged frogs were tracked through three experimental phases: prior to Bd infection, after Bd infection (disturbance), and after clearing Bd infection (recovery period). Bd infection disturbed microbiome composition and altered the relative abundances of several dominant bacterial taxa. After Bd infection, frogs were treated with an antifungal drug that cleared Bd infection, but this did not lead to recovery of microbiome composition (measured as Unifrac distance) or relative abundances of dominant bacterial groups. These results indicate that Bd infection can lead to an alternate stable state in the microbiome of sensitive amphibians, or that microbiome recovery is extremely slow—in either case resilience is low. Furthermore, antifungal treatment and clearance of Bd infection had the additional effect of reducing microbial community variability, which we hypothesize results from similarity across frogs in the taxa that colonize community vacancies resulting from the removal of Bd. Our results indicate that the skin microbiota of mountain yellow-legged frogs has low resilience following Bd-induced disturbance and is further altered by the process of clearing Bd infection, which may have implications for the conservation of this endangered amphibian.

Sequence clustering is an important computational step in numerous bioinformatics applications such as high-throughput immune system characterization, marker-based diversity profiling, and reduced-complexity population genetic experiments. However, clustering very large datasets produced by high-throughput sequencers is CPU- and memory-intensive. As such, the analysis of large datasets is challenging and often requires compromising clustering sensitivity for speed. To address this limitation, we propose a new probabilistic data partitioning technique that divides large datasets of sequences of equal length into smaller, non-overlapping subsets that can subsequently be clustered using higher sensitivity algorithms at lower computational costs. In addition to substantially lowering resources requirements, this partitioning step provides an opportunity for effortless parallelization of DNA sequence clustering, and for the independent processing of each subset using more accurate algorithms. Our results show that our algorithm, which we implemented in a program called SLYMFAST can be used to cluster in a few hours datasets that would otherwise take weeks to cluster without partitioning first.

Computational scalability has become an important requirement for processing the massive amounts of data generated in contemporary sequencing-based experiments. The availability of large computational resources through academic, regional or national cyber-infrastructure efforts, as well as through inexpensive cloud offerings, has shifted the bottleneck, which now lies in the extensive expertise necessary to create reproducible and scalable bioinformatics pipelines and deploy them to such diverse infrastructures. We present here MetaFlow|mics, a comprehensive pipeline for the analysis of microbiome marker data using best practices and state-of-the-art cyberinfrastructure standards to ensure reproducibility. MetaFlow|mics provides seamless scalability and extensibility, allowing users to build and test their pipelines on a laptop with small datasets and to subsequently run them on large datasets on an HPC or on the Cloud with a change to a single line of code. Our framework is built on top of the Nextflow workflow management system and provides an interoperable architecture that leverages self-contained Docker and Singularity instances with all the dependencies and requirements needed to quickly deploy and use the pipeline.

Motivation Metagenomic approaches hold the potential to characterize microbial communities and unravel the intricate link between the microbiome and biological processes. Assembly is one of the most critical steps in metagenomics experiments. It consists of transforming overlapping DNA sequencing reads into sufficiently accurate representations of the community’s genomes. This process is computationally difficult and commonly results in genomes fragmented across many contigs. Computational binning methods are used to mitigate fragmentation by partitioning contigs based on their sequence composition, abundance or chromosome organization into bins representing the community’s genomes. Existing binning methods have been principally tuned for bacterial genomes and do not perform favorably on viral metagenomes. Results We propose Composition and Coverage Network (CoCoNet), a new binning method for viral metagenomes that leverages the flexibility and the effectiveness of deep learning to model the co-occurrence of contigs belonging to the same viral genome and provide a rigorous framework for binning viral contigs. Our results show that CoCoNet substantially outperforms existing binning methods on viral datasets. Availability and implementation CoCoNet was implemented in Python and is available for download on PyPi (https://pypi.org/). The source code is hosted on GitHub at https://github.com/Puumanamana/CoCoNet and the documentation is available at https://coconet.readthedocs.io/en/latest/index.html. CoCoNet does not require extensive resources to run. For example, binning 100k contigs took about 4 h on 10 Intel CPU Cores (2.4 GHz), with a memory peak at 27 GB (see Supplementary Fig. S9). To process a large dataset, CoCoNet may need to be run on a high RAM capacity server. Such servers are typically available in high-performance or cloud computing settings. Supplementary information Supplementary data are available at Bioinformatics online.

