Ahead in the cloud - computational drug discovery reaches for the next level

Cloud computing offers the promise of a flexible computing platform for drug discovery researchers, enabling access to technologies previously unavailable, unaffordable or at a scale previously unobtainable.

Cloud computing – the general term used for delivering computational services over the internet – resembles in some respects the longer established grid computing paradigm. Grid computing divides a task into many smaller tasks which are distributed among a large number of, usually, low power computers (for a life sciences example see the Screensaver Lifesaver project). However, setting up a grid computing infrastructure requires significant expertise and resources and it has therefore not been widely used in the biotech sector, although several large Pharma have created in-house capabilities.

Equally, there are important differences between cloud and grid computing; two of the key ones being flexibility and ease of use. Cloud services are sold on demand, typically by the hour; they are elastic – a user can have as much or as little of a service as they want at any given time – and they are fully managed by the provider. The user needs nothing but a personal computer and internet access.

These features of cloud computing make it highly suitable, in principle, for drug discovery applications such as bioinformatics analysis or computational chemistry, where tasks are compute-intensive but demand is intermittent. Catering for peak demand internally, particularly when the required computing resource is “lumpy” in nature, is very inefficient. The flexibility of the cloud computing resource, allocated on demand, is a compelling alternative.

So what is the status of cloud computing in drug discovery research? At a trivial level, many organizations use the cloud for everyday tasks such as email, calendars and word processing. Although mundane, this can be especially useful for SMEs with limited internal IT expertise and with lower costs than traditional solutions.

For scientific computing, adoption has so far been greatest in the bioinformatics arena, driven by the vast quantities of data needing to be analysed from gene sequencing (e.g. the 1000 Genomes Project)and the mature state of bioinformatics databases and software tools. In fact, it is hard to see how these sorts of application can be handled anywhere other than the cloud, with predictions that relatively soon some applications will be producing hundreds of gigabases of sequence data every day.

Adoption in the area of cheminformatics is at a much earlier stage but the potential is huge. One area of computer-aided drug design (CADD) well suited to cloud computing is virtual screening. An early example of a computational chemistry application is the DrugFinder service provided by InhibOx. This service performs structure-based virtual screening against a very large library of available compounds using either its private cloud or public cloud resources, and returning a focused list of putative active compounds for experimental testing,. Outside of the major Pharma companies virtual screening is, for the most part, being performed on a small-scale, piecemeal, basis with wasteful duplication of effort and sub-optimal technology. The cloud enables better solutions through the development of Software-as-a-Service models such as DrugFinder. Many more examples of this sort can be expected to come on-line in the coming years.

Further structural industry drivers also favour the wider adoption of cloud computing, to improve both the efficiency and effectiveness of drug discovery research. An additional benefit of cloud computing is that it naturally lends itself to collaborative projects, with facile mechanisms for data sharing and a reduction in error-prone data duplication. Cloud scalability enables results to be obtained more quickly, with consequent more rapid feedback to research teams. This makes iterative, “what if”, studies possible and so contributes to better decision-making, ultimately meeting the most important challenge currently facing the drug discovery sector - improved productivity.

One concern often expressed when cloud computing is discussed is around data security. The major cloud computing providers rigorously manage the security of their physical computer infrastructure and data centres to very high standards – levels equal to or exceeding those of most in-house systems. For those with the greatest reservations, one approach is the development of “private clouds” that remain inside company firewalls. For example, Amazon has recently started offering VPN and Google enables users to put data behind a secure firewall in its cloud infrastructure.

Cloud computing is not a panacea. Users of cloud computing services need to think carefully about how scalability works in practice, how easy will it be to move to another platform if that becomes necessary, and are there any hidden costs (network charges, software licence issues) that need to be considered? Nonetheless, the opportunities and potential of cloud computing are manifest and the coming decade will see it playing a major role in delivering the next generation of CADD methods to the life sciences.

To find out more about Dr Paul Finn and his work with cloud computing, and other technologies involved in drug development, visit the InhibOx website.

Published: Friday, 9th April 2010 by Dr Paul Finn

Comment on this article

Comments

No comments for this article, be the first to comment