chromoWIZ: a web tool to query and visualize chromosome-anchored genes from cereal and model genomes

Abstract

Background

Over the last years reference genome sequences of several economically and scientifically important cereals and model plants became available. Despite the agricultural significance of these crops only a small number of tools exist that allow users to inspect and visualize the genomic position of genes of interest in an interactive manner.

Description

We present chromoWIZ, a web tool that allows visualizing the genomic positions of relevant genes and comparing these data between different plant genomes. Genes can be queried using gene identifiers, functional annotations, or sequence homology in four grass species (Triticum aestivum, Hordeum vulgare, Brachypodium distachyon, Oryza sativa). The distribution of the anchored genes is visualized along the chromosomes by using heat maps. Custom gene expression measurements, differential expression information, and gene-to-group mappings can be uploaded and can be used for further filtering.

Conclusions

This tool is mainly designed for breeders and plant researchers, who are interested in the location and the distribution of candidate genes as well as in the syntenic relationships between different grass species. chromoWIZ is freely available and online accessible at http://mips.helmholtz-muenchen.de/plant/chromoWIZ/index.jsp.

Background

Since the release of the sequenced genome of Arabidopsis thaliana in 2000 [1], more than 50 plant reference sequences have become available [2]. While the average genome size in Angiosperms is about 6 Gb [3], sequencing efforts have focused mainly on smaller-sized genomes (< 1 Gb), which serve as models for large and still unsequenced species or on more accessible crop plant genomes such as rice (Oryza sativa). The cereal species of the Pooideae subfamily, including bread wheat (Triticum aestivum), barley (Hordeum vulgare), and rice are among the most important crops and share a high degree of syntenic conservation on a genome-wide level [4],[5]. Among the crops, hexaploid bread wheat (T. aestivum, 2n = 6x = 42, AABBDD) contains the largest and most complex genome with a size of roughly 17 Gb [6]. Despite its high economic relevance – 20% of the calories consumed by the world’s population derive from bread wheat – its genome has so far not been completely assembled. It has taken several years to provide a reference sequence for even one chromosome (3B, [7]), which by itself exceeds the genome size of rice almost 3-fold. Recently, shotgun sequencing and flow cytometry provided the basis for a gene annotation of the complete bread wheat genome comprising ~124 k gene models [6]. Furthermore, for selected chromosomes or chromosome arms, a physical map has been established and genetically anchored (e.g. 1A [8],[9], 1BS [10], 3B [7],[11], 6A [12]). For barley an anchored physical map that covers 3.9 Gb cumulative map length has been released [13],[14], including 26 k high-confidence genes and comprises shotgun assemblies from three cultivars. Most shotgun contigs have already been anchored by population genetics. This approach, called POPSEQ [15], was also used to improve the anchoring of the physical map [13]. Like bread wheat and barley, Brachypodium (Brachypodium distachyon) also belongs to the Pooideae subfamily within the Poaceae family. It has a relatively small genome (~300 Mb) and has been widely used as a model organism to study the structure and evolution of other grass species [16]. Rice is another important member of the Poaceae family and represents one of the most important staple foods worldwide. To successfully integrate all the different resources, e.g. genetic information and gene expression measurements, for these crop species, heterogeneous datasets need to be combined. Therefore, tools and standards for interlinking anchored datasets are required (reviewed in [17]). One of the approaches for combining heterogeneous datasets is the “GenomeZipper” [4]. It establishes a virtual order of genes in plants without assembled chromosomes by exploiting the highly conserved synteny to smaller, already sequenced genomes. Large-sized syntenic regions, together with genetic marker sets enable an anchoring of most genes for larger-sized cereals including e.g. barley [14], rye (Secale cereale) [18] and Aegilops tauschii [19]. Since after the split from their common ancestor, the position of most genes was conserved, this approach provides robust approximations of the gene positions and order [20].

