Drug Discovery in the Cloud: The Case for an Agile Network

August 20, 2019

Drug discovery, development, and manufacture are data intensive. As the need to utilise this data for analytics, collaboration, and compliance grows, we ask: is now the time to adopt an Agile network? 

The biopharmaceutical industry faces myriad challenges. Global pricing pressures, regulatory complexity, and a growing disconnect between modern capabilities and legacy systems are driving up costs, decreasing productivity, and preventing teams from adopting powerful machine learning (ML) based analytics. Simultaneously, an increasingly research-capable China and a warming attitude toward low-cost generics are increasing competition, further depressing margins.

Despite these pressures, biopharmaceutical R&D pipelines have grown 2.7% between 2017-2018, with the largest increase seen in the preclinical (7.04%) and phase I (3.0%) stages. Likewise, the number of novel drugs delivered by the top 10 companies has increased, growing from 6 in 2016 to 21 in 2017, a 111.1% increase.

In this article, the first of our biopharmaceutical reports, we explore how firms can exploit an Agile Network to further increase pipeline growth, accelerate time to market, and increase margins in the face of pricing pressures.

The State of Cloud in Biopharmaceutical R&D.

Globally, cloud technologies are experiencing significant adoption, with private & public clouds accounting for almost 50% of IT infrastructure spending in 2018. Hybrid-cloud, a synergistic combination of on-premise, private-, and public-cloud technologies, is leading the way with 53% of enterprises pursuing a hybrid-cloud strategy. This contrasts to multi-public and multi-private strategies which account for just 17% and 9% respectively. Whilst the majority of industries have been fast to pursue a hybrid-cloud strategy, the biopharmaceutical sector has lagged behind. At present, just 35% of firms in biopharmaceuticals have moved their IT infrastructure and applications to the cloud. However, a major effort is underway as biopharmaceuticals play catchup. This is reflected in the levels of cloud capital expenditure across industries, with the healthcare and biopharmaceutical sectors (2018: $12.1bn) dwarfing expenditures from finance (2018: $7.3bn), retail (2018: $6.1bn) and manufacturing (2018: $3.6bn).

Why then, has the biopharmaceutical industry been so slow to adopt cloud technologies? Our research suggests three themes preventing the widespread adoption of a full hybrid strategy. First, despite significant investment into securing their services, the security of cloud service providers (CSPs) remains a major concern for executives. Biopharmaceutical data is incredibly sensitive, proprietary, and often a component to high-valuable IP. Thus, a breach, leak, or misappropriation of data result have severe financial and reputational damages. Despite CSPs acting to secure their offerings and reassure biopharmaceutical companies, a feeling of apprehension endures – especially in the face of high-profile leaks (e.g., MediTab) and IP theft (e.g., EpiPen & the Four Thieves Vinegar Collective).

Secondly, legacy systems prevail. In the context of security, legacy systems provide biopharmaceutical companies with a tried-and-tested, well understood (by both producer and regulator), system of data protection. Legacy systems are highly integrated into research environments and they are unlikely to change overnight. Furthermore, biopharmaceutical companies are unable to see the financial advantage to rapidly adopting the cloud due to high levels of investment made into on-premise solutions and proprietary apps.

Finally, our research identified an inability to form productive partnerships, as well as an inability to acquire talent, as a major barrier to cloud adoption. When surveyed, 58% of biopharmaceutical firms stated they were ‘unable to find security integration expertise’. However, in a follow-up question, the same firms signaled a reluctance to bring in third-parties, with just 31% leveraging partner experience. Combined, these two factors have resulted in a ‘state of limbo’, hamstringing internal initiatives and further widening the disconnect between biopharmaceuticals and other sectors.

Despite these challenges, the process of decentralising R&D from on-premise, site specific networks, data silos, and legacy working practices, has proven utility. Amazon, in particular, has built a reputation in the sector, with heavyweights including Pfizer, Johnson & Johnson, and Eli Lily exploiting the scale and agility of Amazon’s EC2 for proteomics, bioinformatics, statics, and adaptive trial design, among others. By adopting a hybrid-cloud architecture, these pioneers realised the ability to rapidly design and iterate trials between global teams, to predict development challenges and runtimes, and to lower cost. Cost, which we previously identified as a hamstring to cloud adoption (i.e., legacy sunk-costs), is also a key driver toward cloud infrastructure. By utilising EC2, typical problems can be solved rapidly and at low cost: in one example, a 100-hour CPU time problem took just 1-hour, and $40, to solve. Whilst this can not be extrapolated to all tasks, it is evident there are large gains in agility and cost savings to be made by working in-cloud.

Opportunities for an Agile Network.

