How Do Spreadsheets Hold Back Pharma Innovation?
Complete the form below to unlock access to ALL audio articles.
Excel is widely used throughout pharma research projects, largely due to its flexibility, but reliance on the software can present several issues. Better data management is needed to address these shortcomings and ensure industry regulations and guidelines are met.
Technology Networks spoke to Jesse Harris, marketing communications specialist, ACD/Labs, to learn why spreadsheets can be a bottleneck for pharma research and explore how they compare to laboratory informatics tools. In this interview, Jesse also discusses what the future of data management in pharma R&D could look like and introduces us to Luminata®, software designed to bring all process and analytical data together in one interface.
Anna MacDonald (AM): Why is there such a reliance on Excel in pharma research projects?
Jesse Harris (JH): Excel’s legacy status and flexibility have allowed it to become widespread.
Excel has been around for decades. Workflows and systems within pharmaceutical and chemical industries have evolved with Excel as part of their DNA. When a piece of software has become so ingrained in the structure of an organization, people begin to take it for granted, excusing problems as inevitable. Even scientists entering the industry already have years of experience with Excel. Many researchers are not aware that alternatives are even available.
Excel’s flexibility has also helped it become ubiquitous. With enough plug-ins, extensions and workarounds, it seems the application can do almost anything. This flexibility allows Excel to patch holes between systems in your IT infrastructure and connect incompatible systems.
Unfortunately, the problem with flexible tools is that they are not effective at specialized tasks. This is the case with Excel.
AM: Can you highlight some of Excel’s main weaknesses in pharma R&D?
JH: The fundamental problems are lack of chemical intelligence, inability to work with analytical data and collaboration challenges.
Not chemically intelligent: Excel does not “understand” chemistry. Useful operations such as representing chemicals as structures or structure-based search are nearly impossible. Chemically intelligent programs increase productivity by improving the findability and operability of chemical information.
Incompatible with analytical data: Analytical data is at the core of all stages of the research and development process. “Live” analytical data allow results to be reprocessed and interrogated. When data is exported to a spreadsheet, it is flattened to a table of values. This “dead” data cannot be directly reprocessed, so researchers must find the original analytical file, reprocess the data in separate software, and re-import the data into an Excel spreadsheet. This process is error-prone, time-intensive and tedious, and must then be repeated whenever new questions arise.
Does not support collaboration: One of Excel’s most significant weaknesses (both in and out of the pharmaceutical industry) is sharing data in an ongoing project. Different team members maintain their own spreadsheets, which leads to a mountain of files, the majority of which are out-of-date. Shared files only partially solve these versioning problems while increasing the upkeep and complexity. This issue is especially problematic in the pharmaceutical industry, where many scientists fulfilling different functions spread across multiple sites contribute to the same project. Files must be constantly exchanged and maintained to keep the project moving forward, which increases the chance of mistakes. This also increases the risk of security vulnerabilities.
AM: What are the dangers of leaving sensitive information in Excel spreadsheets?
JH: Researchers constantly share spreadsheets with each other, and from one device to another. This means that proprietary information finds its way onto unsecured devices. Most companies have policies to restrict where and how data is stored, but a ubiquitous application like Excel makes breaching policies too easy, even unintentionally.
This can be partially solved by file permissions, encryption and other security measures, but implementing these measures is a real burden. Researchers must waste time micromanaging file permissions, and IT departments need to set up safe files and troubleshoot issues when they inevitably arise. These costs are often ignored because they are difficult to measure and are spread throughout an organization, but this is a real investment of time and resources.
In addition to security, there are also regulatory compliance issues. Regulators require a chain-of-custody for data so they can validate results. Without an audit trail, experiments may have to be repeated, which is costly. Excel does allow links to external files, but these are limited and can be disrupted.
AM: How can these issues be overcome? What other tools and options are available?
JH: Overcoming Excel’s issues first requires understanding how it is used in your organization. This software is often used to connect incompatible systems or consolidate results within a team. Once you have mapped Excel’s role within your organization, you can begin to select software solutions to replace it.
