Sean Ekins from Collaborative Drug Discovery explains to Robert Roe how science apps can be used for cheminformatics worflows
Delegates attending the SLAS2014 Conference in San Diego California from 18 to 22 January were able to download a conference app - available free from the iTunes store and Google Play - to keep them updated on the conference programme and to build their own itinerary. Many large events offer similar apps for users of smart phones or tablet computers. But one of the speakers at SLAS2014, Dr Sean Ekins from Collaborative Drug Discovery (CDD), believes that the appliance of apps to science more generally will bring ‘a third computing revolution’ to the laboratory.
Ekins’ presentation was the culmination of three years’ work focusing on mobile applications and how they can be employed by chemists but, although his talk was entitled ‘Cheminformatics workflows using mobile apps for drug discovery’, he believes apps have a wider application to scientists in other disciplines. ‘The amount of science apps and their functionality is increasing all the time’, he told Scientific Computing World ‘now there is a whole ecosystem of apps just for chemistry that run the whole gamut.’ Among the chemistry-based apps already available are the ability to search for structures, view 3D structures, reaction equations, stoichiometry, green chemistry, model building, find reference data, look up educational materials, and share data.
Ekins said: ‘Now we have something mobile, we can go into new situations entirely with these devices; we can do work in a whole different way. I think what we are seeing is exponential growth in apps - and why shouldn’t we see the same growth in science apps?’ He hopes his talk will get people thinking about their own workflows and how they could be streamlined or improved by using phones or tablet devices.
The key to how this technology can accomplish more complex procedures is app-to-app communication. Ekins explained: ‘Data has to flow from one app to another in some way. Apps can be used in the interchange of data and can be used in the workflow to increase productivity.’ Ekins, who has a background in computational chemistry, has collaborated on many of these apps with Alex Clark, president at Molecular Materials Informatics, and Anthony Williams, VP strategic development, head of Cheminformatics for the Royal Society of Chemistry.
Clark in particular has worked on the development of a Chemspider app where a structure is drawn in the Chemspider app and can then be downloaded and opened in Mobile Molecular datasheet (MMDS) app where the molecular structure can be viewed as a scratch sheet. Using the molecular structure obtained from MMDS, a substructure search can be completed using the core scaffold to produce a new datasheet in the ChEBI app. The results from the search can be opened in a SAR table app which provides information on the scaffold, and a list of substituents can also be obtained. Structure activity relationships can be developed by classifying the scaffold and substituents from within the SAR app. The next stage is to use SAR to colour code the compounds for bioactivity. This can also be used as a dataset for modelling and prediction for new molecules. The resulting table informs the user on potential molecules of interest.
Once a decision has been made on a compound, the SPRESI mobile app can be used to obtain synthetic information including a reaction mechanism for a potential analogue. This is accomplished using a search of the core structure which also produces similar potential structures that may be of interest, and the app will also open links to literature where these molecules appear. The reaction mechanism can then be imported into Yield101 where the reagents can be customised, quantities added, and the expected yield calculated. The resulting information can then be exported as a PDF, meaning it can be printed, emailed or just kept on the mobile device to be used directly in the lab. The reaction can also be imported into MMDS where it can be emailed, exported, tweeted, shared or uploaded to repositories like Dropbox. Datasheets can be linked to dropbox using the MolSync app. Storing data in this way opens up more possibilities for secure or open sharing options.
Ekins commented: ‘Alex has created the apps from the ground up for chemistry. If he can do that for chemistry why can’t people do that for biology and for other areas?’
Where are apps going? – App-to-app/cloud
However there comes a point where the processing power of mobile devices cannot keep up with the demands of the computation. This is where the use of cloud services to provide computation can be implemented. Ekins sees a time when ‘heavyweight functionality takes place on remote servers which host the data and search algorithms.’ This opens up opportunities for more complex computation without the need for drastically improved processing power in mobile devices, as computation can be done remotely by cloud processing and the data sent to the mobile device.
Ekins outlined how cloud computing could be implemented for science apps: ‘If you create just a simple docking tool, that would enable you to take a set of compounds and dock them against a small set of target proteins, do that all on the cloud and return the results so that you can visualise them in an app.’ He also suggested a ‘pipelining’ tool similar to Accelrys’s Pipeline Pilot that could automate some of the processes when using mobile apps.
Some of these ideas are only in the concept stage, but innovative thinking can create some beneficial science apps: ‘You get inspiration in sort of interesting places and come up with some crazy ideas and see what really works. I’m hoping that going to SLAS will inspire other people to think about this.’
When Ekins attended a green chemistry conference, held by the American Chemistry Society’s Green Chemistry Institute (ACS GCI), the idea for an app that would list environmentally friendly ‘green’ solvents came to him. He realised all the solvent information was freely available as a PDF but that not many people knew about it. It is now featured in the app.
Apps against tuberculosis
The primary focus of Ekins’ scientific research is tuberculosis, for which he has created a science app called TB Mobile which offers access to a database of bioinformatics data and structures of candidate drug molecules. Dr Ekins said: ‘I would like to create a whole ecosystem of apps for TB and each one has a little functionality.’ He continued: ‘We have TB Mobile to look at targets but it would be useful if we could look at being able to come up with scaffold creations for TB, just give people an idea from known active compounds what might be interesting scaffolds to modify.’
He went on to ask: ‘Could we have an app that listed the compounds that have been tested in vivo, or we could add models for in vivo in vitro and add targets?’
He explains that new features will be introduced as web services and will be available in the next release of TB Mobile. ‘You will have a molecule of interest and you will go off and, from TB mobile, predict what the activities are likely to be, and then also come back with a prediction for targets that compound may hit as well’. At first these services will probably be limited to searching only one molecule at a time, but Ekins believes that ‘in many ways this could break new ground, show people what an app can do, it’s more than just being able to draw a molecule.’
One of the biggest considerations for both vendors and customers in adopting mobile technology is the potential security risks from putting proprietary data into applications for mobile devices. Ekins explained: ‘Security has to be taken into account if you are going to build apps that take advantage of existing business tools.’ But this has been true for mobile devices such as laptops for a long time and has not stopped the spread of web-based ELN services or other laboratory management software. If similar security features are implemented and technology such as fingerprint scanning on iPhones becomes widespread, the barriers to mobile technology will be lowered, even for proprietary data.
Educating people about the potential benefits of using mobile science apps can be difficult and some companies, especially in the pharmaceutical industry, have been slow to adopt the technology, Ekins said. ‘I don’t think that there is really a push from the pharma side to develop these apps. They are still very desktop-orientated and I think they will be for a while unless someone can really disrupt it.’ He went on: ‘You’re paying a million dollars to be able to do this with established software, why can’t you, for a much smaller fee, do this with mobile on the cloud?’
The pharmaceutical industry is highly regulated and large companies already have an expensive, existing infrastructure, both of which makes them slow to adopt new technology. However Ekins believes there are other factors: ‘part of it is having strong informatics leadership; part of it is a generational thing -- and this is something we always see with technologies. Ultimately there are going to be some people that are the leading edge of it, whether they are small companies or biotechs, and those are the disruptive ones.’
Ekins concluded: ‘They [science apps] present a huge disruption to existing software vendors. We [scientists] had to buy quite expensive software; now we can get free software on these devices, or very cheap software that enable us to do the same kind of things’.