Search

Large scale data resources such as the NCI’s Cancer Research Data Commons (CRDC) and the Genotype-Tissue Expression (GTEx) portal have the potential to simplify the analysis of cancer data by providing data that can be used as standards or controls. However, comparisons with data that is processed using different methodologies or even different versions of software, parameters and supporting datasets can lead to artefactual results. Reproducing the exact workflows from text-based standard operating procedures (SOPs) is problematic as the documentation can be incomplete or out of date, especially for complex workflows involving many executables and scripts. We extend our open-source Biodepot-workflow-builder (Bwb) platform to provide a dynamic solution that disseminates the computational protocols to process large-scale sequencing data developed by the National Cancer Institute (NCI) Genomic Data Commons (GDC). Specifically, we converted the GDC DNA sequencing (DNA-Seq) and the GDC mRNA sequencing (mRNA-Seq) SOPs into reproducible, self-installing, containerized, and interactive graphical workflows. Secure integration with protected-access CRDC data is achieved using the Data Commons Framework Services (DCFS) Gen3 protocol. These graphical workflows can be applied to reproducibly analyze datasets across other repositories and/or custom user data. Analyses can be performed on a local laptop, desktop, or cloud providers. With RNA-Seq datasets from the GDC and GTEx, we illustrate the importance of uniform analysis of control and treatment data for accurate inference of differentially expressed genes. Furthermore, we demonstrate that these best practices for analyzing RNA-seq data from different sources can be achieved using our accessible workflows. Most importantly, we demonstrate how our reproducible distribution of the methodology can transform the analyses of cancer genomic data by enabling researchers to leverage datasets across multiple repositories to enhance data interpretation.

SummaryGraphical processing units can greatly accelerate image processing but adoption has been hampered by the need for specialized hardware and software. The cloud offers inexpensive on-demand instances that can be pre-configured with the necessary software. Specifically, we use the Biodepot-workflow-builder (Bwb) to deploy a containerized version of Fiji that includes the CLIJ package to use GPUs on the cloud. In addition, we provide an Amazon Machine Image (AMI) with the correct drivers and Docker images pre-loaded. We demonstrate the portability and reproducibility of the platform by deploying an interactive Fiji/CLIJ workflow on both Amazon Web Services and IBM cloud. The workflows produce identical results while providing a 29-fold reduction in execution time.

Acute leukemia (AL), a cancer of the blood, has continued high mortality rates despite a plethora of available treatments. Variable prognosis, treatment response and survival are largely based on distinct cytogenetic and molecular aberrations that characterize AL subtypes. Fast and cost-effective methods of detecting AL fusions and mutations are needed to improve clinical outcomes for patients. We report preliminary data on an ultrarapid and portable nanopore long-read sequencing assay with a same-day turnaround time of ~6-7 hours for identifying key leukemic fusions. Leukemias frequently present with recurrent fusions that impact risk stratification and therapy choice, our assay can help oncologists overcome time related challenges to get patients on treatments faster.

Recurrent gene fusions are common drivers of disease pathophysiology in leukemias. Identification of these structural variants helps stratify disease by risk and assists with therapy choice. Current fusion detection methods require long turnaround time (7-10 days) or advance knowledge of the genes involved in the fusions. To address the need for rapid identification of clinically actionable fusion genes in heme malignancies without a-priori knowledge of the genes, we describe a long-read sequencing DNA assay designed with CRISPR guides to select and enrich for recurrent leukemia fusion genes. By applying rapid sequencing technology based on nanopores, we sequenced long pieces of genomic DNA and successfully detected fusion genes in cell lines and primary specimens (e.g., BCR-ABL1, PML-RARA, CBFB-MYH11, KMT2A-AF4) using cloud-based bioinformatics workflows with novel custom fusion finder software. We detected fusion genes in 100% of cell lines with the expected breakpoints and confirmed the presence or absence of a recurrent fusion gene in 12 of 14 patient cases. With our optimized assay and cloud-based bioinformatics workflow, these assays and analyses could be performed in under 8 hours.Key pointsWe describe a CRISPR-Cas9 enrichment Nanopore sequencing assay with streamlined bioinformatics that outperforms other fusion detectors.We successfully detected both fusion genes and specific breakpoints in CML, APL, and AML in under 8 hours in 80% of patients.Visual Abstract (Figure 1)We successfully detected fusion genes in hematological malignancies with a fast and efficient long-read sequencing workflow in under 8 hours. The method makes the genomic characterization of BCR-ABL1 DNA breakpoint in patients quick and simple, which potentiates design of patient specific primers for personalized monitoring MRD assays.Our assay is based on a CRISPR-Cas9 non-amplification enrichment library preparation strategy and uses Nanopore sequencing single stranded genomic DNA coupled with streamlined bioinformatic workflow containing a novel fusion detector software which outperforms current fusion detection software.