A small number of tools exist that allow users to inspect the genomic position of query genes in target genomes. For barley it is possible to map query sequences by using IPK Viroblast (http://webblast.ipk-gatersleben.de/) or barleymap (http://floresta.eead.csic.es/barleymap/). However, to our knowledge, no web-based tool exists that covers several genomes and allows calculating and visualizing the gene density along the chromosome. This is especially of importance when several dozen genes need to be mapped, e.g. for analyzing a quantitative trait locus (QTL). Transcriptome-oriented studies might reveal a set of gene candidates and the corresponding genomic position supports in removing false-positives gene candidates and defining the genetic or physical location of the QTL. None of the listed tools provide queries based on functional annotation or the integration of expression data. As part of the GenomeZipper, we have previously implemented a module ‘chromoWIZ’ which was introduced to ease detection of syntenic regions for a yet unassembled genome and several sequenced and assembled genomes including Brachypodium [16], rice [21] and sorghum (Sorghum bicolor) [22]. Here, we describe the web-based version of chromoWIZ along with new features. Originally, chromoWIZ was restricted to local use as part of the GenomeZipper package and allowed a mapping of genes or shotgun contigs of one chromosome or chromosome arm against the reference genomes Brachypodium, rice and sorghum. To find genomic positions for genes of interest, in the latest, web-based version functional annotations and sequence homology can be used to find the corresponding regions within the corresponding genome. For grouped or clustered genes chromoWIZ now visualizes the physical position in a group-wise manner. In its latest version, chromoWIZ integrates the anchoring results of both the International Barley Genome Sequence Consortium (IBSC [14]), and the International Wheat Genome Sequencing Consortium (IWGSC [6]) and allows comparing sequences against the genomes of Brachypodium and rice. This tool is mainly designed for breeders and plant researchers, who are interested in the location and the distribution of candidate genes as well as in the syntenic relationships between different grass species. In order to illustrate the features of chromoWIZ and to explain the basic work-flows, we present different use cases. The application website can be accessed at: http://mips.helmholtz-muenchen.de/plant/chromoWIZ/index.jsp without any restrictions.

Construction and content

chromoWIZ runs on a webserver at the PGSB site [23]. The tool’s back-end is implemented in the programming language Python. The front-end uses native HTML and Javascript for data visualization and navigation. Mapping information and gene information were collected from the official releases of the Brachypodium, rice, barley and bread wheat genomes [6],[14],[16],[21]. For Brachypodium protein and coding sequences, as well as functional annotation information were collected from the PGSB database [23] using gene models’ version 1.2. For barley we integrated the datasets that were provided with the genetically anchored physical map [14], which is hosted at ftp://ftpmips.helmholtz-muenchen.de/plants/barley/public_data. For bread wheat, gene models from version 2.2 (ftp://ftpmips.helmholtz-muenchen.de/plants/wheat/IWGSC) were included. The MSU7 annotation has been integrated for rice [21]. More details for the currently used datasets and the corresponding updates are provided on the chromoWIZ web site.

Utility

Application of chromoWIZ

chromoWIZ allows visualizing the location of anchored genes along chromosomes on the basis of functional gene annotations, sequence homology or gene lists. So far, the web tool includes the crop species bread wheat (T. aestivum), barley (H. vulgare) and the closely related but much smaller Brachypodium (B. distachyon) and rice (O. sativa) genomes. Anchored genes are clustered together along the chromosome in non-overlapping genomic or genetic intervals, referred to as bins. In Brachypodium and rice, every bin represents one megabase (Mb) of non-overlapping chromosomal sequence. For barley 10 Mb and for bread wheat 5 CentiMorgan (cM) intervals are used. Bins are visualized as heat maps to enable an intuitive view along the entire chromosomes. The genomic positions in barley are highlighted relative to the anchored physical BAC contigs which were strung together to form virtual chromosomes. All genes within chromoWIZ are linked to external databases providing additional information on the gene models (e.g. for bread wheat EnsemblePlants http://plants.ensembl.org/Triticum_aestivum/Info/Index). The sequences of tagged genes within a bin can be downloaded in the FASTA format. To obtain the genomic location for genes of interest, referred to as “tagged genes”, chromoWIZ provides several search methods (Table 1): By sequence homology a set of query sequences can be mapped against the annotated gene models using nucleotide or protein BLAST searches, requiring a predefined e-value and sequence identity. Alternatively, if known, a list of species-specific gene identifiers can be directly provided instead of sequences. To query families of genes (e.g. genes sharing a specific Gene Ontology (GO) term or PFAM domain [24],[25], an annotation-based approach has been included. The distribution of query genes is visualized by heat maps, which depict the relative distribution of the query-matching genes compared to the overall number of genes along the chromosomes. In addition, the overall gene distribution is shown, as the number of anchored genes varies between the different bins. To see whether certain chromosome (−arms) are enriched for tagged genes an enrichment analysis is provided. The significance of over-representation of genes tagged is assessed by a one-sided Fisher’s exact test and a Bonferroni adjustment of P values. Furthermore, labeled groups of genes can be included, e.g. genes being clustered or co-expressed or that were grouped together based on sequence similarity to allow for a group-wise visualization and analysis. The Data Manager is a part of chromoWIZ that enables the upload of various user-specific datasets and performs a validation of input data prior to integration into the chromoWIZ search interface. These data are subsequently only visible for the respective user and available for 24 hours before they are being automatically removed from the servers. Gene expression is an important factor for judging the relevance of candidate genes. In chromoWIZ, by using the Data Manager, users can optionally upload expression values for their genes of interest. Similar to expression data, information about differential expression can be provided. With expression data at hand, functional information can be combined with the genomic positions.