Background Single-cell RNA sequencing (scRNA-Seq) is an increasingly popular platform to study heterogeneity at the single-cell level. Computational methods to process scRNA-Seq data are not very accessible to bench scientists as they require a significant amount of bioinformatic skills. Results We have developed Granatum, a web-based scRNA-Seq analysis pipeline to make analysis more broadly accessible to researchers. Without a single line of programming code, users can click through the pipeline, setting parameters and visualizing results via the interactive graphical interface. Granatum conveniently walks users through various steps of scRNA-Seq analysis. It has a comprehensive list of modules, including plate merging and batch-effect removal, outlier-sample removal, gene-expression normalization, imputation, gene filtering, cell clustering, differential gene expression analysis, pathway/ontology enrichment analysis, protein network interaction visualization, and pseudo-time cell series construction. Conclusions Granatum enables broad adoption of scRNA-Seq technology by empowering bench scientists with an easy-to-use graphical interface for scRNA-Seq data analysis. The package is freely available for research use at http://garmiregroup.org/granatum/app Electronic supplementary material The online version of this article (doi:10.1186/s13073-017-0492-3) contains supplementary material, which is available to authorized users.

We present GranatumX, a next-generation software environment for single-cell data analysis. GranatumX is inspired by the interactive web tool Granatum. It enables biologists to access the latest single-cell bioinformatics methods in a web-based graphical environment. It also offers software developers the opportunity to rapidly promote their own tools with others in customizable pipelines. The architecture of GranatumX allows for easy inclusion of plugin modules, named Gboxes, that wrap around bioinformatics tools written in various programming languages and on various platforms. GranatumX can be run on the cloud or private servers and generate reproducible results. It is a community-engaging, flexible, and evolving software ecosystem for scRNA-Seq analysis, connecting developers with bench scientists. GranatumX is freely accessible at http://garmiregroup.org/granatumx/app.

BackgroundSingle-cell RNA sequencing (scRNA-seq) offers new opportunities to study gene expression of tens of thousands of single cells simultaneously. However, a significant problem of current scRNA-seq data is the large fractions of missing values or “dropouts” in gene counts. Incorrect handling of dropouts may affect downstream bioinformatics analysis. As the number of scRNA-seq datasets grows drastically, it is crucial to have accurate and efficient imputation methods to handle these dropouts.MethodsWe present DeepImpute, a deep neural network based imputation algorithm. The architecture of DeepImpute efficiently uses dropout layers and loss functions to learn patterns in the data, allowing for accurate imputation.ResultsOverall DeepImpute yields better accuracy than other publicly available scRNA-Seq imputation methods on experimental data, as measured by mean squared error or Pearson’s correlation coefficient. Moreover, its efficient implementation provides significantly higher performance over the other methods as dataset size increases. Additionally, as a machine learning method, DeepImpute allows to use a subset of data to train the model and save even more computing time, without much sacrifice on the prediction accuracy.ConclusionsDeepImpute is an accurate, fast and scalable imputation tool that is suited to handle the ever increasing volume of scRNA-seq data. The package is freely available at https://github.com/lanagarmire/DeepImpute

Previous research has demonstrated that individual modal components of signals propagating in an acoustic waveguide can be isolated from a single hydrophone using non-linear time sampling (Bonnel et al., 2010) and that mode shape functions can be directly extracted from vertical array data in a laboratory setting (Bonnel et al., 2011). Here, we apply and extend the technique to experimental data collected in 2010 from a 15-element vertical array, covering only 63% of the 53 m water column off the continental slope of the Beaufort Sea, during the fall bowhead whale migration. We demonstrate how up to five distinct mode shapes can be extracted from frequency-modulated bowhead whale calls, along with the vertical array tilt and tilt direction, without resorting to assumptions about mode orthogonality across the array aperture. The extracted mode shapes can then be used to estimate whale call depth. Filtered modes (and mode cutoff information) can also be used to invert for whale call ranges and ocean bottom properties. [Work supported by the North Pacific Research Board and permitted by Shell Exploration and Production Company.]