Today, breakthrough technologies are maturing simultaneously. Unprecedented advances in genomics/proteomics, nanotechnology, and analytics (via ML & AI) have generated huge quantities of data. As the biopharmaceutical sector increasingly looks to this data to guide the development of drugs, the need to securely collaborate across trials, departments, and even firms, becomes urgent. At the centre of this exists the requirement for a scalable, secure, and centrally controlled Agile Network.

Several advantages are realised when operating an Agile Network, especially when compared to traditional hardware defined, on-premise, networking. By connecting traditional and hybrid-cloud technologies, biopharmaceutical companies have the potential to ensure better patient outcomes, reduce drug development costs, and accelerate time to market. Furthermore, by implementing an Agile Network, firms stand to increase their utilisation of existing expertise and prevent costly duplication of research – instead focusing their resources on innovation.

A high-level schematic of the Agile network

What then is an Agile Network? We define the Agile Network as consisting of five key components:

One: On-premise legacy applications & data centres.

  • Realise your initial investment into on-premise hardware and expertise. By connecting legacy systems via Cloud Gateway data is made available at global scale, the use of proprietary applications may continue, and technologies can be retired on your schedule.

Two: Private cloud assets.

  • Connect your in-house private cloud to data centres or applications located in the public cloud, your on-premise estate, or to a third party; all through an access-managed and fully-logged central security point held within Cloud Gateway.

Three: Public cloud assets.

  • Network between the thousands of cloud providers, including AWS & Azure, to interface previously incompatible systems and transfer data across internal and external stakeholders. Leverage modern compute capabilities, including Ai as a Service, in the cloud whilst retaining on-premise control of your sensitive data.

Four: Cloud Gateway.

  • Become your own cloud broker and encrypt, transfer, and monitor your data as you move it between your public, private, and legacy network ecosystems. Further enhance your security with software-defined-VPN and connectivity with access to the HSCN, JNET, PSN, MPLS, and internet networks.

Five: An interconnection fabric.

  • Securely deploy, directly connect, and efficiently scale your digital network infrastructure with a global interconnection platform. When combined with Cloud Gateway, an interconnection platform, like Equinix’s dedicated private infrastructure, provides a direct physical connection across all your estates, clouds, and business services.

By connecting these five components into a single, synergistic, and centrally managed ecosystem, previously unattainable levels of data and of internal and external collaboration become achievable. Furthermore, by avoiding the transit of data over the internet, packet dropping, latency, and security issues can be addressed.

The value to be realised from an Agile Network depends on the size, research activities, and IT goals of the user. However, all biopharmaceutical firms, whether ‘big-pharma’ or a university spinout, can gain by using an Agile Network.

In the case of small firms, the ability to leverage on demand Ai as a Service to rapidly analyse experimental data, accelerate quantum mechanics, and automate image-based processes – without significant capital expenditure – is a key benefit. Large firms, thanks to their complex network architectures, also stand to gain. By enabling fully auditable transfers between on-premise and online assets, an Agile Network affords previously unavailable levels of control and visibility over the availability, location, and history of sensitive biopharmaceutical data – even when encrypted in transit.

Independent to the size of the firm, our research identified three primary areas of opportunity for biopharmaceutical companies employing an Agile Network: pipeline growth; time to market; and margin.

Pipeline growth:

By leveraging an Agile Network for R&D, research and development teams can efficiently draw upon distributed hybrid-cloud resources, train models from prior experimental data, and operate advanced analytics. Managed access to private, public, and third party data viaCloud Gateway allows for the rapid collaboration of geographically independent teams. Similarly, this same connectivity enables the rapid assimilation of an acquired biopharmaceutical companies IT ecosystem, giving you instant access to their IP – no matter its network location.

Ultimately pipeline growth is governed by two factors: innovation and M&A. Where M&A aims to unlock synergies, realign portfolios, and acquire innovation, in-house development aims to foster talent, develop ‘blockbusters’ and cultivate IP. In any case, an Agile Network, through its innate connectivity, security, and flexibility, offers the ideal platform from which to innovate.

Time to market:

Time is of the essence in drug development. Only a small window exists in which profitability can be achieved – a product of the mismatch between the 20 year lifetime of a biopharmaceutical patent and the 10-15 years of development time and clinical trials. With an average of five years remaining to recoup cost, decreasing the time to market is a key priority for biopharmaceutical companies.

At a basic level, an Agile Network enables calculations too demanding to be computed efficiently in house to be conducted in a scalable manner in the cloud. With a range of cost affordable solutions now available, cloud based processing – especially in the case of AI, where GPU’s & technical knowhow can be hard to obtain in-house – represents a key tool to speed time to market. Furthermore, by using an Agile Network, resources can be securely located at the edge, where researchers need them most. This reduces latency and eliminates the time wasted waiting for information.