Table 1 A variety of search features are provided by chromoWIZ

The following use cases illustrate different aspects of chromoWIZ. The first use case describes how candidate genes can be mapped against the reference genome sequences using the gene identifiers, sequence-based searches or functional annotations. The second use case illustrates how a list of genes can be filtered based on their expression or by including information about differential expression. In the third use case we show how chromoWIZ allows highlighting syntenic regions between bread wheat and Brachypodium or barley. In the fourth use case we use published expression data to illustrate how the gene-to-group information can help in refining the genomic position of a resistance QTL. This is granted by transferring data from ancient to recent reference sequences. The fifth use case finally gives an example of how chromoWIZ can be applied for comparative genomic analysis.

Use case 1: finding genes using identifiers, sequence similarity or annotations

One of the very basic functionalities of chromoWIZ is searching and visualizing genes by their identifiers. Given a set of species-specific gene identifiers their genomic position can be highlighted. In case no identifiers are available, an alternative approach is to provide sequence information for the corresponding genes. To illustrate this feature, we use the following example: A list of 19 gene identifiers from Brachypodium, preselected from a particular genomic region, was provided to the search interface (the gene identifiers are given in Additional file 1). chromoWIZ provides two outputs: First a heat map which depicts the number of all anchored genes along the chromosomes per bin (Figure 1A), and secondly, a heat map showing only the anchored genes that meet the query criteria (tagged genes) is shown (Figure 1B). For the given example the corresponding bin (bin9, 9-10 Mb) on chromosome 5 is highlighted. To illustrate the sequence-based search, we first extracted the gene sequences from this bin, by using the FASTA export functionality of chromoWIZ. This set of sequences was then provided to the search interface in order to perform a homology-based search. By only visualizing matches below an e-value of 10E-5, sequence identity of 100% and by requiring a best bidirectional match (flag ‘BBH’ has to be set) we again retrieved the bin containing the genes.

Figure 1

Heat map visualization of gene density. chromoWIZ visualizes the gene distribution of (A) all genes anchored as compared to (B) the number of genes matching the query criteria. The tooltip reports the relative and absolute number of tagged genes per bin.

Besides the gene identifier and homology-based search, chromoWIZ also offers a search by gene annotation functionalities. A user might be interested in a particular gene family and would like to analyze whether members of that family have increased or decreased copy numbers compared to other genomes. One way to analyze differences in copy numbers is to compare the amount of genes on the basis of protein families (PFAM [25]) or Gene Ontology (GO [24]) terms and chromoWIZ includes annotation information from these sources. In the given example, we aimed at visualizing all genes that are annotated under the Gene Ontology (GO) term GO:0043565 (sequence-specific DNA binding) that e.g. comprises transcription factors. In Brachypodium, we found matches to 349 genes, in bread wheat matches to 421 (732 including genetically unanchored) genes, and in barley we found matches to 225 (340) genes.

Use case 2: filtering for differentially expressed genes and usage of expression constraints

RNA-seq data is commonly used to analyze gene expression on a genome-wide level. It can efficiently be processed by means of analysis pipelines such as Cufflinks [26] or HTSeq [27]. After finding gene candidates based on their expression patterns it is often of interest to explore their respective genomic position. chromoWIZ provides features for combining expression data with positional information: (i) gene-to-group information can be provided. (ii) lists of differentially expressed genes can be included, and (iii) expression data of all genes can be integrated. Figure 2 shows the Data Manager input and the extended query features on the entry site, which are available once the data sets are included. Gene-to-group information is provided by an input file where the first column contains the gene identifier and the second column defines the group. The differentially expressed genes (DEGs) are provided via an input file that contains the gene identifiers. Also expression information can be provided in a file, where columns represent the conditions of interest. Details about the file formats are given on the chromoWIZ help page. When information about differentially expressed genes is included, the user can specify whether only differentially expressed genes should be queried. If expression information is included, genes can be filtered by two criteria: Either by a ‘Minimum expression’ criterion, meaning that at least in one condition the expression must exceed a given threshold. Alternatively, to find genes with expression variation across conditions, a user can set a ‘CV’ (coefficient of variation, given by dividing the sample standard deviation by the sample mean) filter, to only keep genes with a minimum required CV value.

