TB Mobile provides information on over 800 compounds with known activity against the Tuberculosis mycobacterium. The app provides convenient reference information that can be browsed, examined, searched and exported. It provides activity information on ~90 specific targets, fingerprint-based sorting and clustering, and target prediction capabilities.
The TB Mobile app makes available the results of a curation study on compounds with activity against the Tuberculosis mycobacterium, funded by the National Institute of Allergy And Infectious Diseases. The app is designed to lower the barriers to accessing this information, so that all researchers with an interest in combatting this deadly disease can use it to the benefit of their own efforts.
TB Mobile is current available for iOS-based devices (iPhone, iPod, iPad) and can be downloaded freely from the iTunes AppStore and Google Play. The app is developed by Molecular Materials Informatics in conjunction with Collaborative Drug Discovery, based on a proposal by Sean Ekins. This article has been updated to describe version 2 of the app, which is currently available for iOS-based devices.
When the app first opens, it takes a moment to organise its data, then displays a screen that resembles the following:
The screen is divided into two blocks: the control block on top, and the compound list underneath, with an action button overlay at the bottom right. The control block provides means for searching, sorting and filtering the compound list, as well as access to menus, which will be discussed below. The compound list is a vertically scrollable list of compounds, which are indicated primarily by structure, and annotated by name (if available) and target codes.
Tapping on any of the compound buttons brings up the corresponding detail view:
Most of the detail view is composed of a scrollable list, which shows all of the available information about the compound. In the above example, the structure of isoniazid is shown at the top, along with its name, CDD number, and resistance information. Each of the known target interactions is summarised, which in this case includes inhA. For each target, a variety of information is shown, including human homolog information, whether the target interaction is essential for activity, known biochemical pathways, and a number of links to available reference information. The links can be clicked on, and will launch the mobile browser, providing a significant amount of further detail.
The top of the detail view contains several button icons:
The select button toggles the bookmark state for the compound. Bookmarked compounds are annotated on the mainscreen by a folded top right corner:
The copy button places a copy of the structure onto the device clipboard, so it can be pasted into other apps, or pasted into the molecular structure search box (described in the next section). The open-in button presents a list of installed apps that are capable of opening molecular structures. If one is selected, then that app is launched and provided with the current structure. The cluster button opens up the interactive clustering view, using the current molecule as the central reference point.
The control block provides several ways to modify which compounds are listed, and in what order. Note that if none of these features are activated, then all of the compounds are displayed, and their order is selected randomly.
To restrict the list of compounds to those whose names contain a certain search string, or CDD number, enter text in the search box:
Now only clotrimazole and econazole are shown. Many of the compounds currently have no common name, and so will be excluded from this search. To remove the text filter, clear out the text.
On the right hand side of the control block is a button that contains the text similar structure. If there is an active molecule search, the structure is displayed instead of the text. Tapping this button brings up several options:
If there is an interpretable structure on the device clipboard, the Paste button is shown. Otherwise the two options are Edit and Edit with..., which bring up a choice of molecular structure sketchers:
The dialog box explains that there are 3 variations to choose from. For users who are already familiar with products such as the Mobile Molecular DataSheet and its gesture-based sketcher, then Expert mode is the appropriate choice. For anyone who wants to learn how to use this powerful sketcher, Novice mode is appropriate, since it includes tool tips and training bars which provide a gentler learning curve. For anyone who wants to draw a structure using a simple and familiar user interface, the Casual mode should be used.
Once the structure is provided, either by editing or pasting, the compound list is sorted according to fingerprint-based similarity to the reference structure:
Similarity comparisons are done by computing ECFP_6 fingerprints for the provided molecule, and using the Tanimoto similarity to rank the similarity to the fingerprints of each of the known molecules. Most similar compounds are shown first.
Note that the TB Mobile app can also be opened from other apps that can launch structures (e.g. MolPrime). In this case, the app will perform the structure similarity ordering immediately upon launch.