Increased margins:

In 2018 it cost $2.87bn to bring a new biopharmaceutical to market. $1.16bn of this is an opportunity cost – the times cost of capital – and therefore directly impacted by time to market. The remaining $1.70bn represents out-of-pocket and post-approval R&D costs. Here, significant savings can be made.

Major costs for biopharmaceutical firms arise from two solvable problems: the duplication of data and the duplication of cloud subscriptions. By employing an Agile Network, firms receive oversight of their entire network estate. This makes examining the location of data and optimising for cost between CSPs and on-premise hardware simple. Additionally, the use of an Agile Network cuts cloud-migration complexity, cost, and time. In turn, this increases the likelihood of success of a transformation program and reduces the time, and investment, lost to failed projects.

Agile Networks further support margins via scale and flexibility. As the prevalence of biologics and biosimilars, which accounted for 40% of R&D in 2018, grows, the computational demands on R&D departments increases in parallel. Emerging technologies such as CryoEM/MicroED, gene sequencing, and big-data analytics generate, or require, large quantities of data. As teams increase their uptake of these technologies and analyse ever larger data sets, the need to have on-demand access to scalable machine learning stacks grows. By utilising an Agile Network, biopharmaceutical organisations gain access to Ai as a Service – allowing their teams to leverage machine learning at scale, paying only for what they need whilst maintaining control of their sensitive data.

These opportunities represent some of the key benefits of an Agile Network from the perspective of researchers, management, and stakeholders alike. We note these are not conclusive and a wealth of further opportunities, from enabling the secure use of IoT sensors in clinical trials and manufacture, to interconnectivity providing a foundation for mHealth offerings, exist. Regardless of the opportunity, the existence of a secure, traceable, and centralised network provides the essential groundwork for innovation.

Security in an Agile Network.

To realise these opportunities questions surrounding security must be addressed. Cyberattacks result in severe reputational, financial, and regulatory reactions, with the safety of patients and the secrecy of intellectual property put at risk. With a global cost of $1tn and a growing number of breaches (2017: 130, 2018: 145, per capita), cybersecurity is at the top of executives agenda. As the most targeted industry by cybercriminals , biopharmaceutical companies are at incredible risk.

An Agile Network helps address the challenge of security without negatively impacting user experience. By leveraging identity and access management (IAM), firewall as a service, and software defined VPN technologies, an Agile Network equips cybersecurity teams with preventive tools. Furthermore, technologies are employed in a vendor-agnostic manner; enabling organisations to select the ideal solution for their size and security needs without protracted contracts or complex lock-ins. By passing all connections – be they via the internet, 4G, Equinix’s Interconnection Fabric, or any other network method – through a security enforcement core, organisations gain secure central oversight and unified policy management of their global IT estate. This may be further enhanced through the use of PRIME encryption to secure data-at-transit within, uniquely, level 4 of the OSI model.

Whilst an Agile Network can not account for behaviour, nor human error, it can facilitate the secure, end-to-end encrypted, transfer of sensitive or otherwise valuable data between on-premise, cloud, and third party ecosystems – something prohibitively challenging without a centralised network.

The time is now.

As IoT sensors, the need for collaboration, and multi-cloud strategies increase the complexity of a biopharmaceutical companies IT estate, the level of control, oversight, and security required of their network increases. Without a focused effort to unify systems and eliminate the unproductive duplication of research, companies will fall behind – especially as smaller, more nimble, firms leverage Ai as a Service to address challenges in a low-cost, scalable manner.

Before concluding, we note that the subscription to an Agile Network has wide ranging impacts throughout the business, not just on R&D. By bringing together all functions under a unified network, pharmaceutical companies are releasing substantial returns and enabling broad reductions in trial, conversion, and quality costs – to name just a few.

Ultimately, the value of an Agile Network is in the eye of the beholder. To a cash conscious CEO, optimising risk while saving time, reducing direct cash outlays, and efficiently utilising capital are likely to be key motivators behind adoption. However, to a frustrated researcher, the ability to instantly scale resources, transfer data intra- and inter-cloud, and collaborate globally with peers is likely to motivate adoption. Alternatively, the additional security afforded by end-to-end SD-VPN encryption, IAM, and off-internet interconnections, are likely to encourage CIO & CTO uptake.

Agile Networks are set to become an integral component to the success of biopharmaceutical firms. As securely connecting resources becomes increasingly vital to innovation, we conclude: can the biopharmaceutical industry afford to wait?

The Agile Network
Joseph Mitchell