Test of Arabidopsis Space Transcriptome: A Discovery Environment to Explore Multiple Plant Biology Spaceflight Experiments

Abstract

Recent advances in the routine access to space along with increasing opportunities to perform plant growth experiments on board the International Space Station have led to an ever-increasing body of transcriptomic, proteomic, and epigenomic data from plants experiencing spaceflight. These datasets hold great promise to help understand how plant biology reacts to this unique environment. However, analyses that mine across such expanses of data are often complex to implement, being impeded by the sheer number of potential comparisons that are possible. Complexities in how the output of these multiple parallel analyses can be presented to the researcher in an accessible and intuitive form provides further barriers to such research. Recent developments in computational systems biology have led to rapid advances in interactive data visualization environments designed to perform just such tasks. However, to date none of these tools have been tailored to the analysis of the broad-ranging plant biology spaceflight data. We have therefore developed the Test Of Arabidopsis Space Transcriptome (TOAST) database (https://astrobiology.botany.wisc.edu/astrobotany-toast) to address this gap in our capabilities. TOAST is a relational database that uses the Qlik database management software to link plant biology, spaceflight-related omics datasets, and their associated metadata. This environment helps visualize relationships across multiple levels of experiments in an easy to use gene-centric platform. TOAST draws on data from The US National Aeronautics and Space Administration's (NASA's) GeneLab and other data repositories and also connects results to a suite of web-based analytical tools to facilitate further investigation of responses to spaceflight and related stresses. The TOAST graphical user interface allows for quick comparisons between plant spaceflight experiments using real-time, gene-specific queries, or by using functional gene ontology, Kyoto Encyclopedia of Genes and Genomes pathway, or other filtering systems to explore genetic networks of interest. Testing of the database shows that TOAST confirms patterns of gene expression already highlighted in the literature, such as revealing the modulation of oxidative stress-related responses across multiple plant spaceflight experiments. However, this data exploration environment can also drive new insights into patterns of spaceflight responsive gene expression. For example, TOAST analyses highlight changes to mitochondrial function as likely shared responses in many plant spaceflight experiments.

Keywords: Arabidopsis thaliana; Qlik; RNAseq; bioinformatics; microarray; proteomics; spaceflight; transcriptomics.

Figure 1

Publicly available spaceflight transcriptomics datasets. (A) Relationships between species, ecotype, genotype (i.e., mutant or wild type) growth environment and assay technique for datasets from plants experiencing spaceflight. (B) Relationships between species, ecotypeand genotype versus the tissue or organ type that was sampled to generate the tspaceflight-related transcriptomics dataset. Col-0, Columbia ecotype of Arabidopsis thaliana; Ws, Wassilewskija ecotype; Ler, Landsberg erecta ecotype; Cvi, Cape Verdi Islands ecotype; mutants of Arabidopsis: phyD, Phytochrome D; arg, Altered Response to Gravity; act2, Actin 2; phyD, phytochrome D; hsfa2, heat shock factor a2, Wt, wild type; BRIC, Biological Research in Canister; ABRS, Advanced Biological Research System; EMCS, European Modular Cultivation System.

Figure 2

Database structure underlying TOAST 4.5. Each dataset within TOAST includes a series of pre-computed factors for each gene: minimally including fold-change, P-value, Q-value, and a yes/no value for whether the fold-change for each gene is significant at P < 0.05. These pre-computed values greatly speed the real-time processing of interactive visualizations within the TOAST user interface. The identifiers in the raw data, such as Transcript ID from RNAseq, Probe ID for Microarray, or TAIR ID are translated to their unique Entrez and Ensembl IDs to allow for uniform indexing within TOAST itself and to facilitate passing of analyzed data produced by TOAST analyses to exterior sites and tools. Within TOAST, the strings of molecular ID's from a dataset are both directly transferred to a series of data visualization and exploration tools and are imported into a series of analytical packages accessing a range of databases that have been imported into the TOAST environment. These databases include: the Genome Ontology (GO) consortium databases that allow analysis of the relationships between gene lists of interest and known biological processes, the SUBA4 database which catalogs predicted subcellular locales for each gene, the Kyoto Encyclopedia of Genes and Genomes (KEGG) database that analyzes relationships to known cellular pathways, and Ensembl's Orthologous Matrix database, allowing TOAST to make comparisons between species. The outputs of these analytical modules are then passed to TOAST's interactive data visualization tools to help explore each dataset. Results from the visualizations are in turn returned as lists of Gene IDs to allow for reiterative analyses.

Figure 3

The TOAST 4.5 user interface. (A) The web interface for TOAST launches an overview menu of dashboard icons allowing the user to directly access the introductory materials, omics data, and related analysis tools. (B) Each icon provides a visual summary of the data or tools that it links to including elements such as spaceflight vehicle (e.g., Shuttle, ISS, Shenzou vs ground-based experimentation), the growth hardware used, plant/seedling vs cell culture experiment, RNAseq vs microarray vs proteomics, species and ecotype and dataset identifier (e.g., GLDS number).

Figure 4

Graphical user interface for typical dataset. Clicking on a volcano plot also activates an interactive graphical tool for manual selection of groups of genes of interest. *Defaults to showing 33.43K, i.e., all Entrez identifiers, until a filter or gene selection is applied. Inset, a lasso tool allows user selection of data points from volcano plot in addition to activation of filters such as on significance of change, KEGG Pathway, or GO annotation.

Figure 5

Overview of use of the TOAST 4.5 database. (1a) The user selects an initial study of interest and then can review the summary of its metadata to ensure it is the correct focus for study (1b). The dataset is then opened and (2) when the study is selected an interactive dashboard launches and the user has a direct link to any associated manuscript. Gene filtering: statistical (3a), gene ontology (3b), and other related functional filters can be applied to focus the number of loci being visualized in the volcano plot (3c) to genes of interest. In addition, the volcano plot itself can be interactively manually filtered using a graphical selection tool. All filters can be toggled on and off using selectable tabs at the top of the interface (3d). If an interesting subset of loci are selected the user can activate the download option (4a) and save the related data in word or xml format (4b). (5) The user can also perform further bioinformatic and statistical analysis with other online tools linked from the main user interface.

Figure 6

Analysis of metadata within the TOAST 4.5. (A) Initial dashboards allow access to comparisons between a range of experiment-related factors such as lighting conditions, growth environment, and plant genotypes. (B) A typical dashboard for metadata exploration, in this case for light conditions and age of seedling. Preset filters for e.g., lab group performing the research and growth and radiation environments are available to the user and the identity of the filtered datasets is shown in the bottom left window.

Figure 7

TOAST confirms the “high light early” ROS response from spaceflight data. The “high light early” clade in the ROS wheel analysis represents 8.87K transcripts from a total of 21.33K transcripts detected, or 41.5% of all transcripts.

Figure 8

Analysis of mitochondrion-related genes altered by spaceflight. (A) Screenshot depicting an example of a user's interaction with the TOAST graphical user interface to define mitochondrion-related transcripts. (B) Using TOAST for iterative filtering of differentially expressed genes across multiple spaceflight studies where plants were light grown. (C) More extensive analysis of the studies in (B) using differentiation within the individual datasets for different analytical approaches (microarray vs RNAseq) and for different analysis periods (4 days vs 8 days). (D) Similar analysis but for dark grown plant samples. (E) The effects of spaceflight on the alternative oxidase gene family in dark grown samples. Maximum likelihood tree of AOX gene family generated using ClustalW alignment with Mega-X software (www.megasoftware.net). Venn diagrams plotted using jvenn (Bardou et al., 2014).