The control block shows several buttons for filtering:
The essential and homolog filters behave similarly: tapping either of these buttons brings up a selection menu with three options:
The default state is Maybe, i.e. no filtering. Selecting Yes limits the compound list to only those which have at least one target which is considered essential for activity, or has a human homolog, respectively. Selecting No limits the compound list to those which do not have a target satisfying the constraint.
The gene filter button brings up a dialog that presents a list of target genes:
Each of these genes can be toggled on or off. The compound list will be restricted to those which have activity information about at least one of the selected genes.
Similarly pathways filter button allows known pathways to be selected from an inclusive list:
Selecting and exporting
The action menu is activated by pressing the cog icon in the tool overlay atom the bottom right edge:
The Open in and Send by Email actions refer to the compounds that are currently displayed onscreen: these can be bundled into a datasheet and handed off to an app that is capable of handling them (e.g. MMDS, SAR Table), or composed as an outgoing email. Outgoing emails include the compounds as an MDL SDfile attachment. These two techniques make it possible to select a subset of the content in TB Mobile and use it with other apps, or make it available to collaborators.
There are several menu actions for handling bookmarked state: bookmarking all currently listed compounds, viewing only bookmarked compounds, and clearing bookmarks.
The app provides an interface for viewing molecular similarity in the form of a 2D cluster. When the clustering view is opened (either from the front panel or the detail view), a selection of molecules is arranged:
The layout is achieved by selecting a starting compound (either the first molecule on the front panel, or the subject of the detail view), and placing it in the centre. The most similar compounds (as determined by using ECFP_6 fingerprints and Tanimoto similarity) are placed in its proximity, most similar being placed most closely. The initial layout is then refined using an attraction/repulsion term: the visual effect is that the positions juggle around for a short while, as they relax into their ideal positions.
Tapping on any molecule selects it. Once selected, a molecule can be dragged into a new position. This user-controlled movement causes the other molecules to be re-relaxed, into what may be a new layout. Entities can also be removed from the cluster (tap-and-hold to bring up the menu). This is useful because similarity between a collection of molecules is multidimensional, and embedding in 2 dimensions is a mnemonic. Being able to interactively identify structures of interest and focus on their relative similarity relationships is a way to gain insight into the structure-activity relationships.
By default the molecules are not annotated, but they can be colour-coded to indicate the targets that apply to them, by pressing the target action button at the top left. The list of targets reflects those of the molecules that are currently shown in the cluster:
The molecules are now annotated with colour-coded circles, which match the key on the bottom right:
The cluster can be viewed onscreen, which allows pinch-to-zoom and drag-to-pan, and it can also be exported as print quality graphics by generating a PDF file:
The document can be previewed, sent as an email attachment, or printed directly. Note that because all of the objects are drawn as vector graphics, the resulting output makes use of the full capabilities of the printer.
As well as the curated compounds that are built into the app, it is also possible to add user-supplied structures into the personal stash:
Structures can be drawn using the built in sketcher, imported by launching data from web sources or other apps, pasted from the clipboard, or transferred over from the front panel:
At any time, pressing the prediction button (along the top of the screen) will estimate likelihoods of each of these structures being active against one or more of the known targets. A modified naive Bayesian prediction, using ECFP_6 fingerprints, is used to generate the probability. Targets that have an apparent likelihood are ranked in order with a blue bar, the solidity of which is an indication of the corresponding score:
When there are compounds in the personal stash, they are also represented in the cluster view:
The first version of TB Mobile is described in the literature:
Sean Ekins; Alex M. Clark; Malabika Sarker: "TB Mobile: a mobile app for anti-tuberculosis molecules with known targets",
A paper describing the enhancements introduced in version 2 has been submitted, and is awaiting review.
The TB Mobile is a simple to use app with useful functionality for viewing and manipulating data about compounds with activity against tuberculosis. The app represents a significant development in the effort to make useful drug discovery data freely available in a form that is highly useful to scientists in general, not just cheminformatics experts. Version 1 of the app introduced highly curated and otherwise hard to find data, while version 2 introduced more compounds, more targets, and robust cheminformatics techniques.