Objective. To measure the EEG signals of the people with chronic stroke in eyes-closed and eyes-open condition and study their relationship with the cognitive function and mental wellbeing. Methods. The investigators would conduct cognitive and mental wellbeing tests on recruited subjects. Their EEG signal was acquired by the 16-channel EEG system. The absolute power under different frequency bands and EEG indices (delta alpha ratio and pairwise derived brain symmetry index) in different eye conditions was calculated. Pearson's correlation was conducted to investigate the association between the clinical tests and the EEG index. Results. 32 subjects were recruited for the study. There was a significant correlation between the pairwise derived brain symmetry index (pdBSI) in eyes-open condition with the Stroop Test (p = .002), Paced Auditory Serial Addition Test-3 s (p = .008)/2 s (p = .002) and WHO-5 well-being scale (p = .023). Conclusions. There is a significant correlation between the brain symmetry index and the cognitive and wellbeing assessment. Brain symmetry index over the delta frequency has been found to be the most useful parameter relating to the clinical score. Significance: It is recommended to use EEG as an adjunctive neuropsychological assessment in clinics for people with chronic stroke, especially for clients who could not undertake conventional assessments (eg aphasia, attention problem). Highlights: There is a significant correlation between the EEG index and the clinical neuropsychological assessment Pairwise Derived Brain Symmetry index in delta frequency range correlated with most of the neuropsychological outcome. It is feasible for us to adopt EEG as an adjunctive assessment in clinical settings.

Modern biomedical image analyses workflows contain multiple computational processing tasks giving rise to problems in reproducibility. In addition, image datasets can span both spatial and temporal dimensions, with additional channels for fluorescence and other data, resulting in datasets that are too large to be processed locally on a laptop. For omics analyses, software containers have been shown to enhance reproducibility, facilitate installation and provide access to scalable computational resources on the cloud. However, most image analyses contain steps that are graphical and interactive, features that are not supported by most omics execution engines. We present the containerized and cloud-enabled Biodepot-workflow-builder platform that supports graphics from software containers and has been extended for image analyses. We demonstrate the potential of our modular approach with multi-step workflows that incorporate the popular and open-source Fiji suite for image processing. One of our examples integrates fully interactive Fiji macros with Jupyter notebooks. Our second example illustrates how the complicated cloud setup of an computationally intensive process such as stitching 3D digital pathology datasets using BigStitcher can be automated and simplified. In both examples, users can leverage a form-based graphical interface to execute multi-step workflows with a single click, using the provided sample data and preset input parameters. Alternatively, users can interactively modify the image processing steps in the workflow, apply the workflows to their own data, change the input parameters and macros. By providing interactive graphics support to software containers, our modular platform supports reproducible image analysis workflows, simplified access to cloud resources for analysis of large datasets, and integrated workflows across imaging, genomics and transcriptomics data.

After radiologists perform a set of chest-x-rays (CXRs) they write a short report describing their observations and interpretations. Because these reports are free-text documents, there is the risk of miscommunication, which can result in reduced patient outcomes. We applied text mining methods to radiology reports in the MIMIC Chest X-ray (MIMIC-CXR) database [5], consisting of 227,835 de-identified free-text radiology reports. We selected relevant terms (features) and developed predictive models that take a radiology report as input and return the probability the report describes a positive diagnosis for pneumonia, a common respiratory condition characterized by the accumulation of fluid in the lungs. Subsequently, we evaluated the performance of different predictive models using the area under the curve (AUC) and the Brier Score. Due to the large number of reports in the MIMIC-CXR database, we generated and evaluated predictive models by randomly selecting 500, 1000, 2000, and 3000 reports. Specifically, we randomly selected reports and assigned 70% to the training set and 30% to the test set, created a term-document matrix giving the frequencies of different sets of 1 or 2 consecutive words (1-gram or 2-gram) using the R package tm [2], performed feature selection to identify terms that differentiate between cases, and trained the models using different classification methods, including the k nearest neighbor (KNN), random forest [4], gradient boosting machine [3], xgboost [1], and adaboost [6]. We repeated the process six times and computed the average assessment statistics. Our results indicate that all the models perform similarly on the test set except for KNN. KNN had the worst performance with an average Brier Score (ABS) of 0.313 and average AUC of 0.645. The other algorithms had high performance: random forest (ABS =0.174, AUC=0.836), gradient boosting (ABS=0.175, AUC=0.820), xgboost (ABS=0.177, AUC=0.814), and adaboost (ABS=0.163, 0.815). The high performance suggests machine learning models have the potential to impact patient care in radiology.

