Improving data workflow systems with cloud services and use of open data for bioinformatics research

The NIH Comparative Genomics Resource: addressing the promises and challenges of comparative genomics on human health

Comparative genomics is the comparison of genetic information within and across organisms to understand the evolution, structure, and function of genes, proteins, and non-coding regions (Sivashankari and Shanmughavel, Bioinformation 1:376-8, 2007). Advances in sequencing technology and assembly algorithms have resulted in the ability to sequence large genomes and provided a wealth of data that are being used in comparative genomic analyses. Comparative analysis can be leveraged to systematically explore and evaluate the biological relationships and evolution between species, aid in understanding the structure and function of genes, and gain a better understanding of disease and potential drug targets. As our knowledge of genetics expands, comparative genomics can help identify emerging model organisms among a broader span of the tree of life, positively impacting human health. This impact includes, but is not limited to, zoonotic disease research, therapeutics development, microbiome research, xenotransplantation, oncology, and toxicology. Despite advancements in comparative genomics, new challenges have arisen around the quantity, quality assurance, annotation, and interoperability of genomic data and metadata. New tools and approaches are required to meet these challenges and fulfill the needs of researchers. This paper focuses on how the National Institutes of Health (NIH) Comparative Genomics Resource (CGR) can address both the opportunities for comparative genomics to further impact human health and confront an increasingly complex set of challenges facing researchers.

An Epigenetic Alphabet of Crop Adaptation to Climate Change

Crop adaptation to climate change is in a part attributed to epigenetic mechanisms which are related to response to abiotic and biotic stresses. Although recent studies increased our knowledge on the nature of these mechanisms, epigenetics remains under-investigated and still poorly understood in many, especially non-model, plants, Epigenetic modifications are traditionally divided into two main groups, DNA methylation and histone modifications that lead to chromatin remodeling and the regulation of genome functioning. In this review, we outline the most recent and interesting findings on crop epigenetic responses to the environmental cues that are most relevant to climate change. In addition, we discuss a speculative point of view, in which we try to decipher the “epigenetic alphabet” that underlies crop adaptation mechanisms to climate change. The understanding of these mechanisms will pave the way to new strategies to design and implement the next generation of cultivars with a broad range of tolerance/resistance to stresses as well as balanced agronomic traits, with a limited loss of (epi)genetic variability.

Convolutional Embedded Networks for Population Scale Clustering and Bio-Ancestry Inferencing

The study of genetic variants (GVs) can help find correlating population groups and to identify cohorts that are predisposed to common diseases and explain differences in disease susceptibility and how patients react to drugs. Machine learning techniques are increasingly being applied to identify interacting GVs to understand their complex phenotypic traits. Since the performance of a learning algorithm not only depends on the size and nature of the data but also on the quality of underlying representation, deep neural networks (DNNs) can learn non-linear mappings that allow transforming GVs data into more clustering and classification friendly representations than manual feature selection. In this paper, we propose convolutional embedded networks (CEN) in which we combine two DNN architectures called convolutional embedded clustering (CEC) and convolutional autoencoder (CAE) classifier for clustering individuals and predicting geographic ethnicity based on GVs, respectively. We employed CAE-based representation learning to 95 million GVs from the '1000 genomes' (covering 2,504 individuals from 26 ethnic origins) and 'Simons genome diversity' (covering 279 individuals from 130 ethnic origins) projects. Quantitative and qualitative analyses with a focus on accuracy and scalability show that our approach outperforms state-of-the-art approaches such as VariantSpark and ADMIXTURE. In particular, CEC can cluster targeted population groups in 22 hours with an adjusted rand index (ARI) of 0.915, the normalized mutual information (NMI) of 0.92, and the clustering accuracy (ACC) of 89 percent. Contrarily, the CAE classifier can predict the geographic ethnicity of unknown samples with an F1 and Mathews correlation coefficient (MCC) score of 0.9004 and 0.8245, respectively. Further, to provide interpretations of the predictions, we identify significant biomarkers using gradient boosted trees (GBT) and SHapley Additive exPlanations (SHAP). Overall, our approach is transparent and faster than the baseline methods, and scalable for 5 to 100 percent of the full human genome.