Figure 2

Integration of gene expression information. Gene expression information, lists of differentially genes, and/or gene-to-group mapping data can be uploaded for enabling expression-based querying of genes. The different color codes highlight the search options, which become available after uploading the corresponding data.

For illustration we extracted 692 barley transcripts that are differentially expressed between two Tibetan wild barley genotypes in response to low potassium treatment [28]. The transcript sequences as given in Table S3 of [28] (http://www.plosone.org/article/fetchSingleRepresentation.action?uri=info:doi/10.1371/journal.pone.0100567.s009) were mapped against the genetically anchored barley gene models using BLASTN (sequence identity greater than 95%, e-value of 10E-10, BBH criterion). The 450 matching genes were compiled into a list of differentially expressed genes (Additional file 2) and uploaded by using the Data Manager. When searching barley for anchored differentially expressed genes we obtained 286 hits scattered across the different chromosomes.

Use case 3: pronounced syntenic regions shared in grass species

chromoWIZ has been repeatedly used to define and refine syntenic regions among related reference genomes [29],[30]. For illustration, we used gene models of bread wheat chromosome 4A [6] and to initiated a sequence homology search against Brachypodium and barley genes. In total 4,830 genes are annotated on chromosome 4A and the corresponding sequences were extracted and aligned against both genomes using BLASTN (sequence identity of at least 70% and an e-value of 10E-5, best bidirectional hit). We found matches against chromosomes 1 and 4 in Brachypodium and a rearrangement of an approximately 3 Mb genomic region that was shifted from the short arm of chromosome 1 to the long arm (Figure 3A). Additionally, in chromosome 4, the centromeric and peri-centromeric near regions were tagged. When bread wheat chromosome 4A was compared against barley, besides the largely homeologous chromosome 4H, syntenic regions on chromosome 5H and chromosome 7H were found, comprising genomic regions of 40 Mb respectively (Figure 3B). These findings are consistent with the documented chromosome rearrangements of bread wheat chromosome 4A [31].

Figure 3

Synteny between bread wheat chromosome 4A, Brachypodium and barley. Using chromoWIZ, genes from the bread wheat chromosome 4A were mapped against Brachypodium (A) and against barley (B) in order to highlight syntenic regions.

Use case 4: providing cluster information for tagging genes

Clustering genome-wide expression data into meaningful subsets has become a standard procedure in many transcriptome-oriented studies. Several methods enable to perform such a partitioning of data, e.g. by hierarchical clustering, k-means clustering or network-based approaches. chromoWIZ provides support for group-wise analyses as it allows uploading gene-to-group information. The example data for this use case originates from a co-expression network study assessing the effect of fungal pathogens on different bread wheat lines [32]. The five bread wheat lines in this study were characterized by the presence or absence of particular quantitative trait loci (QTL), which confer different resistance levels. This data has been used to infer a co-expression network with the Weighted Correlation Network Analysis approach (WGCNA, [33]). WGCNA can be utilized to find clusters of highly connected genes, so called network modules, based on inferring a correlation-based weighted gene network. After mapping the bread wheat transcriptome data to a 454 sequencing based whole genome assembly [34] and after quantifying the expression using Cufflinks [26], we observed eight different modules which represented distinct expression patterns containing 3,273 genes in total. One module was of particular interest as the related gene expression depicted a pronounced response to the fungal pathogen. The corresponding nucleotide sequences are given in Additional file 3. Using chromoWIZ those transcripts were mapped against the bread wheat genome survey sequence [6] by requiring a best bidirectional match and sequence identity of at least 95%. A significant enrichment for chromosome (−arms) 3B, 5BL, and 7DL was found (Figure 4). This is in support of the experimental set-up as one of the major Fusarium head blight resistance QTLs (Fhb1) that segregates between resistant and susceptible lines and is located on the short arm of chromosome 3B [35].

Figure 4

Chromosome (−arm) enrichment of genes responsive to a fungal pathogen. Bread wheat chromosome (−arm) enrichment for genes, which were responsive to Fusarium graminearum. Chromosome (−arms) 3B, 5BL, and 7DL are found to be significantly enriched for these genes.

