The InterMine Android app: Cross-organism genomic data in your pocket

Data sources

The InterMine app allows users to search a default subset of InterMine data warehouses (Table 1). This list is configurable, and so users are able to refine or add mines to match their interests. See https://registry.intermine.org for the full list of known public InterMine instances.

InterMine databases typically integrate data from many resources. For instance BioGRID¹³, IntAct¹⁴, UniProt¹⁵, and can include high quality curated data (from the Model Organism Databases), genome-wide high-throughput data and data from smaller more focused studies (See individual InterMine websites for more details). All the InterMine databases accessible from the app are open source (License LGPL 2.1) and the data within them is free to access and download. Some individual InterMines may have restrictions for commercial use of the data and each individual InterMine should be consulted for its policy. See Table 1 for URLs.

Search and analysis

The InterMine app provides several ways to search and explore the data available, including a keyword search, sets of pre-defined template searches and list analysis functionality. These features are described in more detail in the use case section.

MyMine accounts

InterMine databases allow users to create an account through which they can, between sessions, store lists and searches. The InterMine App therefore allows users to log in to any accounts they hold on the underlying databases, so enabling user-created lists to be accessed.

Favourite genes

Users are also able to mark genes in search results as favourite. These genes are stored on the Android device and can be accessed without needing to log in to any of the underlying databases.

Communication

The InterMine database design and the webservices used to power the InterMine app have been previously described^1,16. The InterMine App draws all data from the RESTful Application Programming Interfaces (APIs) that InterMine databases provide¹⁶. RoboSpice, an Open Source (Apache 2.0 Licence) Android communications library¹⁷, provides core network communication functionality. Data are loaded asynchronously over HTTP or HTTPS, depending on the preferred protocol of the database being accessed. For performance enhancement, most web service responses are stored on the device if headers state an appropriate cache lifetime.

When the app receives a JSON response from the web service, it transforms the data from a table-structured format to a more hierarchical view, which presents data more effectively on smaller-screened mobile devices.

Authentication

Each InterMine database is discrete and often maintained by different organisations. If a user wishes to authenticate with multiple InterMines - perhaps to view private gene lists stored on different databases - they will need to provide separate authentication details for each InterMine database. However this is only necessary once, as after a user has successfully authenticated in a given InterMine via a username/password pair, the app retrieves and stores an API authentication token. This ensures that the user can authenticate in the future without having to re-enter or store sensitive password details.

All of the user configuration settings and authentication tokens are stored locally on the device via SharedPreferences, Android’s dedicated settings storage mechanism¹⁸.

Internal storage

Tabular data, such as favourite genes within the app, are not suited to the key/value pair storage used in SharedPreferences¹⁹, and therefore are stored within an SQLite database on the user’s device. Data stored include the InterMine instance the data originated from, the (e.g.) gene’s identifiers, description, organism, and genomic coordinates.

Search

Keyword search is available across lists, templates and gene search results. Search results from different databases are presented to the user as a single result set, sorted by the search relevance score generated by each originating database. Search results can be shared via email, instant messaging, and other sharing media in text format, using Android’s ACTION_SEND Intent functionality²⁰. Further data export options are available through links to the relevant full InterMine database instance.

Advanced information

InterMine also includes advanced analysis tools - particularly data visualisations - which may not be available via the API. To access the extended information about genes or gene lists, a user can load InterMine’s advanced report pages within the app itself. This is implemented via Android’s WebView²¹ functionality which allows live web pages to be embedded in an application (for example, Figure 2 shows an example of an embedded InterMine WebView).

Cross-organism gene search

The keyword search simultaneously searches all InterMine databases selected through the settings option. Thus, a cross-organism overview of data available for further investigation is provided. Link-outs from the search results to each originating InterMine database provide access to detailed gene report pages. These pages collate information integrated for that gene and typically include functional summaries, ontology annotation, pathway, expression, interaction and disease data and links to additional related data.

As an example, searching for ‘ dopamine’ returns dopamine-related genes from PhytoMine, MouseMine, HumanMine, TargetMine, FlyMine, RatMine, ZebrafishMine, WormMine, YeastMine, ThaleMine and HymenoperaMine (Figure 1; see Table 1 for urls). Selecting a gene from the results, for instance the human gene DRD4 (dopamine receptor D4) displays summary information about the gene, with a link to the full gene report available from the HumanMine database. Here we learn, for example, that polymorphisms in the DRD4 gene are associated with the disorder attention deficit hyperactivity disorder (ADHD) (Figure 2), a condition associated with low dopamine levels. The search results therefore facilitate rapid exploration across multiple organisms and kinds of data.

