Best Practices For Adopting An Enterprise Screening Platform

Contributed Commentary by Frank Kilty and David Murray

June 30, 2017 | The AstraZeneca R&D organization is responsible for conducting compound screening across multiple disease areas at multiple global sites and across all phases of drug discovery. These activities span high throughput screening (HTS) through lead identification, lead optimization, and drug safety screening and metabolism.

A few years back, we started on a journey to adopt and implement an enterprise-level screening platform for use company-wide. Here we share some of the important milestones in that journey.

First, some background on AstraZeneca. The company has evolved through a variety of strategic mergers, resulting in a diversity of approaches and tool sets for managing screening data.  Initially, AstraZeneca addressed this issue with a central compound database and screening data warehouse (IBIS). By 2011, it was apparent that the tool sets managing screening data workflows (e.g. Hbase, ADA, DataProcessor, etc.) did not support collaboration among research groups, sites, or external collaborators. Not surprisingly, the diversity of tools was costly to maintain and support and our scientists wanted to streamline the tools to improve screening data flows, productivity, and collaboration.

We wanted a single global application to analyze screening data and load that data into the corporate warehouse, which we believed would also reduce support overhead and make working across the AstraZeneca organization much easier. Support for AstraZeneca business rules was a key requirement for any future application. Also, any new application would have to be flexible enough to keep pace with industry trends in screening technology and methodologies.

Initially, we embarked on an in-house development project to deliver a single software solution to analyse all screening data. It soon became apparent that an in-house development project would be a large and expensive undertaking. Therefore, we created a set of requirements based on what we had built internally with consideration for any future requirements. We used these requirements in our due diligence and assessments of commercially-available software.

To address the needs of scientists and the AstraZeneca IT infrastructure, we pulled together a multidisciplinary team of scientists and IT professionals who evaluated both internally-developed and commercially-available software options. The team used the requirements mentioned above as the framework for the Request for Proposals (RFP) process. The RFP was used to select and invite vendors with the best solutions to demonstrate their products using AstraZeneca data. 

The following criteria were considered:

• Functionality compared to in-house system(s)
• Performance against large data sets
• Additional functionality provided by the commercial products
• Ability to integrate with AZ informatics systems
• Long term product evolution and support
• Cost and likely implementation timelines

The evaluation team assessed all criteria, quality of RFP responses, reference feedback, and capabilities for future expansion in areas such as biophysics and compound combinations. After a one-month evaluation, our team selected Genedata Screener as the system that best aligned with our requirements.

The new screening data implementation project commenced in May 2012 and was completed in about 3 months. This was a fast-moving project that used agile methodologies. We went from blueprinting to production rollout within seven months. To date, this remains our departments’ fastest rollout of a complex enterprise-level software solution. The project’s core team consisted of two professional services consultants from our software solution partner and five team members from AstraZeneca. Approximately 50 business users participated in system- and user-acceptance testing. A Business Reference Group (BRG) was created and chaired by an AstraZeneca scientist, David Murray (co-author of this article), who was also designated as the Product Owner, acting as a conduit between the scientists, AstraZeneca IT, and our solution provider.

Each site or group was represented by a BRG member. His or her role was to:

• Define final functional requirements including must-haves.
• Clarify business requirements and changes in requirements that arose throughout the project.
• Provide their time for User Acceptance Testing (UAT) and encourage others to participate.
• Act as Super Users during rollout.
• Provide a single point of contact between the project and their respective groups.

The UAT was done by key users at all our R&D sites. This approach ensured that the new screening platform supported all business processes within project scope, and that the system would be an improvement over legacy systems.

Four different test cases were used to provide documentation that described a workflow through the application. Testers were required to follows steps and register outcomes.

Users ran the test cases in their current application and then again through the new screening platform. They checked that data were identical across both systems and assessed functionality, performance and usability. At the end of the UAT, each BRG member was asked to decide if the new screening platform passed or failed.

Before-and-After Screening Landscape Single-platform Genedata Screener solution enabled elimination of multiple and disparate legacy systems while streamlining user workflows, reducing overhead, and increasing productivity.

The successful rollout of any IT system is vital in gaining user acceptance. We started the rollout by conducting awareness sessions to inform users of the new platform. 

User training was a hands-on experience done in small groups that used their own data and we scheduled training close to platform rollout so that learnings would be fresh when users first started to use the application.

Super Users are the first line of support for all users, and vital to project implementation success. They help perform testing, provide feedback, and make product enhancement requests. We actively recruited these individuals with the approval of their managers as the role of Super User is a commitment that must be factored into an employee’s schedule and job goals.

When the new screening platform was in a “business-as-usual” state, we modified the support model created during development, testing, and roll-out. Again, this was a vital decision as a good support model, which meet user needs, ensures system adoption and maximum return on our investment. Currently, besides the Super Users, we have informaticians supporting the screening platform at each site, and an offsite AstraZeneca team that looks after the databases and carries out any necessary coding work such as for new normalizations and parsers.

We also have a small group consisting of the system owner, site based informaticians, and the global support team who meet once a week to discuss any issues and to approve change requests for things like new normalizations, parsers, etc. – a user goes to their local informaticians with their request, they fill out a form and submit the request that is then prioritized at this weekly meeting. Our weekly meetings continue to be held to address and support any user requests and ensure the solution’s ongoing success.

The business reference group was disbanded after the rollout and we created a new business group (including many of the original BRG members) to ensure user representation across the business and to facilitate activities such as UAT of new versions of the screening platform.

Lessons Learned