Approved Drugs is a mobile app for iPhone, iPod and iPad that presents a selection of chemicals that have been approved by the FDA (Food & Drug Administration) for use as drugs. The primary mode of display is by name and chemical structure:
The source data is collected from a variety of sources, and has undergone some manual and automated curation. Updating and correcting the content is an ongoing process. There are a variety of actions that can be performed on the list, including searching by name, sorting by similarity and filtering by structural fragments. Individual entries can be viewed in detail: the structure can also be visualised in 3D, and its enumerated tautomers examined. There are a variety of sharing and interoperativity features that can be carried out using the structures.
Probe-like molecules can also be viewed in a similar way, as can user-created molecular structures. Calculated properties such as solubility and toxicity can be viewed. Existing and newly proposed molecules can be visualised in context using hexagonal "honeycomb" clustering.
When the app starts, the first thing it does is reads the content data into memory, which takes a few seconds. Once loaded, the screen resembles the following:
The interface is very simple. The top area provides a number of controls for searching and sorting the content. The rest of the view contains a vertically scrolling panel of drugs. Swipe up and down to browse the list. Tap on any of the drug boxes to open the detail view.
There are >1300 drugs listed within the app, which makes it impractical to browse them all to find one of interest. For this reason, when no filters are applied, the list is sorted randomly. To locate drugs, you need to use the search features described below.
At the bottom of the screen are tabs: the Drugs tab is selected by default, which shows the FDA approved drugs. The Probes tab shows the curated NIH MLP probes (see Lipinski et al) with a similar interface, while Custom provides the ability to draw new structures and evaluate them in context.
To look for a drug by name, use the line input field at the top of the app:
The search string can be a partial of complete name. Press the Search button to show all of the drugs that contain the text:
The list of drugs displayed is now restricted to show only those which contain testosterone in the name.
To remove the name search constraint, tap on the text input and clear the text:
Press the Search button again to restore the full list. Note that name filtering is cumulative with other filtering and sorting operations.
A number of structure fragment filtering icons can be activated or deactivated, e.g.:
Enabling a fragment filter will restrict the displayed list of drugs to show only those that have the fragment. The following fragments filter are available:
The list of drugs can be sorted according to similarity by providing a chemical structure as a reference. To use this feature, press the similar structure button at the top right:
Two options appear: Edit and Paste. If there is a chemical structure on the clipboard, it can be pasted in. Molecules can be sketched and copied to the clipboard using a number of other apps, or they can be sketched within the Approved Drugs app.
The embedded structure editor is the same as that used by MMDS. Use it to sketch a structure, or fragment thereof:
Once the edit is accepted, the list of structures is sorted according to similarity to the provided reference:
The similarity metric is based on calculation of subgraph fingerprint counts. The subgraph fragments are un-normalised, which means that there is a strong bias toward fragments of a similar size, which is balanced against having similar structural motifs.
The fingerprints are generated dynamically within the app, which takes some number of seconds, but subsequent comparisons are faster. Structure similarity can be used in conjunction with the text and fragment filters, i.e. the sorting is applied to the filtered list. To remove the similarity filtering, tap on the structure box, and select Clear.
Tapping on any of the drug boxes from the main view opens up the detail view:
The 2D tab is shown by default, and shows the name and structure for the drug entry.
Pressing the button labelled 3D switches to the conformation view:
The 3D conformation can be rotated by dragging one finger. The two finger pinch gesture zooms, and dragging with two fingers translates the structure across the screen.
Some of the drugs have multiple viable tautomers, and for these, there is an additional Tautomers tab:
Swiping left-or-right on the display pans through the tautomer list.
When the Cluster tab is selected, the currently selected molecule is placed into the middle of a hexagonal grid, making up a honeycomb grid. Drugs that are structurally similar to the central molecule are placed around it, with the outward growth done in a way that maximises similarity between adjacent structures:
This visualisation technique can be used to study the way a particular drug relates to other approved drugs. By panning and zooming, the entire landscape can be viewed, albeit at a distance:
Clustering also applies to the probe molecules, and to custom-drawn structures (see below).
Along the bottom of the detail view are a number of icons, each of which provides a specific action that relates to the currently viewed drug:
The Custom tab brings up a selection of compounds that can be specified by the user, which includes several examples by default:
New structures can be added using the large button labelled add new, which brings up the structure editor:
Structures can be opened in detail view by tapping on them, in the same way as for drugs and probes:
Viewing a proposed drug-like structure in the context of known FDA-approved drugs can be done by selecting the Cluster tab:
The honeycomb cluster interface is exactly the same as when invoked for an existing drug, except that the central compound is not necessarily already in the dataset.
Several Bayesian models are included in the Approved Drugs app, and are used for estimation of adequate solubility under physiological conditions, toxity (hERG, KCNQ1) and also probe-likeness. The numeric probability-like values can be observed within the Data tab. Under the Models tab, a graphical representation is shown, which colour-codes individual atoms according to their contributions to these models:
Predicted properties can be viewed for all of the custom molecules at once, by switching on calculations from the mnemonics near the top of the screen. For example, the Aq button represents the solubility prediction model:
These are viewed using the "traffic lights" colour-coding scheme (green = favourable, yellow = indeterminate, red = unfavourable), and can be viewed one property at a time, or by switching on several at once:
The data used by the Approved Drugs app is stored within the app bundle, i.e. once you have downloaded the app, it is present on your device and does not require a network connection. The filtering, searching and viewing features are all implemented natively on the device. Specific interoperability features, such as opening with another app, sending by email, looking up on ChemSpider etc., transfer control to other apps, many of which require an active network connection.
The data is based on a collection originally assembled by Antony Williams, and generously made available. The 3D conformations were originally incorporated from the e-Drug3D list. Many of the 2D diagrams have been redepicted using the 2D depiction algorithm that is part of MOE. Preliminary validation has been carried out by using the ChemSpider API to add an additional self-consistency checking.
It should be noted that the probability of the collection having at least one mistake should be considered to be 100%. Validation and correction of the data is an ongoing process, and it will be refined as subsequent releases are pushed out. If you find a mistake in one of the entries, we would very much like to hear about it: please let us know, by sending an email to firstname.lastname@example.org.
There are no restrictions on what can be done with the data. While the app itself is not free, the data content is compiled from publicly available data, and any modifications or improvements are made available for any purpose. The usual caveats apply: the data is made available as-is, and no responsibility is taken for its veracity or potential misuse.
The Approved Drugs app provides a convenient reference for FDA-approved drugs names and structures. The list can be filtered by various means to locate structures of interest. The structures can be viewed in 2D and 3D, and a tautomer list is available. Various actions are available for making use of the data in conjunction with other apps and network services.