Using IT to improve success rates

NGP gathers some of the industry’s leading experts to discuss how technology can support and develop the discovery process. With Amos Dor of Umetrics and Patrick Flanagan of Tripos International.

“In five to 10 years, drug discovery will rely even more on data”
-Amos Dor, MKS Instruments

As the pharmaceutical industry changes in response to unprecedented challenges, how must the information technology supporting the drug discovery process evolve and be redefined in order to improve research productivity and success rates?
Amos Dor. As the pharmaceutical industry changes, the information technology supporting it must be able to keep up. Advanced sensors are needed to gain information about the materials being used, the process they go through, and their results. This investment in sensors, however well intentioned, doesn't add value without the means to collect, store and understand the data they provide. IT must be able to support this data explosion with databases, networks and advanced data analysis tools. The analysis tools especially need to be trustworthy and intuitive to maximize benefits from their use. Multivariate analysis becomes even more important when the number of variables being considered increases. Correlations between variables must be considered, and it is very difficult to see correlations between more than four or five variables at a time when they are looked at individually. To truly optimize productivity and success, the majority of time and resources should be focused on decision making based on relevant information, instead of on how to collect and plot the data.

Patrick Flanagan. All pharmaceutical organizations have their own data stores and data access applications. Standardizing on an extensible product that fulfils their data requirements reduces costs and allows internal IT groups to focus on adding value rather than supporting the base platform. Introduction of data standards makes data more readily capturable, sharable, understandable and ultimately valuable.

The tangible benefits of adopting/moving to a product based solution are straightforward: reduced cost of ownership, focus of internal resources on delivery of higher value scientific capabilities, and reduced burden on the scientist of tedious data manipulation tasks. The less tangible benefits offer a vastly higher value and productivity gains to the scientist and project leadership: increased focus of scientists on the meaning of their data and delivering the next progression in their project rather than on how to move data around their computers, and improvement in the use of and interpretation of data. 

In the current climate, pharmaceutical drug development must increase its productivity. How can the more widespread use of technology support tools help streamline and improve discovery processes?
PF. Technology support tools, made available across the enterprise and utilized in daily workflows, can enable true knowledge management. Scientists need to get to their data as soon as it is available, so timely data entry and simple delivery are critical requirements. Data must be presented in the way scientists want to see it, so analysis tools must be a click away, and require no further complex data manipulations, avoiding the need for data transfer and loss of context, and data access and analysis capabilities must be integrated with all the applications that are required for an entire workflow. Logistical data such as synthesis and test dates need to be accessible to better understand and improve research processes, and cross project datasets need to be readily accessible so progression decisions can be made earlier with greater objectivity. Capabilities to mine corporate information beyond the scientists current project provide the key to realizing corporate knowledge.

AD. When drug development includes Quality by Design (QbD) from the beginning, resources can be focused where they will be the most beneficial. Umetrics' Design of Experiment (DOE) software, MODDE, can be used to reduce the amount of experiments needed, while at the same time, provide as much information as possible from those few experiments. DOE helps decide which variables to control, find the optimal setting, and ensure the process is robust enough to keep quality consistent. It also helps to transfer the product from development to production. Knowing the range of settings that will keep quality consistent before handing the process to manufacturing can reduce the time it takes to qualify and validate the process. Also, the data from the design can be combined with historical data from manufacturing to monitor the process and to do quality predictions in real time.

What software tools can discovery scientists use to help organize, analyze and visualize ever-increasing amounts of data, enabling them to make better, faster decisions?
AD. Advanced multivariate analysis tools like Simca-P+ from Umetrics can help increase productivity in several ways. First, you can quickly create an overview of all the data showing trends, groups and outliers. You can use this overview to see which observations are similar to each other (or to the group average) and which are not. Second, you can look for correlations in the data. Using the variable plots, you can see which variables are positively or negatively correlated with each other or with your quality measurements. If you see groups or outliers in the data, you can quickly drill down to the variable contribution plot, which shows which variables are causing the groups to separate or the outlier to be different from the main population. Within minutes, you can get an overview of the data, compare two populations of data, or see which variables are predictive of quality.

PF. Tripos' Benchware Discovery 360 (D360) is a powerful system enabling every researcher in a life science organization to access and analyze discovery data and easily share their findings with project teams. Using D360, scientists gain simple access to everyday logistical data queries, as well as in-depth data mining and analysis capabilities.  In addition to productivity gains achieved by reducing the amount of time spent formatting, manipulating, and manually transferring data, D360 users report that they are able to perform analyses they simply couldn't do prior to implementing D360. Pfizer recently globally adopted D360 under a unique agreement where we are adopting valuable elements from their organic system (RGate) and adopting them into D360, providing a leading technology to Pfizer scientists while leveraging investment in their legacy system to reduce cost and create a modern, best-of-breed platform for drug discovery and development. Previous to its acquisition by Pfizer, Wyeth selected D360, and worked extensively with Tripos over a five-year period to develop the application to meet the needs of Wyeth scientists.

How do you see the use of software in pharmaceutical R&D developing over the next five to 10 years?
PF. We believe that use of software that provides improved capabilities in the ability to predict successful outcomes - like optimizing lead compounds, predicting off-target effects, and managing attrition, will play an increasingly important role in pharmaceutical R&D. Additionally, the integration of data from discovery through clinical and beyond should allow the process of discovery and development to become more cohesive and efficient, moving towards informatics support of translational efforts from hypothesis through to patient outcome.

AD. In five to 10 years, drug discovery will rely even more on data and the information you can get out of that data. The ability to collect data from complex experiments will increase, and it will become even more imperative to be able to quickly understand the story that data is trying to tell. Process Analytical Technology (PAT), including online multivariate process monitoring, will take a larger role in drug manufacturing. With online process monitoring, batches are supervised to see if any measured variables are different than normal. If so, an alert is automatically generated so action can be taken to prevent the batch from being lost. Companies will be able to monitor products from research and development all the way through to delivery to customers. They will be able to provide a complete history for all delivered products. Because of this and other benefits mentioned previously, the FDA will be even more encouraging about implementing programs like QbD and PAT. Companies who haven't implemented this type of process and quality control will be the exception and will have a more difficult time competing with those who have.  

Amos Dor is the Umetrics US manager at MKS Instruments in San Jose, California. He manages the marketing, business development, application, operation, and sales for North America, South America and Israel. He has 20 years of broad experience in management, marketing, business development and engineering in start-ups and large organizations.

Patrick Flanagan is Chief Operating Officer for Tripos International and leads the organization's commercial efforts. Prior to joining Tripos, his career included strategic sales and business development for leading companies in the healthcare, softwar, and consulting fields including Baxter International, Allegiance Healthcare and A.T. Kearney and Oracle.