Implementing the FAIR Data Principles in precision oncology: review of supporting initiatives

Compelling research has recently shown that cancer is so heterogeneous that single research centres cannot produce enough data to fit prognostic and predictive models of sufficient accuracy. Data sharing in precision oncology is therefore of utmost importance. The Findable, Accessible, Interoperable and Reusable (FAIR) Data Principles have been developed to define good practices in data sharing. Motivated by the ambition of applying the FAIR Data Principles to our own clinical precision oncology implementations and research, we have performed a systematic literature review of potentially relevant initiatives. For clinical data, we suggest using the Genomic Data Commons model as a reference as it provides a field-tested and well-documented solution. Regarding classification of diagnosis, morphology and topography and drugs, we chose to follow the World Health Organization standards, i.e. ICD10, ICD-O-3 and Anatomical Therapeutic Chemical classifications, respectively. For the bioinformatics pipeline, the Genome Analysis ToolKit Best Practices using Docker containers offer a coherent solution and have therefore been selected. Regarding the naming of variants, we follow the Human Genome Variation Society's standard. For the IT infrastructure, we have built a centralized solution to participate in data sharing through federated solutions such as the Beacon Networks.

Society for Immunotherapy of Cancer clinical and biomarkers data sharing resource document: Volume II-practical challenges

The development of strongly predictive validated biomarkers is essential for the field of immuno-oncology (IO) to advance. The highly complex, multifactorial data sets required to develop these biomarkers necessitate effective, responsible data-sharing efforts in order to maximize the scientific knowledge and utility gained from their collection. While the sharing of clinical- and safety-related trial data has already been streamlined to a large extent, the sharing of biomarker-aimed clinical trial derived data and data sets has been met with a number of hurdles that have impaired the progression of biomarkers from hypothesis to clinical use. These hurdles include technical challenges associated with the infrastructure, technology, workforce, and sustainability required for clinical biomarker data sharing. To provide guidance and assist in the navigation of these challenges, the Society for Immunotherapy of Cancer (SITC) Biomarkers Committee convened to outline the challenges that researchers currently face, both at the conceptual level (Volume I) and at the technical level (Volume II). The committee also suggests possible solutions to these problems in the form of professional standards and harmonized requirements for data sharing, assisting in continued progress toward effective, clinically relevant biomarkers in the IO setting.

Differences in key genes in human alveolar macrophages between phenotypically normal smokers and nonsmokers: diagnostic and prognostic value in lung cancer

OBJECTIVE

To explore the effect of smoking on gene expression in human alveolar macrophages and the value of identified key genes in the early diagnosis and prognosis of lung cancers.

METHODS

We downloaded three data sets (GSE8823, GSE2125, and GSE3212) from the Gene Expression Omnibus (GEO) database, including 31 non-smoking and 33 smoking human alveolar macrophage samples. We identified common differentially expressed genes (DEGs), from which we obtained module genes and hub genes by using STRING and Cytoscape. Then we analyzed the protein-protein interaction (PPI) network of DEGs, hub genes, and module genes and used David online analysis tool to carry out functional enrichment analysis of DEGs and module genes.

RESULTS

A total of 85 differentially expressed genes was obtained, including 42 up-regulated genes and 43 down-regulated genes. The Human Protein Atlas and Survival analysis showed that GBP1, ITGAM, CSF1, SPP1, COL1A1, LAMB1 and THBS1 may be closely associated with the carcinogenesis and prognosis of lung cancer.

CONCLUSION

DEGs, module, and hub genes identified in the present study help explain the effects of smoking on human alveolar macrophages and provide candidate targets for diagnosis and treatment of smoking-related lung cancer.

Laniakea: an open solution to provide Galaxy "on-demand" instances over heterogeneous cloud infrastructures

BACKGROUND

While the popular workflow manager Galaxy is currently made available through several publicly accessible servers, there are scenarios where users can be better served by full administrative control over a private Galaxy instance, including, but not limited to, concerns about data privacy, customisation needs, prioritisation of particular job types, tools development, and training activities. In such cases, a cloud-based Galaxy virtual instance represents an alternative that equips the user with complete control over the Galaxy instance itself without the burden of the hardware and software infrastructure involved in running and maintaining a Galaxy server.

RESULTS

We present Laniakea, a complete software solution to set up a "Galaxy on-demand" platform as a service. Building on the INDIGO-DataCloud software stack, Laniakea can be deployed over common cloud architectures usually supported both by public and private e-infrastructures. The user interacts with a Laniakea-based service through a simple front-end that allows a general setup of a Galaxy instance, and then Laniakea takes care of the automatic deployment of the virtual hardware and the software components. At the end of the process, the user gains access with full administrative privileges to a private, production-grade, fully customisable, Galaxy virtual instance and to the underlying virtual machine (VM). Laniakea features deployment of single-server or cluster-backed Galaxy instances, sharing of reference data across multiple instances, data volume encryption, and support for VM image-based, Docker-based, and Ansible recipe-based Galaxy deployments. A Laniakea-based Galaxy on-demand service, named Laniakea@ReCaS, is currently hosted at the ELIXIR-IT ReCaS cloud facility.