Figure 1. Example keyword search results.

A search for dopamine returns genes from PhytoMine, MouseMine, HumanMine, FlyMine, TargetMine, RatMine, ZebrafishMine, YeastMine and ThaleMine. A section of the results from HumanMine and FlyMine is shown.

Figure 2. The human DRD4 report page from the HumanMine webapp, displayed in the InterMine app.

(a) This view shows the first part of the report page with a summary of the Gene identifiers and data available. (b) This view shows the Disease section of the report page.

Template searches

In addition to cross-organism gene search, the InterMine databases provide libraries of pre-built searches, called template searches. Such searches provide a user-friendly interface where the parameters for search filters can be specified. These templates range from simple searches, such as for a specified gene (or genes), return the corresponding Gene Ontology terms (represented as “ Gene → Gene Ontology terms”), to more complex searches combining data of more than one type, such as “ Tissue + interaction → genes”, which returns any genes expressed in the specified tissue that also interact with the product of a specified gene. Templates from each InterMine database are available within the InterMine App. The results are provided with a simple keyword search to facilitate further data browsing. For instance, continuing the above dopamine example, we can use a template search to identify genes in Drosophila associated with ADHD: on the templates page for the FlyMine database, we find the template “Disease -> Human genes and Orthologues” (Figure 3). This template allows one to specify disease names that contains “attention deficit”, and on running the template, this returns the disease Attention deficit-hyperactivity disorder along with associated human genes and their orthologues from the available InterMine databases. Using the ability to search within the results we are able to verify that the human gene we are interested in (DRD4) is associated with this condition, and that this gene has a predicted orthologue in Drosophila melanogaster, FBgn0053517, Dop2R (Figure 4). Through such iterative searching we can continue our investigation of this fly orthologue to identify, for instance, interacting partners, pathway and Gene Ontology annotations.

Figure 3. A FlyMine template search, “Disease -> Human genes and Orthologues”.

The disease name is an editable constraint, in this case set to search for human genes associated with diseases with a name that “contains” Attention deficit, as well as their orthologues.

Figure 4. Results from the template “Disease -> Human genes and Orthologues”, with disease constraint set to “contains” Attention deficit, available from FlyMine.

Shows the human dopamine receptor D4 associated with the disease attention deficit-hyperactivity disorder and with the D. melanogaster orthologue FBgn0053517, Dop2R.

Lists

InterMine databases are especially suited to the analysis of lists of genes or other entities. Users can create their own lists, which can be stored between sessions if the user has an account for the relevant InterMine database. Again, direct links from lists to the underlying InterMine databases provide access to list analysis tools, for instance enrichment statistics that help identify surprising properties, such as publications that cite an unexpectedly large number of the list members, or GO terms or protein domains that are associated with an unexpectedly large number of list members.

Public lists, which are typically interesting sets of genes derived from publications and other studies, are often provided by the database operators. For instance, in FlyMine, eleven of the public lists provide sets of genes whose expression increases at defined times during drosophila embryogenesis, as derived from Hooper et al.²⁴. Further lists show genes that are expressed at increased levels in various adult fly tissues according to data from the FlyAtlas resource²⁵. Within one of these sets, PL FlyAtlas_brain_top, we can identify a set of genes up-regulated in the brain. Checking within this list, we find that the dopamine receptor gene Dop2R (FBgn0053517) identified above is present. By following the link to the corresponding list analysis page on the underlying FlyMine website, and examining the enrichment statistics, we find that the Dop2R gene is part of a set in which the Gene Ontology term dopamine receptor signaling pathway (GO:0007212) occurs unexpectedly frequently (p-value 0.001303, with Holm-Bonferroni correction). It is also apparent that two other fly dopamine receptors, Dop1R1 (FBgn0011582) and Dop1R2 (FBgn0266137) are also found in this list (Figure 5).

Figure 5. Gene Ontology enrichment analysis of the FlyMine public list PL FlyAtlas_brain_top.

A set of three genes (FBgn0011582 (Dpo1R1), FBgn0053517 (dop2R) and FBgn0266137 (Dop1R2) are enriched for the GO term dopamine receptor signaling pathway.