Use case 5: comparative genomics in chromoWIZ for analyzing UDP-gylcosyltransferases

chromoWIZ can be used to detect homologous genes and their locations in the four cereal and model genomes using the implemented BLAST searches. To illustrate this, we searched for Brachypodium UDP-glycosyltransferases (UGT) homologous genes in rice, barley, and bread wheat. The Brachypodium UGT gene family contains five members of which several encode for the ability to inactivate the mycotoxin deoxynivalenol (Additional file 4) [36]. Deoxynivalenol is a potent inhibitor of protein biosynthesis produced by Fusarium graminearum, which is a pathogen also to wheat and barley [37]. The presence/activity of such UGTs may confirm high resistance. Yet, their identification remains challenging also due to the sheer size of the UGT superfamily, which comprises 178 members in Brachypodium and probably several hundred in bread wheat [36]. chromoWIZ mapped these six genes to the third and fourth bin on chromosome 5 in Brachypodium. In order to find putative orthologous genes, we extracted the sequences and mapped them against rice, bread wheat and barley. In barley, matches were found to the 2H (3) and 5H (1) chromosomes using 70% identity and e-value of 10E-5 as search criterions. In addition a match to a yet genetically unanchored gene was found. In rice, matches on chromosome 4 (8) and chromosome 9 (1) were observed, confirming previous findings [36]. In bread wheat matches to 2A (1), 2B (1), 2D (1) and 5A (1) indicate possible homoeologous gene-clusters on linkage group 2, however most genes (13) did not receive any genetic position yet. No matches were observed for chromosome 3B containing the Fhb1 locus [35], which was previously shown to govern the higher ability to inactivate the toxin [38].

Discussion

chromoWIZ allows searching for candidate genes and visualizing their density and localizations along chromosomes of selected grass genomes. Genes can be searched by using several options, e.g. by gene identifiers, by functional annotation, by sequence homology search or by gene-to-group mappings. The tool is implemented in a flexible way to ensure that novel genomes or updates of existing genomes can be easily undertaken. Export features are provided and extended functionality is activated if gene expression data or clustering information is provided.

chromoWIZ enables the integration of expression-based information to filter for candidate genes

While there are several tools that provide information, mapping, and visualizations capabilities with respect to syntenic relationships in plant genomes [39],[40], there is a lack for tools to query and interactively inspect genetically and physically anchored genes. One of the major advantages of chromoWIZ over other tools such as barleymap (http://floresta.eead.csic.es/barleymap/) or IPK Viroblast (http://webblast.ipk-gatersleben.de/barley/) is that expression data can be included to filter by several criteria and thereby selecting the most relevant genes. In addition, clustering information and gene-to-group mappings such as sets of co-expressed genes, selected gene families and/or differentially expressed genes can be included and independently analyzed. The different datasets can be imported by using the Data Manager as intrinsic part of the chromoWIZ web application. After uploading the data additional filtering and search options appear on the entry page (Table 1 and Figure 2).

chromoWIZ enables transferring previous results to the current reference sequences

chromoWIZ allows linking gene anchoring information with the annotated gene information and provides access to the gene candidates and their localization as well as to their neighboring genes. With actively ongoing projects and the consequential updates of the reference sequences of bread wheat and barley, data need to be mapped to a common reference sequence to compare previous results against current ones. We demonstrated this approach by using a particular gene co-expression module that comprised the major response of bread wheat genes against a fungal pathogen [32]. As shown in use case 4 chromoWIZ allowed transferring previous analysis [32] onto updated resources by mapping from an earlier bread wheat genome draft [34] to more recent chromosome-arm sorted shotgun contigs [6].

chromoWIZ enables to detect larger syntenic blocks within yet unfinished genomes

For (novel) grass genomes, chromoWIZ can be used to detect and analyze syntenic regions with respect to Brachypodium, rice, barley, and bread wheat. In use case 3, annotated gene models of bread wheat chromosome 4A were used to detect syntenic regions in comparison to barley and Brachypodium (Figure 3). This chromosome is of particular interest, because in most cases barley and wheat chromosomes are collinear [4]. For this specific chromosome syntenic regions appeared also on barley chromosomes 5H and 7H [31]. Furthermore, when arm sorted chromosome datasets become available for a newly sequenced but not yet assembled genome, chromoWIZ can help to allocate genes to corresponding syntenic regions in barley, rice, bread wheat, and Brachypodium. Thereby, it offers a first glance at the genome structure of these plants, particularly for revealing rearrangements and introgression and to analyze more complex nested syntenic structures.

Conclusions

chromoWIZ provides a valuable and user-friendly interface to access anchored genes for agriculturally important crops and model genomes. By using the different query options it is possible to flexibly narrow down regions of interest and/or gene candidates. With future updates it is planned to include more species and to extend the range of features prior to allow interactive and integrative searches on evolving large and complex crop plant genomes.

Availability and requirements

chromoWIZ is freely available without any restrictions at http://mips.helmholtz-muenchen.de/plant/chromoWIZ/index.jsp.

License: Not required.

Any restrictions to use by non-academics: None.

Availability of supporting data

The data sets supporting the results of this article are included within the article (and its additional files).