CONCLUSIONS

Laniakea offers to scientific e-infrastructures a complete and easy-to-use software solution to provide a Galaxy on-demand service to their users. Laniakea-based cloud services will help in making Galaxy more accessible to a broader user base by removing most of the burdens involved in deploying and running a Galaxy service. In turn, this will facilitate the adoption of Galaxy in scenarios where classic public instances do not represent an optimal solution. Finally, the implementation of Laniakea can be easily adapted and expanded to support different services and platforms beyond Galaxy.

Twenty years of bioinformatics research for protease-specific substrate and cleavage site prediction: a comprehensive revisit and benchmarking of existing methods

The roles of proteolytic cleavage have been intensively investigated and discussed during the past two decades. This irreversible chemical process has been frequently reported to influence a number of crucial biological processes (BPs), such as cell cycle, protein regulation and inflammation. A number of advanced studies have been published aiming at deciphering the mechanisms of proteolytic cleavage. Given its significance and the large number of functionally enriched substrates targeted by specific proteases, many computational approaches have been established for accurate prediction of protease-specific substrates and their cleavage sites. Consequently, there is an urgent need to systematically assess the state-of-the-art computational approaches for protease-specific cleavage site prediction to further advance the existing methodologies and to improve the prediction performance. With this goal in mind, in this article, we carefully evaluated a total of 19 computational methods (including 8 scoring function-based methods and 11 machine learning-based methods) in terms of their underlying algorithm, calculated features, performance evaluation and software usability. Then, extensive independent tests were performed to assess the robustness and scalability of the reviewed methods using our carefully prepared independent test data sets with 3641 cleavage sites (specific to 10 proteases). The comparative experimental results demonstrate that PROSPERous is the most accurate generic method for predicting eight protease-specific cleavage sites, while GPS-CCD and LabCaS outperformed other predictors for calpain-specific cleavage sites. Based on our review, we then outlined some potential ways to improve the prediction performance and ease the computational burden by applying ensemble learning, deep learning, positive unlabeled learning and parallel and distributed computing techniques. We anticipate that our study will serve as a practical and useful guide for interested readers to further advance next-generation bioinformatics tools for protease-specific cleavage site prediction.

doepipeline: a systematic approach to optimizing multi-level and multi-step data processing workflows

BACKGROUND

Selecting the proper parameter settings for bioinformatic software tools is challenging. Not only will each parameter have an individual effect on the outcome, but there are also potential interaction effects between parameters. Both of these effects may be difficult to predict. To make the situation even more complex, multiple tools may be run in a sequential pipeline where the final output depends on the parameter configuration for each tool in the pipeline. Because of the complexity and difficulty of predicting outcomes, in practice parameters are often left at default settings or set based on personal or peer experience obtained in a trial and error fashion. To allow for the reliable and efficient selection of parameters for bioinformatic pipelines, a systematic approach is needed.

RESULTS

We present doepipeline, a novel approach to optimizing bioinformatic software parameters, based on core concepts of the Design of Experiments methodology and recent advances in subset designs. Optimal parameter settings are first approximated in a screening phase using a subset design that efficiently spans the entire search space, then optimized in the subsequent phase using response surface designs and OLS modeling. Doepipeline was used to optimize parameters in four use cases; 1) de-novo assembly, 2) scaffolding of a fragmented genome assembly, 3) k-mer taxonomic classification of Oxford Nanopore Technologies MinION reads, and 4) genetic variant calling. In all four cases, doepipeline found parameter settings that produced a better outcome with respect to the characteristic measured when compared to using default values. Our approach is implemented and available in the Python package doepipeline.

CONCLUSIONS

Our proposed methodology provides a systematic and robust framework for optimizing software parameter settings, in contrast to labor- and time-intensive manual parameter tweaking. Implementation in doepipeline makes our methodology accessible and user-friendly, and allows for automatic optimization of tools in a wide range of cases. The source code of doepipeline is available at https://github.com/clicumu/doepipeline and it can be installed through conda-forge.