Gene regulatory network reconstruction is an essential task of genomics in order to further our understanding of how genes interact dynamically with each other. The most readily available data, however, are from steady-state observations. These data are not as informative about the relational dynamics between genes as knockout or over-expression experiments, which attempt to control the expression of individual genes. We develop a new framework for network inference using samples from the equilibrium distribution of a vector autoregressive (VAR) time-series model which can be applied to steady-state gene expression data. We explore the theoretical aspects of our method and apply the method to synthetic gene expression data generated using GeneNetWeaver.

The response to respiratory viruses varies substantially between individuals, and there are currently no known molecular predictors from the early stages of infection. Here we conduct a community-based analysis to determine whether pre- or early post-exposure molecular factors could predict physiologic responses to viral exposure. Using peripheral blood gene expression profiles collected from healthy subjects prior to exposure to one of four respiratory viruses (H1N1, H3N2, Rhinovirus, and RSV), as well as up to 24 h following exposure, we find that it is possible to construct models predictive of symptomatic response using profiles even prior to viral exposure. Analysis of predictive gene features reveal little overlap among models; however, in aggregate, these genes are enriched for common pathways. Heme metabolism, the most significantly enriched pathway, is associated with a higher risk of developing symptoms following viral exposure. This study demonstrates that pre-exposure molecular predictors can be identified and improves our understanding of the mechanisms of response to respiratory viruses., The response to respiratory virus exposure can currently not be predicted by pre- or early post-exposure molecular signatures. Here, the authors conduct a community-based analysis of blood gene expression from healthy individuals exposed to respiratory viruses and provide predictive models and biological insight into the physiological response.

A large amount of panomic data has been generated in populations for understanding causal relationships in complex biological systems. Both genetic and temporal models can be used to establish causal relationships among molecular, cellular, or phenotypical traits, but with limitations. To fully utilize high-dimension temporal and genetic data, we develop a multivariate polynomial temporal genetic association (MPTGA) approach for detecting temporal genetic loci (teQTLs) of quantitative traits monitored over time in a population and a temporal genetic causality test (TGCT) for inferring causal relationships between traits linked to the locus. We apply MPTGA and TGCT to simulated data sets and a yeast F2 population in response to rapamycin, and demonstrate increased power to detect teQTLs. We identify a teQTL hotspot locus interacting with rapamycin treatment, infer putative causal regulators of the teQTL hotspot, and experimentally validate RRD1 as the causal regulator for this teQTL hotspot., Temporal omics data have the potential to dissect complex biological networks. Here the authors develop methods for detecting temporal genetic loci (teQTLs) of quantitative traits monitored over time and inferring causal relationships between traits linked to the locus.

Public Infrastructure-as-a-Service (IaaS) clouds abstract various details regarding the implementation of resources provided to users. For example, users are not informed about the exact physical location of their virtual machines (VMs), the specific hardware used, the number of co-resident VMs they reside with, or the workloads that co-resident VMs are running. Detecting when VMs underperform can help identify resource contention from co-resident VMs to spur their replacement. Resource utilization metrics can be used to help classify performance of runs for use in VM performance model datasets to sample the distribution of performance outcomes in the cloud. VM performance models are key to predicting the cost of bioinformatics analyses in the public cloud. This paper investigates the performance variations of running a RNA sequencing workflow in the public cloud. We examine causes of performance variations including VM provisioning, CPU heterogeneity, and resource contention. We leverage Amazon Elastic Compute Cloud (EC2) placement groups, a feature designed to help influence VM placement to help examine how VM placement impacts performance variations. As a use case, we investigate the performance of a multi-stage bioinformatics RNA sequencing (RNA-seq) analytical workflow consisting of four distinct phases, executing in 90 minutes on average using 8-core public cloud VMs. In addition, we investigate whether Linux resource utilization metrics collected by profiling workflow runs can help identify performance implications.