Modern pharmaceutical R&D organizations use a combination of electronic lab notebooks (ELNs), laboratory information management systems (LIMSs), and chromatography data systems (CDSs). These solve part of the data management problem, but each is designed to cover a limited scope. Excel is often used to patch gaps between these systems, but this reintroduces all of Excel’s problems. Tools need to be designed to cover the entire development process.
Ideally, pharmaceutical development teams would be able to use an overarching informatics tool that connected everything together. This would include:
· Consolidating process and chemical data from across a research project
· Eliminating the reliance on Excel
· Searching by chemical structure
· Allowing researchers to reintegrate experimental data and visualize results
· Interfacing with existing equipment and data infrastructure
This is the role a chemical manufacturing and control (CMC) decision support software should occupy. By streamlining data management, work is more efficient, data is higher quality, and scientists make better decisions.
AM: How easy is it for pharma labs to transition from Excel to a laboratory informatics tool?
JH: It depends. Successfully operating a laboratory informatics tool is not just a piece of software; it is also about the business practices that surround that software. Integrating a new piece of technology into your business and training employees in its operation and management does require a time investment. You are also changing the way your employees do their work, which means change management planning is essential.
It may take several months of integration to your current informatics landscape, training and piloting before an enterprise data management solution is ready for production. Customers who see value in these initial deployments can expand the use of software over time. These systems must not disrupt research, so it takes time to deploy everything correctly. This also depends on the organization's size, the complexity of the systems that are currently in place, and expertise in-house.
We should also remember that laboratory informatics transformations are never done. Data management strategies should constantly be advancing. Do pharmaceutical researchers ever stop learning new chemistry or biology? A mindset of continual advancement in data management is necessary for pharmaceutical research and development. Software is also maturing as time goes on, providing expanded functionality.
It's not about completing a technological transition, it’s about continuous technological evolution.
AM: Can you tell us more about Luminata and how it can help CMC product development?
JH: Luminata is a CMC decision support tool designed to consolidate product development data (both chemical and analytical) in one piece of software. Initially designed as an impurities management solution, its functionality has expanded to include tools for process development, forced degradation, supply chain management and more. It is available as both a desktop deployment and browser-based application.
Consolidating data in one common piece of software is not just a quality-of-life improvement; it saves significant time. An example of this improved efficiency is seen during regulatory submission preparation. One customer reported that using Luminata helped them gain > 30% efficiency in preparing data for regulatory submission. Another user said Luminata saved months’ worth of working hours in a single project. This means better efficiency of employees and that medicines get to market faster, which benefits both the companies and patients.
Luminata organizes chemical information into structures and process maps. This allows users to see relationships between chemical compounds within a reaction scheme and click on any molecular structure and retrieve relevant analytical data. That analytical data is live, which means scientists can reinterrogate chromatograms and spectrums to answer new questions without hunting down old files.
AM: Where do you see the future of data management in pharma R&D headed?
JH: Every company is a data company now. Pharmaceutical data is complicated to wrangle, but there is no question that businesses with the best data management practices will have a competitive advantage in the future. AI-related initiatives and investments are exciting, but their success will be built on the back of high-quality analytical and chemical data.
R&D organizations need to think about their needs systematically to add niche expert tools that integrate into their IT landscape and add value to their operation. While a stop gap like Excel may be necessary to fill holes for a limited time, organizations should never stop looking for tools that will help them make their data management more robust, increase their confidence in regulatory submissions, and shorten response times to regulatory inquiries.
Every scientist should also start thinking of themselves as a data scientist. Data management is everyone’s “problem”. While a single monolithic piece of software to do it all is the ideal, the reality is that informatics systems are only as good as their users and the quality of the data those users produce. Scientists, especially those entering the work force, need to learn about databasing, computers, AI and automation.
Pharmaceutical R&D is becoming more collaborative and more decentralized. Companies work across multiple sites and partner with CXOs during their research. Some data management tools were designed based on the assumption that work would be done under one roof. Companies should continue their move to interoperable, vendor-neutral platforms that enable collaboration.
Jesse Harris was speaking to Anna MacDonald, Science Writer for Technology Networks.