Genomics pipelines and data integration: challenges and opportunities in the research setting

Abstract

The emergence and mass utilization of high-throughput (HT) technologies, including sequencing technologies (genomics) and mass spectrometry (proteomics, metabolomics, lipids), has allowed geneticists, biologists, and biostatisticians to bridge the gap between genotype and phenotype on a massive scale. These new technologies have brought rapid advances in our understanding of cell biology, evolutionary history, microbial environments, and are increasingly providing new insights and applications towards clinical care and personalized medicine. Areas covered: The very success of this industry also translates into daunting big data challenges for researchers and institutions that extend beyond the traditional academic focus of algorithms and tools. The main obstacles revolve around analysis provenance, data management of massive datasets, ease of use of software, interpretability and reproducibility of results. Expert commentary: The authors review the challenges associated with implementing bioinformatics best practices in a large-scale setting, and highlight the opportunity for establishing bioinformatics pipelines that incorporate data tracking and auditing, enabling greater consistency and reproducibility for basic research, translational or clinical settings.

Keywords: ExomeSeq; High throughput sequencing; RNAseq; analysis provenance; bioinformatics best practices; bioinformatics pipelines; genomic data management; reproducible computational research; variant calling.

Figure 1. Annual DNA Sequencer Data Generation

A: Graphical presentation of annual HTS data generation is presented for 2016, and extrapolated through 2021. B: Tabular presentation of HTS for 2015 and 2016 and extrapolation through 2025. Both original data from HTS sequencing (Petabytes) and data expansion from analysis (Petabytes) are presented.

Figure 2. The Shifting Cost & Complexity of Sequencing

The shifting cost & complexity of sequencing (sample preparation, data processing, downstream analysis and data management) from the early days of HTS through the present and extrapolation through 2025.

Figure 3. OnRamp BioInformatics Genomics Research Platform Architecture

The architecture of the Genomics Research Platform provides a unified system for genomic analysis and data exploration. The GRP consists of four major modules, Intuitive User Interface, Analysis Engine, BioInformatics Data Management, and Distributed File System & Storage Management.

Figure 4. Flow Chart summarizing the analyses of RNAseq data

In each box, the upper part in bold describes the analytical step, and the bottom part in plain text notes the programs or services that can be used. FPKM = Fragments Per Kilobase of transcript per Million mapped reads, THP Atlas = The Human Protein Atlas, GSEA = Gene Set Enrichment Analysis).

Figure 5. FASTQC Report

A: High Quality Data. B: Data with potential RNAseq library preparation issues, overrepresented sequences and duplication that should be removed prior to any downstream analysis.

Figure 6. Flow Chart summarizing the analyses of DNAseq data (whole genome and exome)

In each box, the upper part in bold describes the analytical step, and the bottom part in plain text notes specific programs or services that can are employed. Labels on the right indicate a general categorization of the analysis into primary, secondary, and tertiary.