Modern day analyses of biomedical data typically involves a series of computational tasks called workflows. Each of these modules could be written by different laboratories and thus, potentially requires different computing environments. Reproducible analyses of biomedical data is a growing field of interest [1]. This makes it important to develop tools to facilitate analyses in a reproducible manner. Software containers have been used to increase reproducibility in bioinformatics analyses [2]. However, these software containers often use command line tools, requiring expertise that make them inaccessible to many biomedical scientists.

Cloud computing offers on-demand, scalable computing and storage, and has become an essential resource for the analyses of big biomedical data. The usual approach to cloud computing requires users to reserve and provision virtual servers. An emerging alternative is to have the provider allocate machine resources dynamically. This type of serverless computing has tremendous potential for biomedical research in terms of ease-of-use, instantaneous scalability, and cost effectiveness. In our proof of concept example, we demonstrate how serverless computing provides low cost access to hundreds of CPUs, on demand, with little or no setup. In particular, we illustrate that the all-against-all pairwise comparison among all unique human proteins can be accomplished in approximately 2 minutes, at a cost of less than $1, using Amazon Web Services Lambda. We also demonstrate the feasibility of our approach using Google Functions and show that the same task of pairwise protein sequence comparison can be accomplished in approximately 11.5 minutes. In contrast, running the same task on a typical laptop computer required 8.7 hours.

We present a novel method that yields 1100-fold computational speedup and allows biomedical scientists to interactively adjust alignment parameters in real time to iteratively improve final analytical results. Specifically, the alignment time for a 640 million read human RNA-sequencing dataset is reduced from 19 hours to 1 minute using serverless cloud computing. We provide a graphical interface for the accelerated workflows, thus making our methodology accessible to non-cloud experts.

The inference of gene networks from large-scale human genomic data is challenging due to the difficulty in identifying correct regulators for each gene in a high-dimensional search space. We present a Bayesian approach integrating external data sources with knockdown data from human cell lines to infer gene regulatory networks. In particular, we assemble multiple data sources including gene expression data, genome-wide binding data, gene ontology, known pathways and use a supervised learning framework to compute prior probabilities of regulatory relationships. We show that our integrated method improves the accuracy of inferred gene networks. We apply our method to two different human cell lines, which illustrates the general scope of our method.

We present the BioDepot-workflow-builder (Bwb), a software tool that allows users to create and execute reproducible bioinformatics workflows using a drag-and-drop interface. Graphical widgets represent Docker containers executing a modular task. Widgets are linked graphically to build bioinformatics workflows that can be reproducibly deployed across different local and cloud platforms. Each widget contains a form-based user interface to facilitate parameter entry, and a console to display intermediate results. Bwb provides tools for rapid customization of widgets, containers and workflows. Saved workflows can be shared using Bwb’s native format or exported as shell scripts.

Background Using software containers has become standard practice to reproducibly deploy and execute biomedical workflows on the cloud. However, some applications that contain time-consuming initialization steps will produce unnecessary costs for repeated executions. Findings We demonstrate that hot-starting from containers that have been frozen after the application has already begun execution can speed up bioinformatics workflows by avoiding repetitive initialization steps. We use an open-source tool called Checkpoint and Restore in Userspace (CRIU) to save the state of the containers as a collection of checkpoint files on disk after it has read in the indices. The resulting checkpoint files are migrated to the host, and CRIU is used to regenerate the containers in that ready-to-run hot-start state. As a proof-of-concept example, we create a hot-start container for the spliced transcripts alignment to a reference (STAR) aligner and deploy this container to align RNA sequencing data. We compare the performance of the alignment step with and without checkpoints on cloud platforms using local and network disks. Conclusions We demonstrate that hot-starting Docker containers from snapshots taken after repetitive initialization steps are completed significantly speeds up the execution of the STAR aligner on all experimental platforms, including Amazon Web Services, Microsoft Azure, and local virtual machines. Our method can be potentially employed in other bioinformatics applications in which a checkpoint can be inserted after a repetitive initialization phase.